Source: ChatGPT’s conception of “repairable mistakes” and self-correcting democratic institutions. I have some disagreements with some of the characterization and language here (and the quotes felt a little Orwellian to me, especially in the banner), but I thought it was thought-provoking enough to include it as is! What do you think?

Reminder: I am using multiple translations; my main reading is from the hardbound copy of the Mansfield translation. When I refer to page numbers it’s to the Mansfield translation, and when I want to pull quotes that are longer I’ll use the online Schleifer translation, so the text may not match exactly but they are close (and I also read French so I calibrated across the translations to get comfy with Tocqueville’s word choice).

Subscribe now

Part II, Chapters 6 through 8 of Democracy in America build on each other into a more coherent and insightful progression than I honestly was expecting from these chapters. Tocqueville asks first what democratic government does well, then turns to its central danger, the omnipotence of the majority, before closing with the institutions that temper that danger. If you read them in isolation, these chapters look like praise, then warning, then reassurance. Read together, they form an argument about democratic self-correction. Democracy’s advantage lies not in avoiding error but in retaining, through the right formal and informal institutions, the capacity to recognize mistakes and change course without surrendering self-government. This progression and its dynamic feedback effects marks Tocqueville as an evolutionary institutionalist.

Democracy’s real advantage

Chapter 6 opens with an asymmetry: democracy’s faults are visible immediately, while its virtues only become apparent through experience. Tocqueville concedes that aristocracy will generally out-legislate a democracy due to better training, more continuity, more leisure for the work of governing. But skill in lawmaking is not the same as justice in purpose, and his claim is comparative: “The real advantage of democracy is not, as has been said, to favor the prosperity of all, but only to serve the well-being of the greatest number” (p. 223). Majoritarian rule does not harmonize every interest, but it does shift government toward a wider slice of society than aristocratic rule does.

Is Tocqueville being a bit idealistic here? A broader beneficiary class is not the same as a just one, and he knows this. He has a sharper point, though, that I think is one of the most important insights in these three chapters: “the great privilege of the Americans is to be able to have repairable mistakes” (p. 222). Democracy serves people by keeping its errors reversible: bad laws repealed, poor officials removed, failed policies revisited, local experiments contained before they become national ones. Opening governance to more people invites more error, and uncertainty makes some of that unavoidable regardless of who governs. What matters is that the errors stay cheap to fix.

That privilege has a condition attached, which Tocqueville spends the next two chapters complicating. Mistakes stay repairable only so long as the machinery of repair stays intact. A society can enact a bad law and later reverse it, or it can weaken rights, erode legal constraints, centralize authority, and punish dissent, wearing down its own capacity to notice and fix future errors. Repairability is not fixed or guaranteed. It has to be maintained.

Tocqueville grounds this maintenance process in rights: “The idea of rights is nothing other than the idea of virtue introduced into the political world” (p. 227). Rights set a moral ceiling above both self-interest and majority will, and democracy’s distinctive achievement is making that ceiling something ordinary citizens actually practice against, rather than something only elites understand. I found this idea abstract and confusing, but his analogy to property rights helped: “The government of democracy makes the idea of political rights descend to the least of citizens, as the division of property puts the idea of the right of property in general within reach of all men” (p. 228).

Citizens learn liberty the way they learn anything difficult: by practicing it, defending it, and living its reciprocal obligations (and in our time, I think too many of us forget those reciprocal obligations). “There is nothing more fruitful in wonders than the art of being free,” Tocqueville writes, “but there is nothing harder than apprenticeship in liberty” (p. 229). Despotism, by contrast, arrives promising order and quick relief, comforting now and only later revealed as costly. Liberty’s origins are usually tempestuous, and its benefits show themselves only with time and maturity.

This contrast explains one of Chapter 6’s more striking claims. Democracy’s greatness, Tocqueville argues, shows up less in what government executes than in what citizens accomplish outside it: an “energy that never exists without it and that, if only circumstances are favorable, can bring forth wonders” (p. 234). Measuring government purely by administrative output misses the point. The relevant question is what the institutional order frees citizens to do on their own.

From majority rule to majority omnipotence

Chapter 7 turns from democracy’s energy to its power. The same restless activity that Chapter 6 celebrated becomes, if unchecked, the source of Chapter 7’s danger: tyranny of the majority.

Tocqueville is not opposed to majority rule; some decision procedure is necessary, and majority rule fits democratic equality. His target here is majority omnipotence, the belief that a majority may act without obstacle and treat its will as the final measure of justice. “Omnipotence in itself seems to me something bad and dangerous. Its exercise seems to me beyond the power of man, whoever he may be” (p. 241). No earthly authority, whatever its source, is wise enough to hold unlimited power safely, a claim I find persuasive on both epistemic and moral grounds.

His diagnosis of America is correspondingly counterintuitive: “What I most criticize about democratic government as it has been organized in the United States is not its weaknesses... but on the contrary, its irresistible strength” (p. 241). The danger is not too much liberty. It is too little guarantee against tyranny, because the same majority reaches the legislature, the executive, and public opinion all at once.

That last point worries him the most, since democratic tyranny doesn’t have to censor anyone. It can just make dissent expensive. “As long as the majority is uncertain, people speak, but as soon as the majority has irrevocably decided, everyone is silent” (p. 243). That silence is as much an epistemic failure as a political one. Once disagreement disappears, officials hear only their own opinions echoed back, and democracy loses its capacity to test itself. Tocqueville reports finding startlingly little “independence of mind and genuine freedom of discussion” in America (p. 244), and describes a “courtier spirit” spreading through democratic society: where courtiers once flattered kings, politicians and writers now flatter the public.

Self-correction depends on friction, on someone being willing to say a decision is wrong. Majority tyranny does its worst damage exactly where it disables the mechanisms that would otherwise supply that friction.

Democracy correcting democracy

Chapter 8 answers the problem Chapter 7 raises. The American majority may have “the tastes and instincts of a despot,” but it “still lacks the most perfected instruments of tyranny” (p. 250), and Tocqueville credits three institutions with keeping it that way.

The first is administrative decentralization, which limits the reach of majority opinion. Local governments and overlapping jurisdictions create friction between what a majority wants and what gets executed, containing the damage of any single bad decision rather than preventing bad decisions from occurring.

The second is the influence of lawyers, whose professional habits — precedent, procedure, continuity, formal reasoning — slow the passage from popular desire to political action. Tocqueville sees this working through the inherited structure of English common law, adapted and carried forward in America: change happens, but embedded in institutional memory via precedent rather than improvised from scratch. Lawyers as stewards of self-correcting democratic institutions may be the nicest thing anyone has ever said about them!

The third, and the one Tocqueville treats most affectionately, is the jury, which he sees simultaneously as a judicial and a political institution. Jurors listen, weigh evidence, apply general rules to particular facts, and answer for the consequences. The jury moderates democracy by involving citizens in judgment rather than excluding them from power, a paradox Tocqueville captures in one of the best lines in these chapters: “the jury, which is the most energetic means of making the people reign, is also the most efficacious means of teaching them to reign” (p. 264).

That sentence closes the circle opened in Chapter 6 by “repairable mistakes”. Rights teach reciprocity, decentralization contains error, lawyers carry memory, and juries teach judgment — four institutions, doing different jobs, all disciplining the exercise of majority will without ever silencing it. None of the four legislates virtue directly; each cultivates it, building the patience, the tolerance for dissent, and the willingness to lose an argument that repairable mistakes ultimately depend on.

In these three chapters, Tocqueville’s defense of democracy rests on adaptability, not wisdom. A free society stays capable of repair only as long as it protects the institutions that make repair possible. That, more than any particular law or policy, is the argument worth carrying forward.

Source: ChatGPT’s conception of Tocquevillian association

Reminder: I am using multiple translations; my main reading is from the hardbound copy of the Mansfield translation. When I refer to page numbers it’s to the Mansfield translation, and when I want to pull quotes that are longer I’ll use the online Schleifer translation, so the text may not match exactly but they are close (and I also read French so I calibrated across the translations to get comfy with Tocqueville’s word choice).

Democracy in America, Part II, Chapters 4 & 5

Democracy works, when it works, not because citizens choose wisely but because participation forms citizens capable of governing themselves despite not always choosing wisely. This insight runs through Tocqueville’s chapters on political association and democratic government, and it feels evergreen, like the kind of thing each generation forgets and has to relearn.

Subscribe now

Reading these chapters as descriptive is tempting: parties, associations, universal suffrage, public officials, administrative instability, public expenditure. But Tocqueville is analyzing democracy to learn what democratic participation makes of the people who practice it. These chapters are prime Tocqueville as observer, ethnographer, political scientist, theorist.

True to form, he refuses to romanticize the answer. Universal suffrage has advantages, he argues, but guaranteeing good choices is not one of them. Democracy may select mediocre officials, reward short-term thinking, multiply offices, underpay excellence, and produce administrative instability. Aristocracy, by contrast, can offer continuity, skill, grandeur, and long time horizons. Tocqueville sees the advantages and disadvantages of each clearly, and does not pretend democracy is costless. His comparative stance weighs democracy’s costs against aristocracy’s, with its benefits assessed not only in policy outputs but also in the citizens democracy forms.

That comparative analysis is essentially economic, treating forms of government as bundles of tradeoffs. Aristocracy governs with refinement by concentrating power, status, and competence in a narrow class, which produces excellent rulers while politically infantilizing everyone else. Democracy governs less elegantly but diffuses agency, drawing ordinary citizens into public life and giving them practice in judgment. Its output transcends laws and budgets to include habits.

A French liberal inheritance

Tocqueville is heir to a French liberal tradition—Jean-Baptiste Say, Benjamin Constant, Charles Comte, Charles Dunoyer, François Guizot — that fused political economy with historical psychology and judged institutions by the forms of activity, dependence, knowledge, and character they produce. He brings that lens to America, thinking historically about competing social powers (monarchy, aristocracy, church, courts, public opinion) the way Guizot did, while attending to incentives and unintended consequences the way the liberal economists did. Tocqueville also adds a civic psychology sharper than most political economy offers: institutions do not just constrain behavior, they educate desire, ambition, judgment, and self-command.

Political association is the chief mechanism through which that education happens (Chapter 4). Tocqueville treats association as dangerous but indispensable, allowing citizens to organize around shared opinions, publicize claims, and contest majority opinion without violence. In Europe, association often turned revolutionary because it developed under regimes that excluded most citizens from legal participation. In America, association operates inside a framework of broad suffrage: its goal is to persuade the majority, divide it, or eventually become it, not to overthrow it.

That contrast between the Union (America) and Europe leads to different outcomes. Political association converts disagreement into organized persuasion. A minority in the Union can count itself, discipline itself, and seek allies without pretending to embody the whole people, because democracy offers it a path to become more than a minority. Association thus pluralizes popular sovereignty, keeping “the people” from hardening into a moral abstraction.

Association as a substitute for aristocracy

Association is also democracy’s remedy for its own individualism (which Tocqueville understands as a very atomistic concept). Equality loosens inherited bonds, weakening class hierarchy, family authority, and aristocratic patronage. But it can also leave individuals isolated and politically small, formally free yet practically powerless before the majority or the centralized state. Association lets citizens recover collective strength and coordinated action like the capacity for large projects, and without restoring hereditary privilege. Associations are post-aristocratic institutions, substitutes for the social power democracy has dissolved.

The institutional economics here is straightforward: associations lower the cost of collective action among equals. They aggregate dispersed interests, produce information, coordinate effort, and generate civic knowledge unavailable to passive spectators. Political and civil association develop together, Tocqueville later observes: where political association is forbidden, civil association is rare, and where citizens learn the art of association in politics, they carry it into ordinary life.

In other words, democracy has an institutional production function. Under some conditions, participation produces capable citizens; under others, it produces agitation, conformity, or dependence. The better question asks what kind of participation, embedded in what institutions, produces what kind of citizen.

Seen this way, aristocracy and democracy each solve some problems while creating others (again with the tradeoffs ...). Aristocracy coordinates through durable hierarchy and cultivates public men by training a class of men for command. Democracy, lacking hierarchy, must build voluntary mechanisms of coordination and cultivate citizens through use rather than birth. This process creates powerful collective actors through association instead of powerful individuals through lineage.

Institutions that make democratic equality compatible with individual liberty become the central problem Tocqueville asks democracy to solve.

Capacity, not just choice

The tradeoffs and the downsides of democracy show why Tocqueville treats universal suffrage so carefully [Note: The conception of universal suffrage here is far from universal, excluding women, Native Americans, and enslaved people. Tocqueville will discuss slavery and Native Americans extensively later, but takes as given the absence of women from formal political participation.]. He does not believe suffrage reliably selects the most capable rulers. Friends of democracy must solve a harder problem: helping the people choose capable governors while preserving enough popular authority to guide them without administering in their place. That problem sounds, and is, remarkably modern. Democracy needs popular control, but also competence, mediation, and institutional filtering. Tocqueville wants practical participation without the fantasy that direct popular impulse settles every question well.

Guizot’s liberalism struggled with the problem of political capacity: who is capable of governing, and how can a modern society recognize competence without freezing hierarchy into privilege? Tocqueville democratizes that problem rather than abandoning it. He characterizes capacity as a habit institutions produce, meaning that democracy has to both find capable citizens and make them.

That formative point is the deep connection between chapters 4 and 5. Chapter 4 shows democratic participation at its most constructive; Chapter 5 shows democratic government at its most ambiguous. Political association educates citizens in common action, while universal suffrage does not guarantee wise choice. Democratic equality produces agency, but it can also produce envy and impatience. Neither side cancels the other. Democratic self-government lives inside this tension, and doing so successfully requires citizens to practice the habit of self-rule.

Our own moment

That tension speaks directly to our current politics. Much contemporary participation does not look Tocquevillian in the formative sense—it is intense, expressive, nationalized, and identity-laden, but detached from practical self-government. Citizens experience politics more as national drama than local action, and political identity becomes less a habit of shared problem-solving than a badge of belonging and moral sorting.

In The Increasingly United States, Daniel Hopkins has argued that American political behavior has become increasingly nationalized, with citizens interpreting state and local politics through national partisan frames. Hopkins asks what happens to a constitutional order premised on meaningful state and local allegiances once those allegiances fade. Pew’s research on public attitudes finds that large majorities describe politics in negative terms and often feel exhausted or angry thinking about it. The Carnegie Endowment’s review of polarization research adds a related finding: Americans may not be as ideologically polarized as they believe, but they are deeply affectively polarized, disliking and distrusting members of the other party.

Tocqueville would not be surprised. He feared that once detached from robust local and associational life, democratic citizens would become weak where they could act and inflamed where they could not. National politics offers vast moral scale but little practical agency, inviting citizens to identify with abstractions — the nation, the people, history, decline, restoration. Politics cannot avoid such abstractions entirely, but when they displace local action and voluntary association, they deform participation. The citizen becomes a spectator with strong feelings.

Tocqueville’s analysis also helps explain populism’s appeal. Populism offers a shortcut through the hard work of association, telling citizens that “the people” already exist as a unified moral body and that the central task is empowering whoever truly speaks for them. Tocquevillian association begins from the opposite premise: the people are plural, scattered, and unfinished, becoming politically capable only through organization, argument, and repeated action.

Polarization looks different through this lens too. Tocqueville understood association itself as organized conflict, so disagreement is not the issue per se. The problem arises when disagreement loses its intermediary institutions. Parties, local governments, churches, unions, and civic organizations become schools of democratic conduct by forcing citizens to work with others in concrete settings. When politics is mediated chiefly by national media and presidentialized conflict, citizens absorb the emotional intensities of politics without the formative disciplines of participation. We learn to denounce rather than deliberate, to sort rather than associate, to consume politics rather than govern.

Institutional ecology, not nostalgia

Tocqueville’s remedy is thus better participation. Democracy needs institutions that keep politics local, associational, practical, and plural. Local self-government matters because it gives citizens real, substantive, meaningful things to do. Associations teach citizens to act together without waiting for centralized authority. Newspapers connect dispersed citizens and make association possible; juries educate judgment; federalism multiplies the scales of public action. Division of sovereignty plus right of association creates an institutional ecosystem (and a polycentric one at that) that can get the best of democratic equality and individual liberty.

The institutional dimensions of his analysis make Tocqueville such a productive interlocutor for economists. He thinks in tradeoffs, incentives, and institutional effects, but he does not stop where economic analysis usually stops: preferences are not simply given, they are formed in use, and institutions do not simply aggregate interests, they shape the people who hold them. Democracy thus serves as an allocation mechanism and an educational regime, allocating authority more broadly than aristocracy while promising that broad participation can form citizens capable of using that authority well.

That promise is fragile. Democracy can choose badly, centralize power, flatten excellence, and give public opinion tyrannical force. More beneficially, it can also awaken energy in ordinary people, make public life common rather than exclusive, and replace privilege with association. Can democracy build institutions that compensate for what it lacks while cultivating what only it makes possible? That’s the deep and profound question at the heart of these chapters.

Political association, in Tocqueville’s hands, is one of democracy’s central technologies of liberty, supplying coordination without command, power without privilege, and civic education without hierarchy. It teaches democratic citizens to use freedom together, which may be the most important lesson in these two chapters. Democracy is sustained by institutions that help the people become capable of ruling themselves.

Previous articles in this series: I’m not appending these to every one because that just gets annoying, but here we are 1/3 of the way through, so it’s a good time to capture links to the previous articles.

Subscribe now

Ride through is an operational rule in power systems requiring that a customer’s load stay connected to the grid during a brief disturbance rather than disconnecting immediately. The physics of alternating current is the issue. In an AC system, voltage and current oscillate as a sine wave at a nominal frequency (60 hertz in the United States) and the grid operator has to keep generation and load in continuous balance, moment to moment. Frequency is the signature of that balance: it holds steady at its nominal value when generation and demand line up, and drifts up or down when they diverge. Voltage is a more local phenomenon, rising and falling at a given point on the network rather than across the system as a whole. When either quantity moves outside its normal range, protective relays are often set to trip a customer’s load off the grid entirely, shielding that customer’s own machinery from damage.

Most voltage and frequency deviations are small, correcting themselves within seconds through the grid’s own automatic controls. The trouble starts when a very large load reads a passing disturbance as a serious threat and drops hundreds of megawatts (MW) all at once, turning a manageable fluctuation into a systemic problem.

In the days of electro-mechanical technology, the arrangement between a utility and a large industrial customer like a steel plant was straightforward and well understood by both parties. If the plant’s machinery was at risk, its owner could set the plant’s protective relays to trip its connection and drop off the grid. That trip happened quickly by the standards of its era, but slowly compared with today’s digital controls, and it dropped a load far smaller than a modern data center’s. The grid operator, for its part, had well-defined contingency protocols of its own, built around disturbances that likewise unfolded at electro-mechanical rather than digital speed.

The contract between them, and the reliability it protected, was built for those technologies. Now the technologies have changed, and the contract has to change with them.

Source: My conversation with ChatGPT asking it to explain ride through to a non-engineer so I could characterize it well

Ride through and the new operational bargain between data centers and the grid

Data centers consume large amounts of electricity and can change how they consume it in an instant. That speed changes what a utility has to manage. Large industrial customers have always been part of the grid — smelters, steel mills, chemical plants — and their outages have never been trivial. Today’s data centers present a different problem: they are big, fast, electronically controlled, internally flexible, and often opaque to the grid operator watching from outside.

In July [edit: I originally said February, corrected to July], 2024, a lightning-induced fault on a transmission line caused a series of voltage depressions. Sensing these “voltage excursions”, 60 data centers in northern Virginia simultaneously and autonomously dropped off the grid and switched to onsite back and uninterruptible power supply (UPS) resources, reducing grid demand by 1,500 MW (based on the NERC analysis of the event). These responses appear to have been uncoordinated, customer-initiated load withdrawals caused by voltage-sensitive protection and UPS controls.

From the data center’s perspective, fast disconnection looks like prudent self-protection, while from the utility’s perspective, the sudden disappearance of a very large load looks like a major imbalance. If generation keeps producing while demand vanishes, frequency and voltage can rise, and if multiple facilities have similar controls and respond at once, the effect compounds well beyond what any single customer intended.

Ride-through requirements have consequently become serious business. A ride-through rule creates a tolerance window: within a defined range of voltage or frequency deviation, for no more than a defined time, the customer must not trip off or cut load sharply in an uncontrolled way. It may still protect its equipment, just not by exporting a sudden disturbance onto the rest of the grid. If you want more technical depth, this ESIG large load report from February 2026 discusses ride through and other operational strategies for large loads.

While it looks technical, the ride through question is fundamentally economic. Ride through is an engineering setting that is part of a new operational bargain between the utility and the large-load customer.

What the old bargain assumed

The older electro-mechanical grid’s bargain looked different because the technologies had different capabilities. Large industrial customers ran physical processes — rotating motors, furnaces — while utilities operated synchronous generators, carried more rotating inertia, and built practices around known contingencies. A steel mill tripping off was a serious event, but it could not sense a disturbance, transfer demand to batteries, shed selected systems, and recover by a programmed algorithm, all of which a data center can do. Behind the meter, it runs UPS systems, rectifiers, cooling, and backup generators, governed by control software whose aggregate response the utility cannot see unless interconnection demands disclosure and telemetry (remote measurements fed into control systems in near real time).

Why Coase is the right lens

Changing technologies change what good operating rules are, opening up a Coasean bargain between utilities and data centers.

Ronald Coase’s 1960 paper, “The Problem of Social Cost,” pointed out that harms are reciprocal — if a railroad’s sparks damage a farmer’s crops, stopping the railroad harms it in turn. The real question is how to assign property rights/liability for harms so the parties minimize their combined cost. When transaction costs are low, they will bargain to the cheapest approach and split the surplus.

Ride through has exactly this structure. A data center that trips or cuts load during a disturbance imposes costs on the grid and other customers, while a utility requiring it to stay connected imposes costs right back, in equipment, batteries, controls, testing, and telemetry. Neither party is guilty; their technologies simply interact to create new margins of cost and value. One assignment of rights lets the data center disconnect whenever its controls deem necessary, leaving the utility and other customers to absorb the risk; the opposite assignment requires it to ride fully through defined disturbances, leaving it to bear the cost of grid compatibility.

One of Coase’s insights was that with zero transaction costs, the initial assignment wouldn’t matter, since parties would bargain to the efficient result regardless. But transaction costs here are real: the utility knows the grid constraints, the data center knows its control architecture, neither fully knows the other’s exposure, and other affected customers aren’t at the table at all, leaving regulators to stand in for them.

The form of the bargain shapes the outcome, then, and so do its default rules. “The data center shall ride through grid disturbances” assigns rights too crudely to let the parties discover lower-cost ways of splitting the costs and the surplus. A more useful arrangement specifies rights with measurable precision: voltage and frequency conditions for staying connected, minimum load retained at interconnection, the rate and sequence of any demand reduction, circumstances permitting transfer to backup generation, and required telemetry and recovery ramp rate.

Complete versus partial ride through

Complete ride through means the facility stays connected and keeps drawing approximately its pre-disturbance power: a 300 MW data center facing a short voltage disturbance holds demand close to 300 MW, batteries and controls smoothing the event internally. Partial ride through means the facility stays connected while temporarily reducing demand in a controlled, bounded way — that same facility might drop to 220 MW, recover to 260 MW within seconds, and return gradually to 300 MW, an 80 MW reduction rather than a sudden disappearance. Partial ride through can be the efficient compromise, because (in Coasean terms) it can avoid the harm while reducing the combined capital and operating costs to both the data center and the utility.

Economically, partial ride through works because the parties can negotiate over control specificity. The data center says, in effect: I cannot promise full load under every disturbance, but I can promise not to disappear, and I can tell you how much I will retain, how fast, and how I will recover. That promise lets the utility operate with far less uncertainty, which has real cost: a utility assuming an entire campus could trip instantly needs more transmission investment and reserves than one planning around a bounded response. That avoided cost is part of the surplus the two parties can divide, whether as faster interconnection, lower upgrade charges, a different tariff, or a mutually acceptable allocation of reliability risk.

The central object of this bargain is the right to impose a particular dynamic load profile on a shared network, not energy consumption itself. Understood this way, it can resolve data center interconnection timing without necessarily increasing utility investment or residential customer costs.

Source: Me explaining to ChatGPT the sense in which ride through is a Coasean bargain in our new technological context

The disappearance of tacit terms

In the older grid, some operational “contract terms” never needed writing down, since they were embedded in the technologies themselves. Physical characteristics narrowed the range of possible behavior, so no contract needed to say a large customer could not drop 800 MW in 120 milliseconds and restore it through a synchronized, software-controlled ramp. Those terms were tacit in the Michael Polanyi and Hayek sense: implicit and unstated, embodied in equipment, engineering conventions, and shared experience rather than in contract language.

Digitalization weakens that tacit coordination. Once behavior becomes programmable, a data center can decide, through control logic alone, whether to stay connected, reduce load, transfer to batteries, or recover in blocks — choices invisible to the utility unless disclosed and verified. “Large load” no longer tells the operator enough; the relevant context is now large load plus control architecture plus recovery algorithm. Ride-through requirements substitute for that lost tacitness: as the system grows more programmable, the contract has to grow more explicit alongside it.

A layered approach, not a micromanaged one

None of this calls for micromanaging every server rack. A well-designed bargain specifies performance at the interconnection point, articulates what the facility will do electrically when disturbed, and leaves flexibility behind the meter.

Of course, given my love of integrating systems theory with economics, I’m going to suggest a layered approach: an equipment layer, where the data center designs UPS systems, converters, batteries, and cooling to tolerate defined excursions; a facility-control layer, where it decides which loads stay at full power, which curtail, and how workloads adjust; and a grid-interface layer, where the contract specifies what the utility can observe — retained load, ramp rate, ride-through duration, telemetry, event reporting.

But the grid is a shared system, and many affected parties are absent when one utility and one large customer negotiate bilaterally, which is why minimum standards and regulatory oversight matter. Uniform rules have limits of their own, though: a 75 MW facility on strong grid infrastructure is nothing like a 1 GW campus at a constrained corridor’s end, and deep battery capability is nothing like backup architecture that produces a sudden full transfer away from the grid. The efficient arrangement often requires site-specific terms on general standards, since Coasean bargains are context-specific.

The most contentious policy work lies in that middle ground: too little specificity leaves the utility guessing and socializes risk, while too much regulation freezes technology in place. The goal is performance-based specificity — define what the grid needs, make it observable, and let both parties innovate around it.

The larger institutional point

Ride through, then, is a small technical phenomenon pointing toward a large institutional change: the old bargain relied on physical constraints and well-characterized contingencies, while the new one has to rely on explicit contracts, standards, telemetry, and adaptive governance. Digital controls create new gains from exchange as well as new reliability risks. A data center that can shape its load precisely during a disturbance is a source of flexibility with real value, not just a risk to be managed.

The Coasean opportunity is turning that flexibility into a bargain: the data center gains access, predictability, and credit for verifiable grid-supportive performance, while the utility gains a large load that is bounded and operationally useful rather than opaque and potentially synchronized with every other large load on the system.

Technological change does more than add devices onto an old grid—it changes the terms of coordination. When physics supplied more of the governance, contracts stayed thinner; as software expands the feasible activity space, more of the bargain has to be made explicit.

Ride through is one place where that transition is easy to see. Engineering settings become economic rights, private resilience meets system resilience, and the grid’s old tacit bargain gives way to one more explicit, more programmable, and more negotiable.

If we are going to get through this situation productively, we have to approach grid operations and regulatory policy looking for Coasean bargains.

This week’s chapters touch on long-standing debates over federal power and go into detail on the structure and jurisdiction, and moral force, of the Supreme Court. Although this year’s Supreme Court decisions have just occurred and have occasioned the inevitable analysis, and rancor, I leave it to you to associate Tocqueville’s analysis with the modern structure and decisions of the Supreme Court.

Happy semiquincentennial, y’all.

The Betsy Ross flag. Source: Wikipedia

Democracy in America, Volume 1, Part I, Chapter 8 from “On the Federal Courts”, Part II, Chapters 1, 2, 3

In a work celebrating its semiquincentennial this year, Adam Smith placed the division of labor at the center of political economy. Dividing production into distinct tasks lets people specialize, and specializing lets them coordinate through exchange rather than command. The result is greater output, yes, but also a complex order that no single participant designs, performs, or controls.

Near the end of Part I of Democracy in America, Tocqueville reaches for a parallel phrase: the “division of sovereignty” (p. 153). He never makes the comparison to Smith explicit, but the resemblance is suggestive. Just as the division of labor assigns different tasks to different workers, the division of sovereignty assigns different political functions to different governments. The federal government and the states govern the same citizens, but they specialize in different kinds of public action.

This arrangement lets Americans combine political goods that usually appear incompatible: national strength with local liberty, large-scale coordination with small-scale self-government, constitutional authority with a multiplicity of independent institutions. Division also creates complexity and conflict, though. The Constitution can allocate powers on paper, but paper cannot guarantee that the allocation stays clear, that federal laws get obeyed, or that competing authorities respect one another’s boundaries. Tocqueville therefore turns first to the federal courts, which give divided sovereignty its legal form, and then to the political culture—popular sovereignty, parties, and the press—that sustains it.

Making federal power real

“The greatest difficulty,” Tocqueville writes, “was not to know how one would constitute a federal government, but how one would make its laws obeyed” (p. 131).

A constitution may describe a federal government and enumerate its powers, but such a government remains an abstraction until it can enforce its laws. Under a traditional confederation, the central authority addresses commands to member states, and those states must then carry them out. The central government thus depends on the very governments it may someday need to restrain.

The American Constitution adopted a different principle. Rather than governing states as corporate bodies, the Union governs individual citizens directly, with federal officers executing federal laws and federal courts adjudicating them. If an individual violates a federal law, the Union proceeds against that individual, not against the state that failed to prevent it.

Tocqueville regarded this structure as a major discovery in political science. The Constitution divided authority between two governments without reducing the federal government to a diplomatic council of sovereign states. Both the Union and the states could make laws, administer programs, collect revenue, and adjudicate disputes, each within its own sphere.

The individual citizen becomes the point where these sovereignties meet: a resident of a state and a member of the Union at once, subject to both governments and entitled to the protection of both. The Constitution divides sovereignty institutionally, then reunites it in the person of the citizen.

Courts at the boundary

Necessarily, this arrangement creates disputes. Tocqueville describes the Constitution as establishing “two distinct, opposing sovereignties,” (p. 133) each able to expand only at the other’s expense. If state courts alone interpreted federal law, its meaning could shift from state to state, with judges beholden to state governments deciding the limits of federal power. The Constitution therefore created an independent federal judiciary to settle questions of national concern.

The Supreme Court occupies the most delicate position of all. It must interpret the Constitution, distinguish federal from state authority, and adjudicate conflicts between the two, so its jurisdiction reflects the changing scope of federal power itself: “The jurisdiction of the courts of the Union extends or contracts as the sovereignty of the Union itself contracts or extends” (p. 137).

The point matters: jurisdictional disputes are never merely technical. When courts decide whether an issue belongs to the Union or to the states, they help set the practical distribution of political power. The boundary may be written into the Constitution, but it acquires meaning through interpretation, precedent, and political development.

Yet courts have few means to enforce their judgments on their own. Their authority rests on what Tocqueville calls “moral force” (p. 131)—the public’s belief that courts are legitimate interpreters of law. Federal judges may look powerful, but “they are omnipotent as long as the people consent to obey the law” (p. 142).

This combination of strength and dependence gives the Supreme Court its peculiar character. Arbitrating conflicts between sovereignties requires independence and broad authority, but the more extensive that authority grows, the more consequential its errors become. Tocqueville warns that a Court composed of “imprudent or corrupt men” (p. 142) could expose the Union to anarchy or civil war. The danger does not arise from faulty institutional design; it arises from the problem the institution exists to solve. Divided sovereignty requires an authoritative referee, and an authoritative referee can always abuse the position.

The Court does not abolish conflict between governments; at its best, it converts a contest of force into a question of law. Its success depends on the willingness of citizens and governments alike to accept decisions they oppose.

Political specialization

Tocqueville’s defense of federalism goes beyond the familiar claim that it checks tyranny. Division of sovereignty is also productive: different levels of government carry different capacities, information, incentives, and comparative advantages.

The Union performs tasks that require scale: national defense, diplomacy, regulation of matters of common concern, the strength to resist foreign powers. Large nations can mobilize resources, circulate ideas across broad territories, sustain commerce, and project influence abroad.

States and local governments carry different advantages: they stay closer to citizens and more attentive to local circumstance. Freed from primary responsibility for national defense and territorial expansion, they direct their energy toward “internal improvements,” (p. 153) debating plans in town meetings and state legislatures, airing those debates in newspapers, and searching continually for ways to improve their communities. Tocqueville argues that this activity channels ambition away from the conquest of national power and toward the more ordinary pursuit of well-being.

The states are more than administrative subdivisions executing national policy; they are political communities in which citizens acquire the taste and practice of republican government, as Tocqueville notes in Chapter 5. Towns and provincial assemblies developed the habits the federal republic required; the national capital did not impose them.

Federalism therefore combines the advantages of different political scales. Tocqueville summarizes the achievement in one of his most elegant sentences: “The Union is free and happy like a small nation, glorious and strong like a large one” (p. 154).

The pattern is the political analogue to specialization in the economy. The federal government concentrates on functions where size and uniformity pay off; states and towns take functions where proximity, variation, experimentation, and participation pay off instead. Neither level is simply superior—their capacities differ because their scale and position differ.

The cost of complexity

Specialization creates interdependence, and interdependence creates friction. The most visible defect of federal systems, Tocqueville observes, is the complication of their means, or the mechanisms by which government acts. Two sovereignties stand face to face, and no matter how carefully lawmakers define their spheres, they cannot keep those spheres from touching.

A consolidated government offers an appealing simplicity: one territory, one government, one chain of command. Federalism asks more of its citizens—to understand overlapping jurisdictions, to distinguish among several kinds of political authority, to decide whether a given failure belongs to a town, a state, Congress, the President, or the courts, and to tolerate disagreement over where the boundaries actually lie.

The federal system therefore requires what Tocqueville calls “the daily use of the light of their reason”—a demanding condition, and he is not sentimental about the cost: “An idea that is false, but clear and precise,” he warns, “will always have more power in the world than a true, but complicated, idea” (p. 155).

Federalism cannot be exported simply by copying constitutional language. Americans could sustain it partly because they had long practiced local government and had already built habits of political judgment. Geography helped too: weak neighboring powers and relative security from invasion reduced the need for centralized military authority. The system depended not only on laws but on experience, circumstance, and mores.

The Constitution can divide legal authority. It cannot manufacture the habits of mind needed to live under divided authority.

One people, many centers of power

Here the argument turns, moving from Part I of Democracy in America to Part II. Part I describes the formal machinery of government; Part II turns to the passions, opinions, and political practices that animate it.

Tocqueville begins with an apparently contradictory claim: in America, the people govern. If sovereignty belongs to the people, how can sovereignty also be divided between state and federal governments?

The contradiction dissolves once we distinguish the source of sovereignty from its exercise. The people remain the ultimate source of legitimate authority, but they exercise that authority through multiple governments and institutions—popular sovereignty is unitary in principle, plural in operation.

Political parties help organize that pluralism: they assemble citizens into coalitions, attach beliefs to candidates, and translate dispersed opinion into political action. Tocqueville distinguishes great parties, which struggle over fundamental principles, from small parties, which chase personalities, offices, and immediate advantage. Parties divide the sovereign people in order to make its rule effective, but the quality of that division varies—it may clarify competing principles, or it may reduce politics to factional ambition.

The press performs a related function. Tocqueville calls popular sovereignty and freedom of the press “two entirely correlative things” (p. 174): citizens cannot govern themselves if another authority controls the information and arguments from which they form their judgments.

Freedom of the press brings predictable abuses: falsehood, coarseness, sensationalism, partisan fury. Tocqueville nevertheless denies that a government can capture the benefits while shedding the costs. Once censorship begins, it must keep advancing, because thought escapes every partial effort to suppress it. Courts may “arrest the body,” but “the soul escapes them” (p. 173).

The structure of the American newspaper market reinforces this freedom. Newspapers are numerous, cheap to start, and scattered across the country. Most focus on local affairs while still connecting readers to national political debate. No single editor or metropolitan paper can easily corner public opinion—a point Tocqueville makes with an economist’s eye for market structure, decades before anyone called it that.

The press thus creates something like a federalism of opinion, distributing political judgment across many communities and institutions. Like constitutional federalism, it produces noise and conflict—and, like constitutional federalism, it makes domination harder.

Tocqueville’s America is governed by one people through many centers of authority. Its constitutional achievement lies not in resolving the problem of sovereignty once and for all, but in dividing its exercise without dissolving its legitimacy.

The system channels political conflict into jurisdictional disputes, lawsuits, elections, parties, and public argument. It gains resilience from the diversity of its institutions and the specialization of its governments, but it stays fragile because every part depends, ultimately, on citizens accepting complexity, restraint, and the legitimacy of authorities they do not fully control.

The division of sovereignty, then, is more than an arrangement among governments. Like the division of labor, it is a system of differentiated functions and mutual dependence, one that lets the republic do more than any single center of power could do alone. It also requires habits of cooperation, judgment, and trust, without which specialization collapses into paralysis, and division into disunion.

Next week’s chapters: Volume 1, Part II, Chapters 4 and 5 (5’s a banger and there will be lots to discuss!)

Note: In this article I use the word institution to signify various branches of government as well as different levels of government in our federal architecture. In this usage institution has more of an organization definition. This usage differs from the usual way I and other economists and political scientists use institution to signify the formal and informal rules by which we structure our interactions and incentives.

Democracy in America, Volume 1, Part I, Chapters 6, 7, 8 up to “On the Federal Courts”

Subscribe now

American liberty depends not on weakening government, but on distributing power among institutions that possess different sources of authority, different modes of action, and different incentives.

That principle runs through Tocqueville’s discussion in these chapters of the judiciary, impeachment, the federal Constitution, Congress, and the presidency. Dividing power and weakening government are distinct constitutional projects; enumerating separate branches and assigning powers to each is a beginning, not a finished design. His subject is a government capable of acting, but prevented from acting unilaterally through a single, undivided will.

In this semiquincentennial year, reading these chapters is instructive and occasionally unsettling. Tocqueville arrived in 1831, fifty-five years after the Declaration — close enough to observe its institutional experiments still running, far enough to assess which arrangements had taken root and which remained fragile. What he found was a living system, a work in progress, sustained as much by political habit as by constitutional text. He also issued warnings that two and a half centuries of republican experience have done nothing to make less relevant.

Judges as a Political Power

Tocqueville begins with the judiciary because he considers it one of the most original features of the American system. He identifies three characteristics that judicial power shares across nations: courts act only when called upon; they pronounce on particular cases rather than general principles; and they can move only when a dispute is brought before them (p. 94). These limits confine the judiciary and distinguish it from legislative or executive action.

Yet within those limits, American judges possess an extraordinary political power found nowhere else: the right to decline applying a law that conflicts with the Constitution. Understanding why requires understanding the constitutional theory that makes it possible.

Tocqueville contrasts three models. In France, the constitution was treated as immutable; no ordinary power could alter it, which meant courts had no standing to test legislation against it. In England, Parliament’s authority is unlimited; it can modify the constitution at will, which means, in a sense, that no constitution exists apart from Parliament’s current preferences. The American arrangement is different in kind from both. The Constitution is neither immutable nor alterable by ordinary legislative means. It is, as Tocqueville puts it, a “work apart” (p. 95):

An American constitution is not considered to be immutable, as in France; it cannot be modified by the ordinary powers of society, as in England. It forms a work apart that, representing the will of all the people, binds legislators as well as ordinary citizens; but it can be changed by the will of the people following established forms and in cases for which provisions have been made.

This structure makes judicial review both coherent and necessary. When a judge gives priority to the Constitution over a statute, that judge is not claiming sovereignty over the people. The judge is giving effect to the people’s higher and more durable act over the temporary decision of their representatives. And by forcing that challenge to occur through particular cases rather than abstract rulings, the American system shields the judiciary from appearing to attack the legislature directly, preserving both judicial legitimacy and the political independence that makes review meaningful.

The result, Tocqueville observes, is one of the most powerful barriers ever raised against the tyranny of political assemblies. The power is immense precisely because it is constrained in its form of exercise.

Political Judgment and Its Dangers

Tocqueville next examines impeachment, which he calls political jurisdiction or judgment, depending on the translation. The House accuses; the Senate judges; but the tribunal cannot impose ordinary criminal punishment. Its reach extends only to removal from office and disqualification from future service. Criminal liability remains with the ordinary courts.

He interprets this arrangement as fundamentally administrative rather than judicial in character — a means of withdrawing power from someone making poor use of it rather than punishing a crime (p. 102). The American system surrounds this administrative act with the solemnity and guarantees of a judicial proceeding, while withholding the harshest sanctions that make political jurisdiction dangerous elsewhere.

The formal structure is orderly: House accuses, Senate judges, courts punish — each action separated from the others. But Tocqueville’s actual judgment is more disquieting.

Precisely because American political jurisdiction is limited to administrative removal, it becomes more available, more regular, and more tempting as an instrument of majority will. A legislature empowered only to remove officials, not to imprison or execute them, faces fewer inhibitions against using that power frequently. The moderation of the penalty reduces the political cost of wielding it. Americans may have avoided the most violent consequences of legislative tyranny while leaving in place something potentially as dangerous: a removal mechanism convenient enough to serve as a routine instrument of factional control (p. 104).

His conclusion/hypothesis is among the most arresting in these chapters. When the American republics begin to degenerate, Tocqueville argues, the early sign will be easy to identify: an increase in cases of political jurisdiction. The institution designed as a constitutional safeguard carries within it the mechanism of its own abuse.

A Government Made from Governments

Chapter 8 turns to the federal Constitution and the framers’ core challenge: creating a national government capable of acting on individuals without absorbing the states that had authorized it.

The states preceded the Union. They were existing political societies with governments, laws, and sovereign claims. They delegated specific powers to a federal government while retaining the rest. The result was not a centralized nation divided into administrative provinces but a federal republic in which two orders of government acted on the same people — each legitimate within its own sphere.

The framers’ practical solution was giving the national government powers that acted directly on citizens rather than through state intermediaries. Without that direct reach, the Union would depend on state cooperation to carry out national decisions, possessing the form of government without the means. The Articles of Confederation had demonstrated exactly that failure, as Tocqueville notes (p. 106).

American federalism joined two goods that typically resist combination: national capacity for common action and state independence for local self-governance (p. 109). Governing a continental republic from a single center would have been epistemically impossible, requiring knowledge of things no single center could know. Federalism was not merely a political compromise; it was an institutional acknowledgment that self-governance at scale requires distributed knowledge and distributed authority. Vincent and Elinor Ostrom would later formalize this concept as polycentricity: a system of multiple governing centers, each possessing its own legitimacy and logic, producing order through their interaction rather than through hierarchical command. Tocqueville never uses that language, but he sees the thing it names. American republican government works not because a single authority optimally designs all outcomes, but because many authorities, checking and informing each other, arrive over time at arrangements superior to what any one of them could have devised alone.

Yet Tocqueville is careful not to present this as a self-sustaining design. No law, he observes, can foresee everything; institutions cannot substitute for reason and mores. The constitutional architecture depends on something the architecture itself cannot provide.

Executive Power and the Problem of Re-election

Within the federal government, power was divided again. The House represents population directly; the Senate gives equal representation to states and introduces deliberative stability. Bicameralism requires majorities to form through different channels, a test that slows legislation while screening for breadth and durability (p. 111).

The presidency supplies unity and energy that a legislature cannot. But Tocqueville is emphatic about the president’s subordinate position: “The president stands beside the legislature as an inferior and dependent power” (p. 117). In everything essential, the president is directly or indirectly subject to Congress. The executive’s weakness, not its strength, allows a president to govern in opposition to the legislative power (p. 119). This observation cuts sharply against two centuries of subsequent executive aggrandizement.

Institutional power also shifts with circumstances. In foreign affairs and national crisis, the executive’s advantages — unity, speed, secrecy, continuity — come into their own. In ordinary domestic governance, Congress occupies the stronger position. The balance is dynamic, not fixed.

A president’s re-election eligibility disturbs that balance in a specific way. A president seeking a second term is simultaneously a candidate: reading public opinion, managing allies, distributing favor, and calibrating every significant decision against its electoral effect. Governing by principle and governing toward re-election are not always compatible, and the incentive to survive politically tends to prevail. The structure converts private ambition into a reason to deploy public authority for partisan ends, not through corruption in the ordinary sense, but through the mechanism of treating electoral responsiveness as its own justification.

The Constitution as Information Architecture

These chapters speak with unusual force to a republic in its 250th year.

Political authority has shifted toward the executive across recent decades. Presidents govern through administrative agencies, emergency declarations, and broad interpretations of delegated statutes. Congress retains formidable constitutional powers but frequently declines to use them, choosing instead to state aspirational goals while leaving the executive to determine how those goals will be achieved. Claiming credit for the aspiration while avoiding accountability for the tradeoffs is individually rational behavior. When it becomes collectively routine, the constitutional order changes in ways no individual legislator chose and no individual legislator can reverse. And yes, for the game theorists among you, I would indeed characterize Congress today as being in the inferior outcome of a Prisoner’s Dilemma.

The result is poorly ordered government: sweeping executive action followed by litigation, reversal, and another sweeping action from the next administration pointed in the opposite direction. Each cycle erodes institutional legitimacy and the practical capacity to make durable policy.

Tocqueville’s analysis offers a diagnostic framework grounded in something deeper than procedural preference. Each branch’s distinctive mode of action is also a distinctive mode of processing political knowledge. Congressional deliberation aggregates diverse preferences through bicameral bargaining. Courts reason from rules and precedent, producing the predictability that makes planning possible. States govern with local knowledge no national institution possesses. When Congress surrenders its legislative function to administrative agencies, it does not merely lose power; it eliminates an information structure, replacing distributed deliberation with concentrated executive discretion. Policy made without deliberation is likely to be worse policy, not merely less procedurally legitimate policy.

The founders understood this, Tocqueville argues, and their clearest achievement was the courage to act on it. Beyond the people, they established powers capable of refusing the majority’s dangerous demands — powers that, while ultimately deriving from the people, enjoyed enough independence within their sphere to struggle against caprice. That, he writes, is the greatest glory of the American people: knowing how to appreciate that truth, and submitting to it (p. 129).

When the American republic begins to degenerate, Tocqueville hypothesized, an early sign will be the growing use of political jurisdiction. Although the modern history of impeachment is too complicated to fit that prediction neatly, his deeper warning remains relevant. Conflict between Congress and the presidency now takes the form of accusation, investigation, litigation, and judgment, while the ordinary work of deliberation, legislation, oversight, bargaining, and administration deteriorates. Two hundred and fifty years on, Tocqueville’s insight requires less correction than refinement: constitutional dysfunction appears not only when impeachment becomes common, but also when political institutions judge one another instead of governing together.

Thomas Edison’s Menlo Park laboratory, now at the Henry Ford Museum, Greenfield Village, Michigan. Source: Wikimedia Commons

During the telecommunications revolution of the early 2000s, people often joked that Alexander Graham Bell would be bewildered by the modern communications network, while Thomas Edison would still recognize the electric system. The phone had become a pocket computer, a camera, a map, a shopping mall, a newsstand, and a small glowing portal into everyone’s opinions. The electric grid, by contrast, still looked like the same basic machine: big power plants, high-voltage wires, local distribution networks, mechanical meters, bills, and customers at the end of a one-way line.

For a long time, that joke worked because it was mostly true.

Subscribe now

Edison would recognize more than the physical wires; he would recognize the logic of the system. Although the incandescent light bulb became his transformational invention, Edison’s real achievement was the creation of an integrated electric light and power system: generation, distribution, wiring, switches, meters, lamps, financing, installation, and customer service. The bulb was useful only if all of the other pieces worked to power it.

That system logic has shaped the electricity industry for more than a century. Electricity was unlike ordinary goods: a real-time network service produced by large, specialized assets that had to operate together. Generation, transmission, distribution, and consumption had to be coordinated continuously. The physics were unforgiving, the equipment expensive, and the transactions among the pieces difficult to measure and enforce. In that world, vertical integration was a business strategy grounded in a technology coordination problem.

Economists have a useful way to think about this problem. Ronald Coase argued in his 1937 article “The Nature of the Firm” that firms exist because markets are not free to use. Finding trading partners, writing contracts, measuring performance, resolving disputes, and adapting to changing conditions all impose transaction costs. When those costs are high, organizing activity inside a firm can be cheaper than relying on a web of market contracts. This form of organization—vertical integration—brings the many transactions of a supply chain within a single firm that owns and operates the assets and relationships that performance requires. Four decades later, economists Benjamin Klein, Robert Crawford, and Armen Alchian showed why vertical integration becomes especially attractive when parties make specialized investments.

Electricity fit that logic almost perfectly. Power plants, wires, substations, meters, control rooms, and customers were bound together in a tightly coupled electro-mechanical system. Once a generator, a transmission line, or a distribution network was built, its value depended on the rest of the system. The resulting industry, built around integrated utilities, exclusive service territories, and regulation, became a system of government-granted monopolies in exchange for an obligation to serve and public oversight.

That world is now changing amid innovations in digital technology. Sensors, software, communications networks, advanced meters, smart inverters, batteries, power electronics, electric vehicles, and distributed energy resource platforms make it cheaper to observe, measure, verify, forecast, and control electricity services within the grid and around its edge.

An economic interpretation of these innovations is that they lower transaction costs, which changes the boundary between firms and markets. Activities that once had to be coordinated inside a vertically integrated utility may now be coordinated through contracts, platforms, prices, standards, and protocols. The electric system may remain physically interconnected, but its institutional structure can become more modular. A modular system is built from standardized, self-contained parts that can be combined, rearranged, replaced, or expanded, like Lego blocks.

Edison might still recognize the grid’s bones. He would see central stations, wires, substations, meters, and customers. He might not recognize a system in which millions of devices can produce, store, shift, and manage electricity in response to prices, software, and system conditions. The old grid was a machine. The emerging grid is becoming a platform. Can our institutions keep up? Grappling with that question requires a deeper understanding of why the grid was built as an integrated machine in the first place, and why that design made economic sense for as long as it did.

Edison’s system

Making electric light commercially useful required more than a better lamp; it required a network. Edison had to supply not only generators and wires but capital, customers, operating routines, technical standards, and enough confidence in reliability to displace gaslight, candles, and other familiar lighting technologies.

Beginning operations in 1882, the Pearl Street Station in lower Manhattan embodied this systems logic. What looked like a generator wired to a cluster of lamps inside buildings was in fact a central-station power system serving customers in a defined area with an integrated set of technologies. Edison’s direct-current system could not transmit power economically over long distances, so generation had to be located near customers. The physical limits of the technology shaped the organizational form.

That relationship between technology and organization is the central point. Edison’s system was integrated because its parts were deeply complementary: a generator without wires, wires without lamps, or lamps without dependable current each had little value. The system had to come first; the market could develop only after the system existed.

Electricity was therefore unlike most 19^th-century industries in how tightly coupled its technology was. A shoemaker could sell shoes to customers who already had feet; a newspaper publisher could sell papers to readers who already had eyes and literacy. Electricity required an elaborate bundle of complementary assets before anyone could consume the product at all. In economic language, the business involved high asset specificity, strong complementarities, and significant coordination costs. Edison was building an institutional and technological ecosystem.

AC and the system’s first great expansion

The transition from direct current to alternating current (AC) is often told as a high-stakes drama of Edison versus Nikola Tesla, with George Westinghouse entering as the commercial champion of Tesla’s AC. That story has some truth, and it keeps biographers and screenwriters busy. But as an explanation of the electricity system, the more important story is about scale, modularity, and coordination.

Direct current worked well enough over short distances, but it was difficult to transform from one voltage to another. Alternating current could be stepped up to higher voltages for long-distance transmission and then stepped down for local use. That technical feature changed the feasible geography of electric power. Generation no longer had to sit as close to customers, and power systems could extend across larger territories, connecting bigger generating stations to dispersed loads and exploiting economies of scale.

Tesla’s inventions in AC motors and system components were crucial, and Westinghouse’s role in commercialization mattered enormously. But the AC system was not the product of a single mind. It emerged over decades from a broader engineering and entrepreneurial ecosystem involving transformers, motors, meters, generators, insulation, transmission lines, protection systems, standards, financiers, managers, and customers. A new technological architecture changed the system’s economic possibilities.

AC loosened one constraint while reinforcing another. It made the power system geographically larger, but it did not make the system institutionally modular in the way that digital networks would later become. By enabling larger networks and larger generating stations, AC strengthened the case for coordinated planning and operation. A bigger synchronized machine required more elaborate coordination.

The early 20^th-century electric utility emerged from this logic. The industry developed around large fixed investments, extensive network economies, and long-lived assets. Utilities built generation, transmission, and distribution systems; served customers within exclusive territories; recovered costs through regulated rates; and accepted an obligation to serve. Public utility regulation became the institutional bargain that made this structure politically and economically sustainable.

That bargain was imperfect, but it reflected the economic conditions of the technology. When the cost of coordinating through markets is high, hierarchy and centralized control become attractive. When duplication of infrastructure is wasteful, monopoly becomes feasible. When customers are captive, regulation becomes necessary. The vertically integrated utility was not just an exercise in corporate control; it was also an answer to a then-new question: How do you coordinate a tightly coupled, capital-intensive, real-time network whose services are essential and whose customers have few alternatives?

The logic behind vertical integration

The traditional utility model emerged because it economized on real coordination costs. In ordinary economic models, markets often look almost frictionless. Buyers and sellers meet, prices adjust, contracts are enforced, and resources move to their highest-valued uses. In the real world, using markets is costly. People must find one another, discover prices, evaluate quality, write agreements, monitor performance, adapt to surprises, and resolve disputes. Those are transaction costs.

Coase’s insight was that firms and markets are alternative ways of organizing economic activity. A firm replaces some market transactions with managerial direction, and the boundary of the firm depends on the relative cost of coordination inside the firm versus coordination through markets. Klein, Crawford, and Alchian added that vertical integration becomes especially attractive when investments are specific to a particular relationship. If one party builds an asset that is valuable only when paired with another party’s asset, the parties may become vulnerable to opportunistic bargaining after the investment has been made. Economists call this the hold-up problem.

Electricity is full of such relationship-specific investments. A power plant is built to connect to a grid. A transmission line is built to move power between particular locations. A distribution network is built to serve a particular territory. Customers install equipment that depends on reliable electric service. These assets are expensive, long-lived, and difficult to redeploy.

The product itself adds another layer of complexity. Electricity must be balanced continuously because of the physics of AC networks. Supply and demand must match in real time, not approximately next quarter. The network has physical constraints, and actions at one point can affect conditions elsewhere. Reliability is a system property, not a simple bilateral transaction. Whether my lights stay on depends not only on my relationship with the utility but also on generation adequacy, transmission capacity, distribution operations, voltage control, system protection, and the behavior of many other users.

Under those conditions, vertical integration reduced transaction costs. Instead of contracting separately for every component of generation, wires, balancing, reliability, metering, billing, and customer service, the utility organized many of those activities internally. Regulation then supervised the resulting monopoly. The model had costs, including weak incentives for innovation and efficiency, but it economized on a genuine coordination problem.

The traditional electric utility was, in this sense, a Coasean institution built around an Edisonian machine. It endured, too, because state-level regulation reinforced the business model, creating the economic certainty utilities needed to build and expand this tightly coupled system around large-scale technologies and centralized organization.

The first institutional unbundling

The late 20^th-century restructuring of electricity regulation challenged the old model, especially in generation. Catalyzed in part by the combined-cycle natural gas turbine, policymakers and economists increasingly argued that generation did not always need to be owned by the same firm that owned transmission and distribution wires. Independent power producers, wholesale electricity markets, open-access transmission rules, regional transmission organizations, and competitive retail supply in some states all reflected efforts to separate potentially competitive functions from natural monopoly functions.

This restructuring showed that vertical integration was not an essential fact of electricity. Some functions could be unbundled. Markets could coordinate generation dispatch. Independent firms could build and operate power plants. Regional transmission organizations could run wholesale markets and manage transmission access across utility boundaries. Prices could convey information that had previously been embedded in utility planning and internal dispatch.

But restructuring has been incomplete, uneven, and institutionally constrained. Transmission and distribution remained network monopolies. Retail competition in different states produced mixed results. Wholesale markets required complex rules, administrative oversight, and constant adaptation. Reliability still required centralized coordination. Resource adequacy, transmission planning, interconnection, market power mitigation, and state policy integration all turned out to be harder than the 1990s reformers anticipated.

That experience offers an important lesson for the next phase of electricity transformation. Unbundling is not magic. Markets are not self-executing; they are institutionally grounded processes that require rules, measurement, enforcement, information, settlement systems, and governance. When those institutional complements are weak, unbundling can disappoint. When they are strong, unbundling can create enormous value.

This history leaves us with a sharper way to think about what comes next. The vertically integrated utility was neither an accident of corporate ambition nor a mere artifact of regulation; it was an economically sensible response to a real coordination problem, pairing Coase’s logic of the firm with Edison’s tightly coupled machine. Restructuring then demonstrated that some of those coordination costs had fallen far enough to support markets in generation, while also revealing how much institutional structure a working market actually requires. Both lessons travel with us into the transformation now in progress.

Digital technologies make a deeper form of unbundling possible, especially at the edge of the distribution network. But possible does not mean automatic. Digital technology will not decentralize the electric system on its own; whether institutions allow decentralized resources to be observed, valued, coordinated, and compensated in ways that make economic sense will shape what the architecture of future power systems looks like and enables.

First published at Dispatch Energy, https://thedispatch.com/newsletter/dispatch-energy/thomas-edison-electricity-grid-power-economy/

Faneuil Hall in 1826. Source: Wikimedia Commons

In Part I, Chapters 3, 4, and 5 of Democracy in America, Tocqueville moves from the social foundations of democracy to the local institutions that teach people how to govern themselves.

Reminder: I am using multiple translations; my main reading is from the hardbound copy of the Mansfield translation. When I refer to page numbers it’s to the Mansfield translation, and when I want to pull quotes that are longer I’ll use the online Schleifer translation, so the text may not match exactly but they are close (and I also read French so I calibrated across the translations to get comfy with Tocqueville’s word choice).

Subscribe now

In the previous installment of this series, I examined Tocqueville’s account of America’s “point of departure” — his argument that nations, like individuals, bear the marks of their origins. Chapters 3, 4, and 5 now move from origins to structure. Tocqueville begins with what he calls the American “social state”, proceeds to the principle of popular sovereignty, and then descends into the detailed workings of towns, counties, and state governments.

He does not begin with the Constitution, the presidency, or the Supreme Court. He asks what kind of society Americans inhabit, where political power is understood to originate, and how ordinary people encounter that power in their daily lives. His answer is institutional as well as moral and developmental. Free government depends on more than distributing authority correctly on an organizational chart; it depends on forming people who know how to use liberty, and the town is where that formation begins.

The Social State and Its Warning

Chapter 3 opens with one of Tocqueville’s most important claims: “The social state of the Americans is eminently democratic” (p. 46). He never offers a tidy definition of “social state”, expecting readers to infer it from context. The concept describes the underlying pattern of relationships within a society — property distribution, status, inherited privilege, and social mobility. A political regime tells us who formally governs; a social state tells us how people stand in relation to one another.

America’s social state is democratic because it is characterized above all by relative equality of conditions — not equal wealth or talent, but the absence of durable, legally entrenched hierarchy. Wealth exists, but it does not create permanent castes. Social positions remain fluid; families rise and fall.

Tocqueville’s treatment of inheritance law makes the institutional logic precise. Under primogeniture, landed property remains concentrated and attached to a family line. Under partible inheritance, property divides among heirs at each death, producing what Tocqueville calls “a revolution in property” (p. 47): “Not only do goods change masters, but they change, so to speak, nature; they are constantly fragmented into smaller portions” (p. 47). A civil rule governing estates quietly reshapes the distribution of political power across generations. The democratic social state is neither simply natural nor simply designed — it is an emergent order whose development is powerfully shaped by legal rules, even as those rules themselves reflect underlying social conditions.

But equality of conditions does not guarantee liberty. A society of relatively equal individuals may move toward genuine self-government, or it may submit to a centralized and absolute power. Individuals freed from aristocratic superiors may govern themselves, or they may find themselves equally subordinate to a distant state. The corrupted democratic passion does not raise the weak; it seeks to pull down anyone appearing stronger or more independent. In its worst form, it leads people to prefer equality in servitude to inequality in liberty. America had avoided that outcome, Tocqueville thought, because its equality was joined from the beginning to favorable mores, enlightenment, religion, and, above all, local institutions capable of channeling democratic energy toward self-governance rather than submission.

Popular Sovereignty as Practice

Chapter 4 examines the political principle corresponding to the democratic social state. Tocqueville describes popular sovereignty in America not as a constitutional phrase but as a “law of laws” — a belief recognized by mores, proclaimed by laws, spreading freely without obstacle (p. 54).

The principle begins with individual sovereignty. Under the democratic dogma, each person forms an equal portion of the sovereign power, presumed as capable of self-government as any fellow citizen, obeying no one on grounds of natural superiority. Why, then, does anyone obey at all? Tocqueville’s answer cuts to the core of democratic political obligation (p. 61):

He obeys society, not at all because he is inferior to those who direct it, or less capable than another man of governing himself; he obeys society because union with his fellows seems useful to him and because he knows that this union cannot exist without a regulatory power.

Obedience rests on the recognized usefulness of association, not on natural hierarchy.

Despite being mutually reinforcing, individual sovereignty and popular sovereignty stand in potential tension. The same democratic principle that rejects rule by a natural superior can, when aggregated, produce rule by an overwhelming majority. The equal moral standing of individuals gives rise to collective power, and that collective power may bear on those same individuals with considerable force. Tocqueville does not pursue this problem fully here — his analysis of majority tyranny comes later — but its foundations are already visible, embedded in the very logic of popular self-rule.

The American Revolution did not create this principle from an abstract theory. It generalized and elevated what local communities had long practiced, sovereignty developing through lived experience before receiving a more universal political expression. This sequence is important for understanding how institutions actually form and sustain themselves — a theme running through all three chapters, and one that helps explain why Tocqueville turns immediately from popular sovereignty to the town. The town is the arena where the sovereignty principle becomes practice, where individual sovereignty and collective self-rule find their most immediate, tangible, and educative expression.

The Town as a School of Liberty

Chapter 5 is the longest and most institutionally rich of the three chapters. Tocqueville explains why understanding American democracy requires studying towns and states before the federal Union: the states created the federal government, not the reverse, and most important American political institutions developed first at the local level.

Tocqueville begins with an arresting observation: “The town is the only association that is so much a part of nature that wherever men are gathered together, a town takes shape by itself” (p. 57). But he immediately distinguishes the natural existence of towns from the political achievement of town liberty: “If the town has existed ever since there have been men, town liberty is something rare and fragile” (p. 57). Central authorities tend to encroach on local independence, while local institutions often lack the strength to resist. Town liberty survives only when legal protection has existed long enough to become embedded in mores.

Mores, on this account (and as discussed last week), are not doctrines imposed from above. They emerge through repeated interaction — citizens participating in local government, learning its procedures, developing expectations about appropriate conduct, and acquiring attachments to institutions through which they act. The formal institution gradually becomes a social habit, sustained not by compulsion but by familiarity and investment. And sometimes that goes the other way, with social habits that are robustly beneficial being codified as formal institutions.

Tocqueville’s deepest claim about town institutions is not that they produce efficient administration. It is that they educate citizens: “Without town institutions, a nation can pretend to have a free government, but it does not possess the spirit of liberty” (p. 58; note here that Schleifer uses “liberty” while Mansfield uses “freedom”).

Town government puts liberty within the reach of ordinary people, giving citizens manageable responsibilities and repeated opportunities to exercise judgment. They debate practical questions, raise revenue, select officials, oversee schools, repair roads, resolve disputes, and cooperate with neighbors whose opinions they may not share. In doing so, they acquire civic competence — learning that public problems do not solve themselves, that collective action requires patience and compromise, and that the neighbor with the most emphatic opinion is not always the person best equipped to repair the bridge. Liberty becomes habitual by being exercised.

Scattering Power

One of Tocqueville’s most striking observations is that New England town government deliberately “scatters” power (p. 64):

See with what art they have taken care in the American township, if I can express myself so, to scatter power in order to interest more people in public things.

Power is divided both to prevent abuse and to widen participation. By distributing offices and responsibilities among many citizens, the town draws more people into public life — numerous officials with specific duties creating numerous points of entry into governance, making public authority less imposing and more accessible.

This arrangement also does something subtle with ambition. At the national level, only a tiny number of people exercise meaningful power. Town government brings esteem, influence, and responsibility into ordinary life, concentrating the desire for recognition near the domestic hearth and redirecting passions that might otherwise become socially disruptive toward local public action. Tocqueville is not trying to eliminate ambition; he is institutionalizing it, connecting ordinary motives to socially useful activity.

Scattering power thus performs at least three functions simultaneously: limiting domination by preventing authority from concentrating in a single visible official (the constitutional function); mobilizing dispersed knowledge by giving those closest to local problems the capacity to address them (the epistemic function); and forming citizens by involving many people in public responsibilities (the civic function).

Tocqueville’s formulation of the objective is superb (p. 67):

In this way they wanted to make authority great and the official small, so that society might continue to be well regulated and remain free.

This institutional vision differs subtly but importantly from Montesquieu’s and Madison’s separation of legislative, executive, and judicial powers. Tocqueville is examining the vertical and horizontal dispersion of administrative responsibility throughout society — a nation may contain internally separated branches while still centralizing the administration of local life, leaving citizens as recipients rather than participants.

He distinguishes two forms of centralization. Governmental centralization concerns interests common to the whole nation: general laws, foreign affairs, national defense. Administrative centralization concerns the detailed execution of public affairs within particular localities. Americans had “almost entirely isolated administration from government,” maintaining capable central authority on genuinely national matters while leaving most administration to states, counties, towns, and citizens themselves (p. 85).

He concedes the costs: local administration may be uneven, amateurish, or duplicative. A centralized administration may perform certain tasks more neatly — files standardized, procedures uniform, reports arriving in triplicate, everyone enjoying the modest consolation of knowing that nothing can happen until Form 17-B has been approved in the capital. Yet administrative efficiency is not the highest political value. A system that solves public problems for citizens may gradually deprive them of the habit of solving problems together, the central administration growing more capable as the population grows less so. This dynamic becomes self-reinforcing. “What I admire most in America,” Tocqueville concludes, “are not the administrative effects of decentralization, but its political effects” (p. 90).

Tocqueville’s America and Ours

The America Tocqueville describes in 1831 feels simultaneously familiar and remote. The vocabulary persists — federalism, local control, divided powers — but the political structure has shifted, dramatically. National politics increasingly organizes political identity, even when the decisions affecting daily life remain local.

His description of ambition is especially revealing. In 1831, most politically ambitious citizens directed their energies toward towns and states, where they could exercise visible responsibility. Contemporary politics has largely reversed this orientation — national office dominating attention, fundraising, and partisan identity, while even local disputes are rapidly translated into national ideological conflicts. When political attention migrates upward, citizens may begin to regard meaningful power as something exercised elsewhere, local participation seeming small or technical by comparison. Yet national politics offers most citizens little opportunity to do anything beyond voting, donating, and becoming angry at professionally selected intervals.

Tocqueville would likely ask what this structure does to civic agency, whether it forms citizens capable of governing or spectators skilled at expressing approval and contempt.

These questions do not imply that America should recreate the Massachusetts town of 1831, even if it could. Many problems genuinely cross local and state boundaries, requiring coordination at larger scales. Tocqueville’s principle is not that everything should be local. His principle is that the burden of proof should fall on removing responsibility from those with the closest interest and knowledge.

Freedom, in Chapters 3–5, does not follow automatically from equality, nor from elections. It must be institutionalized, requiring legal protection, dispersed authority, repeated participation, and enough time for institutional practice to become part of the mores. The sovereignty of the people can remain a majestic abstraction, proclaimed in constitutions and invoked in speeches. Or it can become part of ordinary everyday life, exercised imperfectly by citizens deciding together what to do about schools, roads, public order, and the repair of the town hall.

The former may produce the appearance of free government. The latter produces its spirit.

Previous articles in this series:

Source: Town Meeting 101, Discourse Magazine, 2022

Note: In this first essay I didn’t both with quotes and page references because I wanted to focus more on broad themes. If I am doing page references or quotes I’ll specify which edition and put a reminder in each article. And please do treat this like a reading club by sharing ideas and interpretations and implications in the comments!

Subscribe now

Alexis de Tocqueville went to America in 1831 to discover democracy.

Democracy in America is often treated as a book about the United States, and of course it is. Tocqueville observed American geography, settlement patterns, town government, religion, slavery, newspapers, courts, parties, associations, manners, ambition, restlessness, and the habits through which democratic people learn, or fail, to govern themselves. America was the great case study; democracy was the real object of inquiry.

For Tocqueville, democracy is not merely a system of elections, reducible to majority rule, representative institutions, written constitutions, or popular sovereignty. Democracy is a social condition before it is a political form. It begins with what he called equality of conditions: the weakening of hereditary rank, the erosion of fixed social orders, and the gradual emergence of a world in which human beings increasingly see themselves as comparable, mobile, and entitled to judge their own affairs.

For that reason, Tocqueville began his great book in a way that can feel surprising to modern readers. He does not start with Congress, the presidency, the Constitution, or the mechanics of American elections. He begins with equality. He begins with social transformation. He begins with the most important fact, as he sees it, of the modern world.

The young United States of America was distinctive because equality of conditions had advanced further there than in Europe, and because it had done so under unusual circumstances. America, for Tocqueville, is democracy with fewer aristocratic ruins lying around. France had encountered democracy through revolution, violence, centralization, and the destruction of an old regime. America seemed to show democracy nearer its point of departure, before centuries of aristocratic hierarchy had so thoroughly complicated the landscape.

More than asking how America works, Tocqueville asked what happens to human beings when equality becomes the organizing fact of social life.

That question remains alive today, and may be more alive now than when Tocqueville posed it. We still live inside the world he was trying to understand: a world in which equality is morally powerful, liberty is rhetorically revered, freedom is demanded in private life, and democracy is invoked as both ideal and weapon. We often use these words as if they all meant the same thing. Tocqueville’s insight begins with his refusal to do that, and with his interrogation of them as distinct concepts.

Equality is not liberty

Tocqueville carefully separated concepts that modern political language tends to blur. Equality, liberty, freedom, and democracy are related but not synonymous. Much of Democracy in America is an extended meditation on how they reinforce one another, how they threaten one another, and how difficult it is to preserve liberty in a democratic age.

Equality, for Tocqueville, is first a social condition. It does not mean that people have the same wealth, talent, education, status, or influence. Equality of conditions means that inherited rank loses its power to organize society. Social positions become less fixed. Wealth circulates. Professions open. Education spreads. Families rise and fall. The old assumption that some people are born to command and others to obey becomes less credible.

This change is moral and psychological before it is institutional. Equality changes how people see one another and how they see themselves. The peasant, the merchant, the lawyer, the minister, the legislator, and the landowner may still differ in wealth and power, but they increasingly inhabit a shared social imagination. Each can imagine becoming something else. Each can compare himself with others. Each can resent, sympathize, admire, imitate, compete, and aspire.

Democracy is the social and political world that grows from this equality of conditions. Tocqueville used the word more broadly than we do: not merely electoral government, but the whole ensemble of habits, ambitions, anxieties, resentments, beliefs, and expectations that social equality generates.

Liberty is something different and more demanding, not simply the absence of restraint. Liberty requires the capacity for self-government, and self-government requires citizens who can do more than cast votes. They must form judgments, accept responsibility, participate in common affairs, respect rights, restrain appetites, cooperate with others, and resist both private domination and public servility.

Freedom, in the ordinary sense, often means the ability to act without immediate coercion: to speak, worship, move, work, buy, sell, associate, and choose. Tocqueville values these freedoms without mistaking them for political liberty, which demands considerably more. A people may enjoy many private freedoms while slowly losing the habits of public liberty. Citizens may withdraw into private life, ask centralized authority to manage common affairs, and retain the feeling of personal independence while surrendering the practice of collective self-rule.

This possibility is one of Tocqueville’s deepest anxieties. Equality may produce democracy, but democracy does not automatically preserve liberty. Indeed, democracy can endanger liberty when equality becomes a passion for sameness, when majority opinion becomes morally coercive, or when citizens prefer administrative comfort to the burdens of self-government.

The central problem of democratic society is therefore not whether equality will come. Tocqueville thinks it had already arrived, treating its advance as nearly providential. The problem is whether equality can be joined to liberty.

Through his 1831 lens: democracy is coming. The question is whether it will be free.

The long democratic revolution

Tocqueville’s Introduction offers a compressed and stylized history of European social leveling — a time-lapse film, not a researched parish record. In the old aristocratic world, power was rooted in land. Over time, other forces eroded that order: clergy gained influence, monarchs elevated commoners to weaken nobles, lawyers rationalized, commerce expanded, towns grew, movable wealth displaced inherited property. Each group acted locally, strategically, and partially, advancing equality without intending to do so. The emergent result was a world none of them chose fully.

This pattern explains one of Tocqueville’s most striking claims: democracy has grown up like a child abandoned in the streets. The metaphor can sound odd, since equality had advanced gradually over centuries; if the process was so slow, why call the result uneducated? The answer is that slow causes can produce revolutionary conditions. A river may rise inch by inch, but once it overflows the levee, the whole landscape changes. By Tocqueville’s age, equality had become strong enough to define the social order itself.

Tocqueville’s purpose is to educate democracy for liberty, the animating project of the book.

America as a democratic landscape

The first chapter of Part I can seem, at first glance, like a long geographical preface. For modern American readers, the detail may feel familiar or excessive. For Tocqueville’s French readers, it was neither: he was writing for people who had little sense of the United States as a continental reality, and before he could explain American democracy, he had to put America on their mental map.

Tocqueville’s landscape does more than orient the reader. It has causal force, which is a debatable claim. North America appears in his account as a physical setting unusually favorable to democratic development: vast, abundant, and conspicuously empty of feudal structure. There are no ancient castles, no inherited manorial order, no dense aristocratic past pressing down on every acre. His description of the Mississippi watershed becomes almost rhapsodic, a continental artery gathering the energies of an immense interior, land that seems to contain not only resources but destiny. A Smithian undercurrent — he almost certainly read Smith as well as Jean-Baptiste Say — runs through this geography: terrain shapes relative returns, returns shape occupations, occupations shape habits, habits shape institutions. In Tocqueville, political economy does not always announce itself, but it is frequently in the room.

The argument is not that geography determines democracy, but that democratic liberty takes root more easily where the material environment weakens aristocracy, encourages activity, disperses property, and rewards self-reliance. America’s physical world does not create its moral world, but it creates the conditions for it.

Occupation, possession, and the Lockean shadow

Tocqueville’s discussion of indigenous peoples reveals both his analytical power and his civilizational limits.

On one hand, he sees something real in indigenous societies that contrasts sharply with European aristocracy. He identifies forms of equality and liberty among indigenous peoples, and he notices that social distinction there does not resemble the rigid hierarchy of aristocratic Europe. His description can verge on romanticization, but it’s more than sentimental. He uses indigenous society as a mirror in which European inequality looks less natural than Europeans often imagined.

On the other hand, Tocqueville’s treatment of land and possession is deeply marked by European assumptions. His claim that indigenous peoples occupied the land but did not possess it rests on a theory of property in which agriculture performs the decisive act of appropriation (John Locke, call your office). To cultivate land is to make it one’s own. To hunt across it, dwell within it, or attach meaning to it in nonagricultural ways does not count in the same way.

Tocqueville’s foundation here is thoroughly Lockean. Property appears not merely as presence or use, but as improvement. Agriculture becomes the social technology through which land enters the world of recognized ownership.

He correctly treats property as a social institution, not just a physical relation between a person and a thing. It is a relation among persons concerning things, stabilized by norms, law, recognition, and enforcement. But Tocqueville’s framework also implies that peoples whose property systems do not resemble European agricultural property can be described as occupants rather than possessors. That distinction helped make dispossession legible as improvement.

That theme will matter more later in Tocqueville’s discussion of the “three races” in America. For now, it’s enough to notice that Tocqueville sees the institutional character of property clearly.

The point of departure

If Chapter 1 gives democracy a continent, Chapter 2 gives it an origin story. Tocqueville calls this the “point of departure” — the foundational conditions a society carries forward, shaping development long after the original circumstances have changed. Origins are not everything, but they are rarely nothing. We now call this institutional path dependence; Tocqueville was its theorist before the term existed.

His image of old chains hanging from the vaults of a building captures the idea precisely. Inherited institutions continue structuring society even after people forget what they once supported, a phenomenon known after Tocqueville’s time as Chesterton’s fence. Tocqueville’s point is subtler than conservatism, though. Old things should not be preserved because they are old, but because social practices contain knowledge that is not fully articulated — knowledge that cannot be recovered simply by reading statutes or constitutional texts.

Social mores (moeurs in French) are not designed or installed by statesmen like machinery. They emerge from the accumulated interactions of families, churches, towns, courts, markets, and political assemblies. Repeated practices become expectations. Expectations become norms. Norms shape judgment. Judgment shapes law. The line between custom and institution is, over time, nearly impossible to draw.

Tocqueville’s great insight is that constitutions rest on this pre-constitutional social order. Institutions begin as habits before they become constitutions. But habits themselves begin as lived practices before anyone thinks to call them political theory.

Why New England matters

Tocqueville places special weight on New England because the American point of departure appears there in its clearest form, but not without distortion, since New England is not all of America. He distinguishes North and South carefully from the start, preparing his later discussion of slavery’s destructive influence on Southern society.

Still, New England provides his central example. The settlers who came were not aristocrats transplanting feudal hierarchy to a new continent; the physical and institutional conditions of settlement made it hard to sustain the old noble order. Small holdings, local institutions, and the demands of a new landscape favored broadly distributed social and political participation.

The liberty that emerged was therefore closer to bourgeois and democratic liberty than to ancestral aristocratic liberty: the liberty of ordinary citizens governing local affairs, levying taxes, serving on juries, making rules, sustaining churches, founding schools, and acquiring the habits of public responsibility.

Aristocratic liberty belongs to orders, estates, and privileges. Democratic liberty must belong to citizens, which makes it more morally expansive, but also more fragile. If liberty belongs to everyone, then everyone has to learn how to practice it.

Sectarians who invented self-government

Tocqueville’s treatment of Puritanism refuses easy categories. The Puritans were not liberals; their moral legislation was severe, intrusive, and intolerant, punishing religious dissent and regulating private conduct in ways that rightly offend modern ideas of freedom.

Politically, though, they were astonishing innovators. Principles that many Europeans barely grasped in the seventeenth century were already practiced as facts in New England: popular participation in public affairs, free voting of taxes, accountability of officials, individual liberty, jury trial. The founders of New England were, in Tocqueville’s phrase, both ardent sectarians and impassioned innovators.

That pairing unlocks his account of American democratic liberty. Religion restricted some forms of conduct but formed moral habits — self-restraint, seriousness, literacy, discipline, covenant, obligation. Political liberty gave citizens responsibility for common affairs. The result was ordered liberty: freedom embedded in moral expectations and sustained by local practice, not freedom as expressive individualism.

Near the end of Chapter 2, Tocqueville names this juxtaposition as one of the book’s central claims: Anglo-American civilization arose from the combination of the spirit of religion and the spirit of liberty, two elements often opposed elsewhere. Religion gave moral structure to freedom; liberty gave public scope to religiously formed citizens. Each restrained the other from its worst tendencies, religion from consuming public life, liberty from dissolving into private license.

That lesson extends beyond seventeenth-century New England into Tocqueville’s time and today. Democracies need freedom, but freedom alone does not tell citizens how to use it well. Democracies need equality, but equality alone can morph into envy or conformity. Democracies need institutions, but institutions alone cannot generate the habits that make them work. Those habits must be formed before the institutions arrive. Before Madison, mores. Before formal institutions, habits. Before national government, towns. Before democratic theory, democratic practice.

Democracy’s promise and danger

Tocqueville’s attitude toward democracy is neither sentimental nor reactionary. He sees its virtues in enabling equality to elevate human dignity, weaken inherited domination, broaden opportunity, and draw ordinary people into public life. Democratic liberty can produce energy, responsibility, innovation, and moral seriousness.

But he also sees democracy’s natural vices. Equality can make people restless, envious, and conformist. Democratic citizens may become so absorbed in private affairs that they abandon public life. Majority opinion may become intellectually oppressive. Centralized administration may expand while citizens congratulate themselves on their independence. A people may remain free in small things while losing liberty in great ones.

This warning endures: democracy is not self-securing. It does not automatically produce liberty. The social condition that weakens aristocracy may also weaken the intermediate institutions, inherited obligations, and local attachments that help citizens resist centralized power. Equality liberates while it isolates. Democracy empowers ordinary citizens, but can also encourage conformity. Freedom expands choice, but does not by itself cultivate judgment.

The work of democratic civilization is thus to join equality to liberty. That work cannot be accomplished by constitutions alone. It requires law, religion, association, local self-government, property, education, and mores. Above all, it requires citizens who have learned, through practice, to govern themselves.

The beginning that keeps beginning

Tocqueville begins Democracy in America with equality because he thinks equality is the fact modern people must learn to govern. He turns to geography because democracy does not unfold in abstraction. It takes root in land, settlement, abundance, scarcity, distance, and opportunity. He turns to New England because institutions begin in practice before they become theory, and because democratic liberty depends on mores that no constitution can simply summon into existence.

The Introduction announces the democratic age. Chapter 1 gives that age a continent. Chapter 2 gives it a point of departure in religion, settlement, law, and local self-government.

At this early point in the text, Tocqueville’s deeper argument is already visible. Democracy is not just what happens when people vote. Democracy is what happens when equality reshapes the social world. Liberty is what happens only if democratic people acquire the habits, beliefs, and institutions necessary to govern themselves.

Tocqueville’s America is young, energetic, flawed, expansive, religious, commercial, practical, and restless. It is also morally contradictory from the beginning. It contains self-government and slavery, equality and dispossession, liberty and intolerance, local responsibility and national ambition. Tocqueville sees enough to admire it, enough to criticize it, and enough to treat it as a warning to Europe.

His question is still ours. Equality has remade the modern world. Democracy has become our common political language. Freedom remains our personal demand, but liberty requires the harder work of self-government. The democratic age would be saved, if at all, by learning how equality, freedom, and liberty can live together without devouring one another.

That task begins before politics as we usually define it, in the habits of ordinary life. It begins in the family, the town, the church, the school, the association, the jury, the meeting, the market, and the thousand small practices through which people learn to act with others.

Next time: Part I, Chapters 3,4,5

This summer marks the 250th anniversary of the Declaration of Independence.

That fact alone is astonishing: a republic born in rebellion against monarchy survived civil war, depression, world wars, constitutional crises, social upheaval, and the immense pressures of industrialization, globalization, and technological change. The American experiment has endured longer than many of the democracies that once dismissed it as unstable or naïve. I celebrate that endurance enthusiastically and gratefully.

And yet this anniversary arrives at a strange and dispiriting moment. Americans today often seem less capable of living together as fellow citizens than at almost any point in recent memory, as Lauren Hall has documented effectively (in two articles here and here). Our politics has become intensely polarized, but polarization alone is not the deepest problem. Democracies have always contained conflict, disagreement, and faction.

Contempt is the deeper problem.

Too many Americans increasingly regard political opponents not merely as wrong, but as morally illegitimate, intellectually corrupt, or fundamentally alien. Public life often feels less like democratic disagreement among fellow citizens and more like tribal warfare funneled into two camps and conducted through institutions that both sides increasingly distrust.

I share Lauren’s despair about this condition. But I also think the crisis runs deeper than contemporary politics, election cycles, or social media incentives. What we are witnessing is, in part, the erosion of civic virtue.

That phrase has acquired an unfortunate ideological coding in modern political discourse. “Civic virtue” is often interpreted as a conservative slogan or, worse, as a moralizing demand for conformity. I think that interpretation badly misunderstands the concept. Civic virtue is not partisan. It is not reducible to politics. And it is not fundamentally about agreement.

At its core, civic virtue rests on two simple but profound premises that are fundamental to the American project. First, every person possesses dignity as an individual human being. Second, human beings are social creatures who must live together in communities, cooperating in countless ways even amid disagreement. We rely on each other not only emotionally and culturally, but also materially and politically. We maintain institutions (informal and formal) together. We build infrastructure together. We govern together. We inherit constitutional orders together. A liberal democracy cannot survive without some shared commitment to those truths.

This idea lies at the heart of the American founding itself. The Declaration of Independence is not merely a legal document announcing separation from Britain. It is a moral claim about human equality and human dignity. The proposition that “all men are created equal” asserts that individuals possess inherent worth and therefore deserve liberty, protection under law, and recognition as moral equals. It is an aspirational statement of a set of core values and a yardstick for our moral improvement in achieving those values.

That tradition is fundamentally liberal in character. It joins liberty with dignity, rights with responsibility, and individual freedom with mutual obligation.

But we live in a political culture saturated with outrage, suspicion, and performative antagonism, so that moral foundation increasingly feels fragile. Many Americans now encounter fellow citizens primarily through nationalized ideological conflict rather than through local civic life, neighborhood institutions, religious congregations, voluntary associations, or community organizations. We increasingly experience one another as abstractions rather than as human beings.

Alexis de Tocqueville saw this danger nearly two centuries ago, and that recognition is one reason I want to reread him now. Revisiting Tocqueville during the semiquincentennial year can help reanimate the ideas and ideals embodied in the Declaration of Independence as an aspirational statement of values.

Alexis de Tocqueville (1805–1859) was a French aristocrat turned political thinker, historian, and statesman. He is best known for Democracy in America, a work that remains one of the most penetrating analyses of democratic society ever written, as Brendan McCord of the Cosmos Institute has argued recently. Tocqueville was a comparative social scientist avant la lettre, asking how modern democracy reshapes law, culture, religion, and everyday life.

Reading Democracy in America today can be almost unsettling because of how contemporary many of its warnings feel. Tocqueville admired American democracy deeply, but he also understood its vulnerabilities with remarkable clarity. He recognized that democratic equality could elevate human dignity while simultaneously weakening the habits and institutions necessary to sustain liberty.

Most importantly, he understood that democracy depends upon culture and character as much as upon constitutions and laws. Americans often assume that liberal democracy survives through formal institutions alone: elections, courts, constitutions, and procedures. Tocqueville thought those institutions mattered enormously, but he also believed they were insufficient in and of themselves. Free societies require mores, habits, associations, local institutions, and civic norms that teach citizens how to govern themselves. Democracy requires civic virtue.

Tocqueville famously observed the role of civil society in the young American republic. Today, many of the mediating institutions Tocqueville admired have weakened simultaneously: local journalism, civic organizations, churches, neighborhood associations, and geographically rooted communities. Americans participate less in shared local life while consuming more national political conflict. The result is a society filled with people who are hyper-politicized yet civically disconnected.

Tocqueville believed Americans historically counteracted some of the corrosive tendencies of democratic individualism through the density of their associational life. Americans formed clubs, churches, reading groups, charities, local organizations, and civic associations for nearly every conceivable purpose. These institutions did more than accomplish practical tasks. They taught citizens habits of cooperation, compromise, responsibility, and mutual respect. They taught people how to disagree while still regarding one another as fellow citizens.

The tragedy of our present moment is not merely that Americans disagree passionately. Democracies are supposed to contain passionate disagreement. The tragedy is that many citizens increasingly struggle to imagine political opponents as participants in a shared constitutional project, which is precisely why Tocqueville matters today.

Reading Democracy in America offers both historical insight and a kind of civic reorientation. Tocqueville approaches democracy neither with blind celebration nor with cynical contempt. He combines admiration with warning, hope with realism, and criticism with genuine affection for democratic life. He writes with humility. He even writes about technology, which you know I am going to want to discuss!

Tocqueville understood that democracy is fragile because human beings are fragile. Free societies depend upon habits of restraint and mutual recognition that no formal constitution can compel. A people incapable of respecting one another eventually becomes incapable of governing itself. That observation feels painfully relevant during this 250th anniversary, and it is one reason I think reading Tocqueville today is an important civic exercise.

At a time when contempt increasingly dominates public life, Democracy in America reminds us that liberal democracy requires something deeper than procedural rules or electoral victories. It requires citizens capable of balancing liberty with responsibility, individuality with community, and disagreement with mutual respect.

Tocqueville recognized both the greatness and the danger embedded within the democratic revolution in America. We would do well to read him again.

That is what I plan to do, and I invite you to join me.

I will be reading the copy I have, which is the University of Chicago Press Mansfield/Winthrop translation (2000), as well as Schleifer’s companion (2012). Electronic copies of Democracy in America are available at the Liberty Fund website, a 2010 new translation in four volumes. So if you want to read along with me and don’t have the same edition, the pagination won’t match but the chapter numbers should.

A small sentimental recollection: this volume of Tocqueville was one my father purchased on a visit to us in 2000, shortly after its publication, when it was one of the new publications featured on the front table at the Seminar Co-Op Bookstore at the University of Chicago (back when it was still in the basement of the old seminary building). I don’t know how much he read it; it doesn’t have any marginalia. I inherited it from him when he died in 2006.

Here’s my schedule starting next week, if you would like to read along and have a Tocqueville book club (which I would love):

Essay 1
Volume 1, Part 1, Chapters 1–2 (42 pages)

Essay 2
Chapters 3–5 (48 pages)

Essay 3
Chapters 6–8 (through “On the Federal Courts”) (37 pages)

Essay 4
Remainder of Chapter 8; Part 2, Chapters 1–3 (50 pages)

Essay 5
Chapters 4–5 (40 pages)

Essay 6
Chapters 6–8 (44 pages)

Essay 7
Chapter 9; Chapter 10 through the section on slavery (61 pages)

Essay 8
Remainder of Chapter 10 (73 pages)

Essay 9
Volume 2, Part 1, Chapters 1–12 (44 pages)

Essay 10
Chapters 13–21 (34 pages)

Essay 11
Volume 2, Part 2 (entirety) (56 pages)

Essay 12
Part 3, Chapters 1–15 (47 pages)

Essay 13
Chapters 16–26 (54 pages)

Essay 14
Part 4 (entirety) (38 pages)

Essay 15
Analysis, synthesis, and contemporary relevance

Please do join me!

Source: https://www.aboutamazon.com/news/aws/aws-data-center-inside

Most of us experience electricity as a subscription service. We move in, call the utility, set up an account, and expect the electrons to arrive. The power system is one of the great coordination achievements of modern life precisely because, for most of us, it disappears into the background. When it works, we do not notice it. When it fails, we notice nothing else.

That mental model does not scale. A data center requesting hundreds of megawatts is not another customer signing up for service. At that scale, electricity service becomes an infrastructure negotiation—and the problem is not just how much power data centers consume. The problem is how a large, fast-moving, uncertain new customer becomes a permanent enough fixture around which to build power systems.

This distinction becomes more important as artificial intelligence drives a surge in data center development and an expectation of ensuing demand growth. Public debate has focused on quantities: how much electricity data centers will use, how much new generation we will need, how much bills might rise. Those questions matter, but their focus on the shortage of electricity infrastructure misses a deeper problem: a shortage of institutional infrastructure.

Subscribe now

Beyond Plug-and-Play

Large new electricity demand does not simply appear on the grid. It has to be forecast, studied, planned, financed, physically connected, and then delivered reliably hour after hour. These steps need much more than engineering; they require credible commitments.

A data center developer may want the utility to reserve hundreds of megawatts of future capability, while the utility may have to build substations, transmission upgrades, and transformers before the data center is fully operating. Regulators must then decide whether those investments are prudent, who should pay for them, and what happens if the project is delayed, uses less electricity than expected, or never materializes at all. The grid-owning utility has to make durable investments under uncertainty (and so does the hyperscaler). Better forecasting can reduce that uncertainty but not eliminate it. The harder institutional question is how to write rules and contracts that make investment possible without asking ordinary customers to become involuntary insurers of speculative demand.

Historic trends. New uncertainty.

Data centers are not the first large customers utilities have ever served. The electricity system has a long history with energy-intensive industrial loads—steel mills, aluminum smelters, chemical plants, auto factories—that shaped the 20^th-century economy and power grid. These customers received special contracts, industrial rates, and sometimes interruptible service in exchange for paying lower prices. But many historical large loads were anchored to physical production processes, local resource bases, and long-lived industrial assets, arriving on timescales that aligned reasonably well with utility construction cycles.

Data centers scramble that model by combining features that collectively strain the inherited utility planning process. They arrive in large concentrated increments, with a single campus demanding hundreds of megawatts and a handful of projects transforming a utility’s load (demand) forecast. They are fast, with developers caring intensely about “speed to power” while transmission upgrades, substations, and generation procurement run on slower clocks. They are mobile, shopping across utility territories and grid regions, with each site inquiry showing up as potential future load before a developer commits to building anywhere. They are uncertain—some backed by major companies with real demand, others contingent on financing, permits, chips, or favorable tax treatment. And they may be flexible, though not automatically, with some computing workloads more amenable to shifting across time than others.

A new paper on large-load tariffs and load forecasting, by Angela Navarro and Molly Knoll for the National Association of Regulatory Utility Commissioners, captures the key shift: Historical load forecast errors were driven by population growth, marginal changes in end-use consumption, and macroeconomic factors that were diffuse and incremental. Today’s large, discrete loads are localized, infrastructure-intensive, and often seeking service faster than utility planning cycles can accommodate. The old forecasting world was about gradual demand trends. The new forecasting world is about project-level shocks. Poor load forecasting, welcome to your main-character era.

When the forecast becomes the problem.

For much of the 20^th century, load forecasting was necessary but quiet, most noticeable when something went wrong, because demand growth was generally steady. The data center boom changes that. In regions with substantial data center interest, the load forecast becomes contested terrain.

The underlying difficulty is the phantom load problem. If developers request service in multiple locations before deciding where to build, utilities and regulators may see more proposed demand than will ever materialize. PJM, the grid operator in the mid-Atlantic region, has responded by requiring firmer commitments for near-term forecast years while treating longer-term projects as less certain. ERCOT in Texas adjusted its large-load forecasting assumptions after observing an average project delay of 180 days and data centers’ consumption of less than half the capacity originally requested for study.

A forecast is easy to inflate when someone else bears the cost of relying on it. Forecasting is both technical and strategic, with developers wanting to reserve capacity, utilities potentially expanding their rate base, economic development officials courting investment, and consumer advocates all having stakes in the outcome. In such an environment, a load forecast is a claim on future infrastructure, and when forecasts are wrong, the consequences are expensive. Transformers, substations, and transmission upgrades cannot be returned with a receipt and a sheepish apology. Bad forecasts become bills.

The real problem is credible commitment.

This observation brings us to the central institutional economics question: Who should bear the risk created by uncertain large-load forecasts?

The easy answer says that big technology companies should pay more because they are large and profitable. This claim has political force but is not quite the right analytical frame. The better question is what commitments a large customer should make before everyone else is asked to commit resources based on the forecast.

When a utility constructs infrastructure to serve a large new customer, it makes investments that are durable, capital-intensive, and difficult to repurpose. The relevant concept from transaction cost economics is asset specificity: A transformer, substation, or transmission upgrade built to serve a particular large load in a particular location may have little value if that load never materializes. This characteristic transforms a forecasting problem into a contracting problem.

The structure is a classic setting for what are known as contractual hazards. One party must make a durable investment before the future is fully known; the other party has better insight into its own plans and commercial prospects, but it, too, has durable investments on the line. Once the investment is sunk, bargaining positions change. MIT professor Paul Joskow’s work on coal contracts showed that contract duration tends to increase when relationship-specific investments matter most; when one party must invest in durable assets to serve another, longer contractual commitments reduce the risk of being left exposed. Nobel laureate Oliver Williamson described the need for “hostages” to ensure credible commitment: A deposit, collateral requirement, minimum bill, or termination fee gives the customer something at stake, making its forecast more than cheap talk.

The relevant question is practical: What kind of contingent contract can make investment possible when the future is uncertain? When parties cannot know the future, they can still agree in advance on what happens under different future conditions. If the data center arrives on schedule, here is how costs will be recovered. If it builds more slowly, here are the minimum charges. If it leaves early, here is the exit fee. If it offers flexibility during grid stress, here is the compensation.

That is not punishment. It is governance.

Turning forecasts into commitments.

Properly understood, a large-load tariff is a rulebook for joining the grid when a customer is large enough to alter system planning, infrastructure needs, and cost allocation. These tariffs may require long-term contracts, minimum billing commitments, collateral, exit fees, phased ramp schedules, and notice requirements. The basic institutional function is consistent: Large-load tariffs convert forecasts into commitments.

The new National Association of Regulatory Utility Commissioners paper adds an important insight. Tariffs are more than customer-protection mechanisms—they can also improve forecasting. Minimum demand obligations, collateral requirements, phased load ramps, eligibility thresholds, contract lengths, and exit fees all contain information about the timing, magnitude, persistence, and credibility of large-load requests. Navarro and Knoll propose that regulatory commissions use tariff terms directly in load forecasting, applying them to baseline planning assumptions and sensitivity analyses, and updating forecasting assumptions as actual queue and realization data accumulate. This approach reveals that a large-load tariff is more than a price, it is a forecasting institution.

The components of a well-designed tariff are not random regulatory appendages. Long-term contracts address duration. Minimum bills address downside demand risk. Deposits and collateral bond the customer’s promise. Exit fees protect against costly early departure. Ramp and build schedules make timing uncertainty contractible. Curtailment provisions that lay out the conditions under which the large load customer will reduce its demand define what flexibility means operationally. The goal is not a complete contract specifying every possible future state; no tariff can perfectly anticipate AI demand trajectories, chip constraints, or capital markets. The more modest and useful goal is contingent contracting: specifying what happens under the most important foreseeable deviations from the forecast.

The experience of AEP Ohio illustrates this well. After regulators approved and the utility adopted revised data center tariff terms in July 2025—including 25-megawatt eligibility requirements, an 85 percent minimum billing obligation, 12-year contract terms, and termination fees equal to minimum charges for 36 months—the interconnection queue fell from 30 gigawatts to 13 gigawatts. The missing 17 gigawatts was not necessarily “phantom” in any simple sense. But once capacity requests became costly to maintain, the queue began to contain better information, filtering speculative entries before utilities built around them. Better load forecasting can tell us which futures are more likely. Large-load tariffs help decide who is responsible when the likely future turns out to be wrong.

Can the cloud learn to bend?

Requiring data centers to pay for the infrastructure they require is the necessary first principle. Existing customers should not subsidize speculative load growth. But stopping there misses the question of whether data centers might reduce the infrastructure they require by becoming more responsive to grid conditions.

Not all megawatts are created equal. A customer requiring firm, round-the-clock service imposes different costs from one able to reduce demand during scarcity, shift computing tasks across time, or coordinate operations with grid needs. Some workloads—AI training, batch processing, cooling systems—may offer flexibility that latency-sensitive inference cannot. This flexibility is not automatically valuable, however. It has to be defined, contractualized, and verified. A data center claiming flexibility should have to specify what that means operationally: how much load can be reduced, how quickly, for how long, and at what compensation. Flexibility becomes useful to the grid only when it is reliable enough to plan around, making this too a credible-commitment problem.

Large-load tariffs could become tools for institutional discovery, revealing which customers are firm, which are flexible, and which are speculative—making their characteristics visible and actionable rather than forcing public utility regulators to guess.

The larger lesson.

The data center boom did not create weaknesses in electricity regulation. It removed the luxury of ignoring them.

Load forecasts are no longer technical appendices to utility plans; they are central to economic development, reliability, affordability, and decarbonization. When a forecast becomes the basis for billions of dollars of infrastructure on both sides of a transaction, the assumptions behind it deserve scrutiny: which projects are included; how credible they are; what commitments customers have actually made; and how much projected demand is firm, flexible, speculative, or duplicative.

But forecasting is only part of the solution. Forecasts reduce but do not eliminate uncertainty. Once we better understand the likely timing, magnitude, and persistence of large new loads, we still need rules assigning responsibility when expectations prove wrong. Data centers are neither villains nor ordinary customers. They are large, sophisticated, fast-moving actors whose demand can reshape the electricity system, and good institutions should be able to tell the difference between those creating real value and those claiming capacity without credible commitment.

Better forecasting tells us which futures are more likely. Better contracting tells us what happens when the actual future is different. The first reduces uncertainty; the second governs the uncertainty that remains.

You cannot just plug in a data center. But with better rules, better forecasts, and better commitments, perhaps we can learn how to connect one.

First published at Dispatch Energy at https://thedispatch.com/newsletter/dispatch-energy/data-center-electricity-use-regulators-utilities/

Source: Yet another productive conversation I had with ChatGPT.

PJM’s recent white paper, Powering Reliability Through Market Design, does not solve the problem underlying its title. But the paper is extremely valuable because it acknowledges that the foundational assumptions of organized capacity markets no longer fit the environment in which those markets operate, and does so more openly than most institutional discussion in this industry. That candor is significant.

The striking thing about the paper is how familiar the concerns are. Reading the document, I recognized the same institutional problems that were already visible nearly two decades ago during the early expansion of organized capacity mechanisms. In a chapter on resource adequacy and capacity markets in my 2009 book Deregulation, Innovation and Market Liberalization (Chapter 7), I argued that those mechanisms were emerging largely because organized wholesale markets had combined two institutional design choices that interacted badly: price caps and suppressed demand responsiveness. The resulting market structure then required administrative supply-side mechanisms to compensate for demand-side flexibility it had institutionally prevented from developing.

Subscribe now

The Missing Money Problem Was Never Purely Technological

The conventional justification for capacity mechanisms is familiar. Electricity markets cap scarcity prices to mitigate market power and political backlash. Most retail customers face flat or quasi-flat retail rates rather than real-time prices. Demand therefore becomes highly inelastic precisely during scarcity conditions, limiting the ability of energy market revenues alone to support investment in resources needed primarily during peak reliability events. Capacity mechanisms emerge to supply the missing money — the revenue shortfall created by combining administrative price caps with inelastic retail demand.

This logic has long underpinned organized capacity mechanisms throughout wholesale power systems. But even in 2009, this explanation struck me as incomplete because it treated demand inelasticity as though it were an unavoidable technological characteristic of electricity systems rather than partly the consequence of institutional design choices. As I wrote then, capacity mechanisms are “an attempt to remedy the consequences of a distortionary policy with another distortionary policy.”

That framing remains central to understanding today’s resource adequacy debates.

Retail price caps, limited dynamic pricing, barriers to demand aggregation, insufficient integration between wholesale and retail markets, and governance structures designed around passive load all suppress the emergence of active demand participation. The resulting lack of demand elasticity creates the appearance that administrative supply-side procurement is necessary to maintain reliability; this appearance, though, is at least partly endogenous to the institutional architecture itself.

The irony is that organized markets have spent two decades building increasingly elaborate administrative mechanisms to compensate for the absence of demand responsiveness while simultaneously preserving many of the institutional arrangements that suppress demand responsiveness in the first place. PJM’s paper indirectly acknowledges this reality throughout. Its discussion of the credibility trap is especially illuminating: PJM correctly recognizes that scarcity pricing during tight conditions is economically rational, yet those same scarcity prices create political backlash because large portions of load remain exposed to volatility without corresponding flexibility or meaningful hedging arrangements. Policymakers then intervene, investors discount the credibility of future market revenues, investment becomes more hesitant, and scarcity persists. Lather, rinse, repeat.

This dynamic reveals a problem deeper than insufficient supply-side incentives: a broader institutional problem involving governance, risk allocation, and the persistent underdevelopment of active demand.

The Administrative Demand Curve as Epistemic Substitute

One of the most revealing features of organized capacity markets is their reliance on administrative demand curves. PJM, ISO-NE, and NYISO all eventually adopted downward-sloping administrative demand curves intended to approximate the value of future capacity resources. Yet the fundamental problem with these curves has always been that they substitute for actual preference revelation.

As I argued in 2009, “the artificial capacity product has no intrinsic value to customers,” because what customers actually value is reliability itself, not an engineering reserve margin target. The resulting administrative demand curves therefore become bureaucratic approximations of preferences that markets themselves are institutionally constrained from revealing directly. At best, such curves are rough estimates derived from incomplete information; at worst, they become bureaucratic substitutes for the decentralized discovery processes that active markets would otherwise facilitate.

My recent work on RTO governance argues that RTOs are best understood not simply as market operators but as governance institutions managing a congestible common-pool resource through rule-defined participation and exclusion. These institutions are non-replicable — no competing alternative can simply be constructed — meaning that the rules defining who participates, on what terms, and bearing what obligations become the primary margin of institutional adaptation rather than competitive pressure from outside.

That framing exposes a basic epistemic limitation: the distributed knowledge embedded in modern electricity systems — about flexibility, reliability preferences, and operational capabilities — is precisely what centralized administration struggles to incorporate, yet organized market governance still largely reflects the institutional assumptions of the 1990s restructuring era: centralized thermal generation, homogeneous demand, and limited customer participation. Administrative constructs increasingly substitute for the decentralized coordination they were never designed to replace.

Administrative demand curves substitute for actual demand expression, accreditation methodologies substitute for market discovery regarding reliability value, centralized reserve targets substitute for differentiated reliability preferences, and capacity constructs substitute for forward contractual coordination. The more technologically heterogeneous the system becomes, the more administratively demanding this substitution becomes, requiring more regulatory complexity to compensate for the coordination it cannot fully supply.

The Accreditation-Penalty Trap

Recent work by Ke Xin Zuo, Joshua Macey, and Jacob Mays identifies a structural dilemma at the institutional core of capacity market design. Capacity markets require non-performance penalties strong enough to create honest performance incentives, but penalties strong enough to be economically efficient expose suppliers to catastrophic financial risk when outages are correlated, as they were during Winter Storms Elliott and Uri. Under those conditions, high penalties drive generators toward default, and bankruptcy itself becomes a hedge against non-performance. Softening the penalties restores the problem they were designed to solve: suppliers again overstate their reliability contributions, and load again bears the consequences.

The institutional response has been to develop increasingly elaborate accreditation methodologies — marginal effective load carrying capacity (ELCC) estimates, performance assessment hours, seasonal accreditation — substituting regulatory constraints on position size for the market-based performance discipline that adequate penalties would otherwise provide. But accreditation introduces compounding challenges of its own. Optimal values depend on the resource mix at the time of the capacity auction, not the mix that will exist at delivery. Correlated outages are rare by design, making the hardest-to-model events the most consequential ones. And adopting improved methodologies requires navigating multi-veto stakeholder governance structures that move more slowly than the resource mix is changing.

The result is a mutually reinforcing administrative expansion: weak incentives require elaborate accreditation, elaborate accreditation requires complex governance, complex governance requires long implementation timelines, and long timelines mean that accreditation methods chronically lag the technical realities they are designed to measure.

PJM’s paper largely takes this institutional inheritance as given. Zuo, Macey, and Mays draw the more fundamental inference: reduce reliance on administrative accreditation by restoring economic incentives through stronger price signals in the energy and ancillary services markets, using accreditation primarily to assess credit risk rather than substituting for performance incentives that markets should supply directly. That inference maps directly onto PJM’s Path C proposal, as well as onto the vision of integrated spot and forward energy markets that I argued for in 2009.

Hedging, Risk Allocation, and the Architecture of Markets

Farhad Billimoria’s recent Substack article on hedging architectures reframes organized wholesale markets fundamentally as systems for allocating and managing risk rather than merely as systems for dispatching generation resources. That perspective clarifies what PJM’s three proposed pathways actually are: alternative hedging and risk-allocation architectures, not just alternative market designs. The Stabilized Markets pathway (Path A) relies more heavily on long-term centralized hedging structures, requiring load to be covered by forward commitments before the spot auction, preserving the capacity market while insulating consumers from its price volatility. Differential Reliability (Path B) reallocates reliability risk across customer classes or geographic areas, making explicit the allocation that currently happens implicitly and arbitrarily. The Energy Market Transition pathway (Path C) shifts revenue recovery toward the energy and ancillary services markets while relying on long-term energy contracting to provide the consumer protection that stable administrative capacity payments currently offer.

Billimoria correctly observes that no one-size-fits-all architecture exists; the appropriate design depends on regulatory structures, retail market development, and consumer engagement in each jurisdiction. This insight echoes arguments I advanced in 2009 about the importance of integrated spot and forward energy markets and the use of financial call options as transitional resource adequacy mechanisms — offering revenue certainty to investors while creating direct, market-based performance incentives rather than administrative ones (an idea I credit to the brilliant Shmuel Oren). Forward financial arrangements are constitutive of resource adequacy.

But even sophisticated discussions of hedging architectures like Farhad’s tend to understate one dimension: the importance of demand responsiveness itself. The problem is more than incomplete financial markets; it encompasses incomplete institutional participation. Technology increasingly enables active, flexible demand participation — smart devices, automated controls, distributed storage, EV charging, flexible industrial processes, and advanced metering all expanding the feasible scope of decentralized responsiveness — yet governance structures, retail rate design, participation rules, and market institutions still largely treat demand as passive load rather than as an active reliability resource, artificially inflating the scale of administrative procurement needed to compensate.

Differential Reliability and the Institutional Transition

Perhaps the most consequential element of PJM’s paper is its discussion of Differential Reliability — significant not primarily as an operational proposal but as an implicit acknowledgment that reliability preferences and operational capabilities are becoming irreversibly heterogeneous.

Some customers may value extremely high reliability and willingly pay for it, while others may prefer lower-cost service with greater exposure to curtailment or dynamic pricing. Some loads may be highly flexible and technologically capable of rapid adaptation (including the hyperscale data centers whose load growth is driving the current situation) while others remain relatively inflexible. This heterogeneity connects directly to arguments I made in 2009 regarding differentiated reliability products and the possibility of technology enabling what I called “privatizing many aspects of reliability”, drawing on the vital, seminal work on priority service of Chao and Wilson (1987).

Once differentiated reliability becomes institutionally feasible, the logic of administratively pooled uniform reliability begins to weaken. Demand no longer appears merely as passive load requiring protection through centralized reserve procurement. Instead, demand itself becomes an active reliability resource capable of dynamic adaptation in response to prices, contractual arrangements, and system conditions, changing the structure of the resource adequacy problem fundamentally. The central question becomes less about determining reserve margins administratively for undifferentiated passive demand and more about creating governance institutions enabling decentralized coordination among heterogeneous participants with differing capabilities, preferences, and risk tolerances. That governance transition is difficult, requiring changes in retail rate structures, market rules, legal obligations, property rights, and technology deployment. But the mounting strain visible in organized capacity markets suggests that avoiding this transition indefinitely may prove more difficult still.

A Bridge, Not a Destination

One argument from my earlier work seems especially relevant today: capacity mechanisms, if necessary at all, should function as transitional mechanisms rather than permanent fixtures, bridging toward more integrated spot and forward markets with active demand participation, clearer property rights, and richer forms of decentralized coordination. Specifically, I proposed that regulators stipulate milestone-based sunset provisions for capacity markets, reducing their scope as the volume of forward commitments in financial markets approaches the desired reserve margin.

That concern now appears even more pressing than it did then.

Administrative capacity mechanisms tend naturally toward institutional entrenchment, with entire business models, stakeholder coalitions, regulatory processes, and governance structures evolving around them, making the mechanism originally intended as a transitional response to incomplete market development into an obstacle to further institutional adaptation. As I said in 2009, imagine if 1850s law had dictated the existence in perpetuity of a capacity market for the production of whale oil. That deliberately provocative analogy has a serious underlying point: institutional resilience requires an ability of governance structures to evolve, adapt, and sometimes disappear as technologies and coordination mechanisms change.

PJM’s white paper represents an important step forward in acknowledging the depth of the current institutional strain, opening space for a more fundamental conversation than this industry has generally been willing to have.

The strain affecting organized wholesale markets arises both because generation investment has become more difficult and because the institutional framework continues suppressing decentralized demand responsiveness, differentiated reliability preference expression, dynamic coordination, and risk-sharing that modern technologies increasingly make feasible. Relaxing those suppressive institutional constraints — through retail rate reform, expanded demand participation pathways, clearer property rights, and milestone-based governance transitions — goes to the institutional root of the resource adequacy problem.

Continuing to add administrative layers to compensate for institutional constraints that could instead be removed is not a sustainable strategy. It is, at best, a bridge. And bridges built without any plan for what comes after them tend to become permanent fixtures.

Source: An iterative conversation about Michael’s series with ChatGPT

Michael Lee’s Distributed Grid series reads like a field report from the institutional frontier. Lee, formerly CEO of Octopus Energy US, is writing from inside the practical problem that has animated my work for decades: how do we build a power system in which customers, new technologies and resources, flexible loads, and digital control and automation become working parts of a more adaptive grid? Reliability, affordability, and abundance will depend increasingly on that coordination.

Lee’s answer, across the first seven essays in his planned ten-part series, is that the next utility will not look like the old utility. It will look more like a platform business — a coordinator of capabilities, transactions, information, contracts, and performance across a dynamic network of resources the utility does not necessarily own. That vision gives operational specificity to a broader institutional-economic transition I have been writing about for over two decades. The grid’s future is decentralized coordination.

Subscribe now

Lee begins with the utility business model. His first essay makes the crucial point that the investor-owned utility is built for a different era because its core financial incentive still rewards capital deployment more than customer value. Traditional cost-of-service regulation does not simply allow utilities to build; it creates a business model in which building is the primary path to earnings. New substations, feeders, transformers, and other capital assets enter rate base. Less visible substitutes — flexible demand, distributed batteries, non-wires solutions, better information, or third-party coordination — may reduce total system cost but generate no equivalent durable earnings stream.

Lee is careful, and rightly so, not to frame this as a morality play. Utility executives and employees are not villains; they are organizations responding to incentives. The rate-base model did what it was designed to do: it financed a capital-intensive universal network. The problem is that the task has changed. Rather than an institutional framework that induces building large assets, we need an institution capable of discovering when building is the best answer, when operating differently is, and when customers and third parties can solve problems more cheaply and quickly.

This framing is where Lee’s practitioner critique meets the knowledge problem perspective. In my 2009 book, Deregulation, Innovation and Market Liberalization, I argued that the traditional natural-monopoly model is inherently static and poorly suited to a system characterized by technological change, heterogeneous agents, diffuse private knowledge, and evolving institutions. The fundamental economic problem in electricity is not allocating known resources to known uses. It is coordinating the actions and plans of people and organizations who know different things, value different things, and face changing technological possibilities.

Lee’s second essay, on performance-based regulation, pushes that analysis further. PBR is often presented as the corrective to cost-of-service regulation: pay utilities for outcomes rather than inputs. Lee’s contribution is to show why that reform, while directionally right, is often too small to change behavior. If a utility can earn large, durable returns from rate-base growth but only small, episodic rewards from performance incentives, the basic earnings logic remains intact. A modest performance incentive sitting beside a much larger capital incentive decorates the old model with a new ribbon. It does not transform the organization.

That diagnosis connects to a central finding in the economics of rate-of-return regulation: cost-of-service rules encourage capital investment even when capital is not the best path to reliability, resilience, affordability, or innovation. Lee adds the organizational mechanism, showing that a utility optimized around capital deployment will have planning, engineering, accounting, and managerial routines reinforcing that deployment. Changing the metric is not enough. The business model has to make operational excellence, flexibility procurement, and customer value financially consequential.

Lee’s third essay applies cost causation to one of the most important current stress tests for the grid. He argues that data centers should fully pay their own way, particularly when new large loads drive upstream transmission, substation, or distribution upgrades. His core intuition is sound: if the system met reliability standards before the new load arrived, upgrades required to maintain reliability afterward are caused by that load.

I agree with Lee that cost causation should discipline the data center debate, although I would put more weight on the institutional problem these loads reveal. Data centers expose a pacing mismatch: they can be built in 18–24 months, while major grid infrastructure often requires 7–10 years. That mismatch is a coordination problem under uncertainty; ignoring workload shifting in time and place, UPS-based frequency response, hybrid on-site generation, and other capabilities that can reduce system stress and upgrade costs, just makes it worse.

Here the Lee framework and my data center flexibility work converge. The question is not whether data centers should receive subsidies through socialized grid costs — they should not. The question is whether our rules can distinguish between large loads imposing rigid peak demands and large loads capable of offering flexibility, speed, and grid support. Cost causation and flexibility valuation have to work together, or we will either socialize too much cost or push customers toward inefficient bypass.

Lee’s fourth essay turns from data centers to what he sees as the larger opportunity: the distribution grid. Public debate remains fixated on bulk generation, long-distance transmission, and large loads, which are important. But the distribution system is where customers experience reliability, where many costs accumulate, where DERs connect, and where local constraints can often be resolved by local capabilities.

Lee emphasizes distribution-system load factor, observability, and topology. Distribution infrastructure is built for peaks occurring only a small fraction of the year. Without real-time topology, granular metering, and operational visibility — all achievable now through digital technologies — DERs remain blurry objects on the edge of the system rather than usable grid resources. Engineering conservatism and capital bias compound each other under poor observability, producing more iron rather than better coordination.

Lee’s fifth essay is deliberately pro-DER while also anti-hype. He is bullish on distributed energy resources, bearish on the infrastructure around them, and that distinction has meaning. Rooftop solar, batteries, electric vehicles, smart thermostats, and flexible devices can create system value, but not by existing heroically in PowerPoint slides. They need telemetry, data access, locational compensation, workable interconnection processes, OEM cooperation, and distribution-level price signals. Without that institutional and technical infrastructure, DERs remain potential rather than capability.

This institutional gap explains why flexibility remains the missing economic category in so many electricity debates. The old system valued energy and capacity. The emerging system also has to value capabilities: the ability to shift, reduce, inject, absorb, store, substitute, and orchestrate electricity use across time and space. The analysis must shift from a resource’s physical capacity to its capability to perform a specific service at a specific time and location, treating flexible demand not as a burden to be satisfied but as one of the system’s most valuable resources. What matters is not just how many kilowatts exist, but what they can do, how fast, and where they are in the network.

Lee’s sixth essay identifies the institutional precondition for this world: transparency as an operating model, not public relations. Utilities ultimately sell trust because customers cannot see electricity, often lack competitive alternatives, and receive bills whose causes are opaque. Distribution constraints, hosting capacity, feeder loading, engineering assumptions, interconnection queues, and performance data have to become legible to qualified participants if third parties are going to propose alternatives and if regulators are going to evaluate whether utilities are delivering value.

This necessity connects directly to what I have called the pacing problem in electricity. Technologies are changing faster than regulatory institutions can metabolize them. DERs, data centers, automation, AI, batteries, EVs, and digital control systems are evolving through modularity and recombination, while regulatory processes still move through rate cases, prudence reviews, tariff filings, and contested proceedings. Institutions need stability, but they also need adaptive capacity — otherwise the system gets ossification or chaos, and neither keeps the lights on.

That brings us to Lee’s seventh essay, where the platform vision becomes explicit (a vision my co-authors and I analyzed in our “From Airbnb to Solar” paper (Theisen, Kiesling, Munger (2022)). He imagines a distribution planning engineer facing a feeder constraint. In the traditional model, the engineer proposes a substation: capital investment, rate base, familiar process. In the platform model, the engineer opens a constraint marketplace. Storage developers, aggregators, flexible load providers, and conventional infrastructure options compete to solve the constraint on comparable terms, with each option evaluated on cost per unit of reliability delivered rather than on how much capital it adds to rate base.

That example is doing a lot of work. Converting the utility from a builder of owned assets into a coordinator of solutions creates a way for the utility to earn by saving customers money rather than by expanding rate base, treating third-party resources as complements rather than irritants, and recognizing that the cheapest reliable solution to a grid problem may sit in software, customer behavior, distributed storage, reconductoring, or traditional capital investment. The point is not to privilege DERs over wires. The point is to discover the least-cost, best-performing portfolio of capabilities.

In my 2020 paper “Plug-and-Play, Mix-and-Match: A Capital Systems Theory of Digital Technology Platforms,” I argued that digital platforms create value through modularity, scalability, and adaptability — a stable core, standardized interfaces, heterogeneous edge participants, governance rules, and mechanisms for matching capabilities to needs. That architecture applies directly to the grid edge: devices must communicate, market participants must know what services are needed and what those services are worth, and utilities must have incentives to publish constraints and treat edge resources as working parts of the system.

The hard part is that the wires monopoly still sits at the center of this architecture. Drawing on the Baxter/Bell Doctrine — that regulated monopolists have both the incentive and the opportunity to distort adjacent competitive markets — I have argued that the platform utility vision requires structural discipline. The utility may coordinate the platform, but absent regulatory constraint, using network control to foreclose competitive entry at the edge is the predictable response.

Lee’s series is important because it gives operational language to a transition that institutional economists have been analyzing from a more theoretical angle. He is describing the grid from the customer and distribution edge outward; my work has described the same transition from institutions and coordination inward. Those perspectives meet at the same point: a power system in which reliability, resilience, affordability, and decarbonization emerge not from suppressing decentralized action but from coordinating it.

The distributed grid is not a gadget story. It is an institutional story. The technology now exists to make millions of small decisions useful to the system. The question is whether regulation, markets, and utility business models will let that knowledge into the room.

Since AI development’s acceleration in late 2022, one question has loomed over electricity policy: can the decidedly twentieth-century electric utility industry build infrastructure fast enough to meet the speed-to-power expectations of hyperscalers and AI companies? Data centers can be planned, financed, and built on timelines that make utility planners blink twice and reach for another cup of coffee. Transmission lines, substations, turbines, interconnection studies, certificates, permits, and rate cases do not move at software speed.

That timing mismatch produces a second question: can data centers reduce their electricity use when the grid is stressed? The answer is “sometimes,” which is both less satisfying and more analytically useful than yes or no. The better question is what kinds of flexibility different types of data centers can provide, under what conditions, at what cost, with what effect on reliability, and through what institutions.

Subscribe now

Flexibility as Elasticity

Flexibility is fundamentally an economic question of elasticity — how responsive electricity demand is to changes in price, grid conditions, utility instructions, or reliability events. Infrastructure is a question of supply elasticity: how responsive quantity supplied is to changes in expected demand and prices.

Embedded in both is the variable of time. Alfred Marshall’s fish-market example is useful: in the immediate run, supply is fixed by the boats and fishers already at sea; in the short run, more fishers may go out with the same number of boats; in the long run, people can build more boats. Electricity infrastructure has the same structure, albeit with fewer nets and way more acronyms.

For data centers, many important questions are long-run: how much generation, transmission, and distribution infrastructure should be built, and how energy-efficient can chips, servers, cooling, and data center designs become? Other questions are short-run: where does the grid have unused capacity, during what hours, and at what locations? Can digital load move across time or place before new infrastructure arrives?

The locational dimension is crucial. Grids have different conditions not only at different times but at different places at the same time. At 5 p.m. on a hot August afternoon, PJM may be near its operating limits while cooler weather in the Midwest leaves MISO with excess capacity. A company with data centers in both regions may shift compute from PJM to MISO, maintaining service while reducing stress on the constrained grid and lowering its energy cost.

Defining Flexibility

That example points to a definition:

The four moving parts are timing (moving work from a stressed hour to a less stressed one), location (shifting workloads between regions or facilities), magnitude (reducing total power draw), and source (drawing from grid, batteries, fuel cells, or on-site generation). The definition’s final phrase matters because a data center is not an aluminum smelter or an irrigation pump — some digital work can move; some cannot. Uptime, latency, customer contracts, and quality-of-service obligations all matter. For a good definitional framework for flexibility, see this impactECI February 2026 report.

Flexibility is therefore broader than curtailment. Curtailment (reducing consumption below the level that otherwise would have occurred) is one form of flexibility, but flexibility also includes shifting, substitution, self-supply, and orchestration. A data center can be flexible at the meter while keeping computation running.

That distinction matters for business models. OpenAI, a pure-play AI company, and AWS, a cloud hyperscaler, both operate large compute infrastructure but face different control problems. A pure-play AI company has direct authority over its training workloads — deferring training, prioritizing urgent inference, slowing fine-tuning, or shifting compute geographically when software architecture allows. In contrast, AWS provides cloud capacity to many customers (including AI companies like Anthropic) whose workloads it cannot simply pause for the good of PJM. AWS flexibility comes less from curtailing customer workloads than from managing infrastructure and contractual portfolios: batteries, on-site generation, internal workloads, opt-in customer products, regional orchestration, and meter-level operating envelopes. The utility may see a 500 MW load request, but inside that request are layers of property rights, contracts, control rights, software, and reputational risk.

Data center flexibility is thus not a generic attribute. It depends on workload, ownership, contracts, design, location, telemetry, incentives, and institutional rules. EPRI’s 2025 Grid Flexibility Needs and Data Center Characteristics paper distinguishes data centers by size, reliability requirements, workload, and ownership model, emphasizing that flexibility requires a common taxonomy among power providers, grid operators, and data center stakeholders.

Evidence on Flexibility

Over the past two years, a body of work has clarified these distinctions. The Norris et al. (2025) Rethinking Load Growth analysis asked how much large flexible load the existing U.S. power system could accommodate by tapping latent grid headroom, finding that 76 GW of new load could be integrated if those loads could curtail for only 0.25 percent of maximum uptime. The result rests on important assumptions and does not eliminate the need for transmission, ramping, and reserve analysis, but it suggests small amounts of well-timed flexibility can have outsized capacity value — unsurprising in a nonlinear complex system. Subsequent work at Duke University has continued to develop these analyses.

EPRI’s DCFlex initiative contributes a power-system taxonomy, framing flexibility as the ability to adapt across time scales from seconds to decades and identifying where flexibility comes from inside data centers: compute assets, balance-of-plant systems, and power assets. The framework moves the conversation from “can data centers turn down?” to “which subsystems can provide which services under which rules?”

ACEEE’s 2025 and 2026 reports widen the aperture. Data center policy should not be only about emergency demand response “programs”, a category into which too many imaginative demand response opportunities have been shoehorned. Efficiency remains the first demand-side resource; improving chips, algorithms, cooling, software, and utilization reduces the baseline from which flexibility begins. Data centers, ACEEE reminds us, are the flashiest part of a larger load-growth problem driven by electrification, industrial development, and manufacturing, not the whole circus.

GridLab and Telos take the question into resource planning. Their NV Energy case study, reviewed in LBL’s 2025 large load literature survey, found that 1–2 GW of data center flexibility produces NPV savings of roughly $300–400 million over 2025–2050 when incorporated into integrated resource planning rather than treated as an afterthought (LBL March 2026 update). Most utilities still plan for large loads as firm, inflexible demand — understandable, since planners are paid to keep the lights on rather than ruminate on optionality. But treating dependable flexibility as fully firm leads to overbuilding.

Camus, encoord, and Princeton’s ZERO Lab push the analysis into interconnection architecture, arguing that flexible grid connections and “bring your own capacity” structures can speed interconnection while protecting other customers from cost shifts. Their analysis suggests that traditional firm-only interconnection adds $764 million in system supply costs per GW of new data center demand, while conditional firm service plus BYOC could reduce costs by roughly $400 million per GW, and a 500 MW data center using both could reach full operation three to five years faster. The approach demonstrates how to operationalize flexibility: not a handshake promise, but planning models, telemetry, contracts, verification, and enforcement that together make flexibility legible to the grid.

The newest operational evidence comes from Emerald AI, EPRI, National Grid, and Nebius. Their March 2026 UK demonstration tested whether AI infrastructure could operate as a “power-flexible” asset without disrupting mission-critical workloads. Over five days, a 96-GPU NVIDIA Blackwell Ultra cluster at Nebius’s London AI Factory responded to 22 live dispatch events, including surprise emergency signals, reducing power by 30 percent in under 40 seconds. Emerald’s Conductor platform, prioritizing urgent workloads while absorbing flexibility through lower-priority work, hit 100 percent compliance across more than 200 power targets while preserving SLAs. Small relative to gigawatt-scale demand, but a real operational proof point.

So what should we make of all this? Data center flexibility is real, valuable, heterogeneous, and institutionally fragile. Real because demonstrations and planning studies show some loads can respond. Valuable because relatively small reductions during constrained hours can avoid expensive capacity additions or accelerate interconnection. Heterogeneous because OpenAI, AWS, a co-location provider, and an enterprise data center do not share business models or control rights. Institutionally fragile because technical capability does not become system value automatically.

The Flexibility Stack

The framework I use to make sense of all this is a flexibility stack. Power systems are layered cyber-physical-social systems of machines, software, markets, laws, regulators, firms, consumers, and communities. The stack metaphor helps us see that flexibility is not just a technical feature of servers, batteries, or cooling systems. Those technologies matter, but they operate within an institutional architecture that determines who may act, who gets paid, who bears risk, and who is accountable when things go sideways.

Source: A thorough iterative conversation with ChatGPT about my flex stack idea.

At the bottom are physical assets: generators, wires, substations, batteries, data centers, cooling systems, sensors, and controls. These define the physical possibility frontier. Above sits the digital-control layer—telemetry, forecasting, workload orchestration, AI scheduling, DERMS, and automated controls—which turns physical possibility into operational capability. A battery is only potential flexibility until integrated into a dispatch platform; a shiftable AI workload becomes usable flexibility only when software can classify, defer, migrate, and verify it.

The firm and business-model layer comes next. Ownership and contracts determine control: a pure-play AI company, a cloud hyperscaler, and a co-location provider each control different parts of the workload, facility, and service obligation. Flexibility is partly technical but also a function of governance inside the firm.

Markets and tariffs determine whether flexibility is rewarded. Energy, capacity, ancillary services, demand response, congestion pricing, and bilateral contracts all shape its value. A tariff is more than a price schedule; it is an institutional technology reflecting risk allocation, curtailment rights, cost causation, collateral, and service quality.

The regulatory layer operates across jurisdictions. State public utility regulation shapes retail tariffs, integrated resource planning, and cost allocation, raising questions like whether data centers should pay minimum demand charges, post collateral, receive conditional service, or bring accredited capacity, and whether other customers should bear the risk if forecasts are wrong. Regional RTO governance and market design shape wholesale participation, capacity accreditation, transmission planning, interconnection, and scarcity pricing, where location matters: a megawatt of flexibility in the wrong place may have little value, while a smaller reduction at a constrained node may be highly valuable. Federal regulation and environmental rules together govern interstate transmission, reliability, and whether flexibility can rely on diesel, gas, batteries, fuel cells, or clean procurement.

Consumers and communities sit at both top and bottom: they use electricity, cloud services, and AI tools, but they also pay rates and live near infrastructure. Regulatory legitimacy depends on making the bargain explicit—who benefits, who pays, who bears risk, who has recourse.

Implications

Flexibility can fail at any layer of this stack. A workload may be technically shiftable but prohibited by customer contracts. A utility may want conditional service the tariff doesn’t offer. An RTO may value flexibility but lack a market product that fits. A data center may have backup generators its air permits won’t allow it to run. A regulator may approve a flexible tariff that, without telemetry and penalties, can’t be verified.

Flexibility is not a resource until the stack makes it actionable. A flexible workload, a battery, or a data center willing to help during grid stress is only potential flexibility. To become a grid resource, flexibility must be defined legally, measured technically, compensated economically, dispatched operationally, and enforced contractually.

The architectural challenge is to move beyond a binary model of service — connected or not, firm or non-firm, load or resource. Data centers could instead buy modular service bundles combining firm and conditional capacity, self-supply obligations, demand-response commitments, clean-energy requirements, and verification-and-penalty structures, with the bundle depending on location, workload, business model, and grid constraints.

The current evolution of power systems is an institutional coordination problem under technological change. Data centers make it visible because they are large, fast-moving, geographically concentrated, digitally controllable, and politically salient. They stress the old architecture but also reveal where it can evolve. The flexibility stack is the architecture through which latent technical capability becomes reliable system value.

The practical question at the center of data center energy policy is not whether data centers can be flexible; it is whether our institutions can make their flexibility real.

Subscribe now

Electricity Building at the 1893 World’s Columbian Exposition in Chicago (Source: Chicagology)

The Stress Test

Electricity used to be easy for most people to ignore. It was the quintessential background technology: vital, expensive to build, occasionally maddening, but mostly invisible unless the power went out or the monthly bill jumped. That invisibility has ended. Between the rapid growth of data centers, the surge of interest in AI, the electrification of vehicles and buildings, and a string of uncomfortable reminders that reliability is neither free nor automatic, more people are paying attention to the electric grid and asking a sensible question: Why does this system seem so hard to change?

That question has become more urgent as utilities, regulators, large customers, and ordinary ratepayers argue over how to serve large new loads, how quickly to build generation and transmission, and who should bear the costs and risks. Data centers are not the only reason these tensions exist, but they have become a powerful catalyst because they are revealing a deeper reality: The American electricity industry still operates through institutions built for a very different technological and economic world. The old system emerged for some good (and some not so good) reasons, and for a long time it worked tolerably well. But the status quo has a habit of lingering long after the conditions that created it have changed.

To understand why today’s electricity debates are so consequential, you have to go back to the beginning, which in the United States is 1882. The electric industry began as a messy burst of entrepreneurship, experimentation, technical rivalry, and races to the patent office, only later evolving into what looked like a neat monopoly blessed by economic theory. In the late 19^th century, inventors and firms were racing to turn the strange, invisible, physical phenomenon of electric current into something commercially useful. Electric lighting was the glamour product of its age, and early systems were local, improvised, and expensive. Thomas Edison’s Pearl Street Station in lower Manhattan is the iconic example, but it was only one node in a broader network of invention and commercial trial and error. What eventually mattered most was the form in which electricity could be delivered. The “war of currents” victory of George Westinghouse and Nikola Tesla’s alternating current (AC) over Edison’s direct current (DC) was decisive because AC, combined with transformers and high-voltage transmission, made it possible to generate power at large scale and move it over long distances. That technical architecture favored centralization and created the modern grid’s basic logic: large generating plants, extensive wires networks, and one-way delivery from producer to consumer.

That architecture had powerful economic implications. Electricity required large upfront investments in generation, transmission, and distribution, while the cost of serving additional customers once the system was built was relatively low. It was an industry of high fixed costs and strong economies of scale, so for a long time one network appeared cheaper than many overlapping ones. From that cost structure came the intuitive case for monopoly; if duplicating infrastructure was wasteful, the least-bad option might be to grant one firm an exclusive territory and regulate it to prevent abuse. That logic became the foundation of the early 20^th-century public utility model: a legally protected monopoly with an obligation to serve, overseen by state commissions. The natural monopoly theory that economists later formalized gave this arrangement a conceptual vocabulary, but the institutions themselves were already emerging in the Progressive Era as both an economic solution and a political bargain.

That bargain had two economic stories behind it, and both matter. The public-interest story holds that regulation emerged to protect consumers from monopoly abuse and to ensure safe, reliable, affordable service. The public-choice story holds that firms often sought regulation because it stabilized earnings, reduced competitive pressure, and lowered the cost of capital. Samuel Insull, the founder of Commonwealth Edison and an early leader in the utility industry, captured both dynamics: He expanded electric service aggressively, but also recognized that a regulated franchise could provide the stability needed to finance massive infrastructure investments, and he lobbied for such regulation. The result was an institutional settlement among governments, firms, and reformers, forged in a world where large centralized systems seemed both technologically necessary and economically efficient.

The practical mechanism for governing this settlement was rate-of-return, or cost-of-service, regulation. The basic logic is simple. Because the firm is a monopoly, it cannot be left to charge whatever the market will bear—a concern that grew as electricity became essential to daily life and economic activity. At the same time, because the firm must make large, long-lived capital investments, it must have a reasonable opportunity to recover prudently incurred costs and earn an allowed return on the capital devoted to public service, its rate base. Regulators therefore review the utility’s costs, determine which assets belong in its rate base, and authorize rates designed to cover operating expenses plus a reasonable return.

This model did some important things well. It supported capital formation in a capital-intensive industry during a century of fairly steady demand growth. It reduced some kinds of investor risk. It provided a framework for universal service and reliability. And for much of the 20^th century, it fit reasonably well with the dominant “iron in the ground” technologies of the industry: large central stations, long-lived network assets, and a relatively standardized product.

Still, the model came with its own incentive structure, and that structure matters enormously today. Rate-of-return regulation creates a capital bias in utility decisions: It is easier to put poles, wires, substations, or power plants into the rate base than to earn on digital grid management, operational improvements, demand-side flexibility, or organizational innovation. That bias toward the tangible and depreciable was less troubling when the goal was to build a standardized network at scale. It is more problematic now, when the greatest value often comes from coordination, information, software, flexible demand, decentralized resources, and other ways of doing more with existing assets. The regulatory model was built for a static industrial system, while electricity has increasingly become a dynamic one.

That static character was built into the theory. The classic natural monopoly model assumes stable technology, reasonably predictable demand, a clear definition of the product being sold, knowledge of costs, and regulators capable of approximating the public interest. Those assumptions were always simplifications, but for decades they were close enough to reality to be useful. The 20^th-century electric system was centralized, standardized, and largely one-directional. Demand grew, but its growth was comparatively legible. Utilities built more generation, more wires, and more substations, and regulators reviewed those investments through familiar legal and economic frameworks. The institutions and the technologies co-evolved, and were more tightly coupled than many people realized.

Since the 1990s, that fit has weakened, especially as new generation technologies changed the sector’s economics. Combined-cycle gas turbines reduced efficient scale, lowered capital costs, and made wholesale competition more feasible in some places. Wind and solar introduced very different cost structures and operating characteristics from traditional thermal plants. Storage and digital control technologies improved the ability to shift, monitor, and coordinate supply and demand, while smart grid technologies and distributed energy resources reduced transaction costs and made new organizational forms more plausible. In earlier decades, bundling energy, wires, and retail service within a vertically integrated utility had a strong technological rationale. As technologies changed, so did the efficient boundary of the regulated utility: Activities that once had to occur inside a monopoly organization could now be coordinated through contracts, markets, platforms, or decentralized control.

This development is where the pacing problem enters. Technologies can change quickly; institutions usually do not. In electricity, that asymmetry is especially pronounced because the system is physically consequential, legally dense, and politically salient. Utilities and regulators are understandably risk averse, since reliability failures are visible and customers are unforgiving. But caution is costly too. When institutions remain calibrated to older technologies and policy goals, they can become ill-equipped to meet new conditions, slowing adoption, preserving legacy organizational forms beyond their useful life, and raising the cost of experimentation. Governance can thus lag behind technological possibility, and that lag can become a source of inefficiency and conflict in its own right. Even governance rules meant to protect reliability can harden into barriers that slow entry, adaptation, and coordination under new conditions.

Regulatory restructuring in the 1990s and 2000s only partly addressed this mismatch. In some states, generation became competitive while transmission and distribution remained regulated monopolies; in some places, retail choice emerged, while in others it did not. Regional transmission organizations and independent system operators assumed important wholesale coordination functions, but they did not erase the deeper institutional inheritance of the public utility era. What emerged was not a clean replacement of the old model, but a layered hybrid: Parts of the system were liberalized while monopoly governance persisted in others, and unevenly across states and regions. That patchwork can work, but it means today’s disputes over regulation and the role of the utility unfold within a complicated structure of historical compromises, nested rules, and reliance on past practices. The old model was not overthrown. It was refurbished, patched, and in some places tastefully wallpapered over. The wiring underneath is still old.

This historical background makes the current data-center debate revealing. Data centers are not merely large loads. They are often lumpy, geographically concentrated, highly time-sensitive, and unusually demanding in their expectations of reliability. They can require substantial new investment in generation, transmission, distribution, or all three. They also sharpen the old regulatory questions that rate-of-return systems have always struggled to answer neatly. Who should pay for infrastructure built partly to serve a small number of very large customers? Should incumbent utilities build ahead of demand, or should they wait for firmer commitments? How should risks be allocated if projected load growth does not materialize, or materializes in a different place or form? Should utilities own and rate-base the assets needed to serve these customers, or should more of the coordination occur through contracts and competitive supply arrangements? Beyond merely engineering questions, these are questions about governance, incentives, and risk allocation.

Rather than creating the electricity sector’s institutional tensions, data centers exposed them, and at scale. AI’s arrival collides with a system still shaped by assumptions from an earlier era: that demand growth is gradual and predictable, that the product is relatively homogeneous, that large infrastructure should be socialized through monopoly regulation, and that adapting to change can proceed at the stately pace of the regulatory docket. But the world arriving at the grid’s doorstep is more heterogeneous, more digitally coordinated, and more impatient than the one those assumptions reflected. Meanwhile, the grid itself remains a shared network in which one party’s decisions affect many others, so governance and exclusion rules still matter. Reliability is not something markets can simply wish into existence. The real question is whether our existing institutions preserve reliability while also enabling adaptation.

That is the deeper lesson. The current electricity debate is not simply about whether data centers are good or bad, or whether they are raising rates for other customers, or even whether AI is worth the power it consumes. It is also a larger debate about whether institutions designed for 20^th-century centralized power systems can govern a 21^st-century electricity economy with a vastly different architecture and technology set. The natural monopoly model solved real problems, and it should not be caricatured as institutional sclerosis or rent-seeking (although both are present). It helped build one of the most important infrastructure systems in human history. But it was technology-contingent in ways that data center innovation is making apparent, reflecting a particular set of technologies, costs, and policy priorities. As those conditions change, the case for preserving every inherited boundary, every bundled function, and every regulatory habit grows weaker.

What matters now is matching governance more honestly to present conditions and designing institutions that can adapt better as those conditions change. In some parts of the system, monopoly still makes sense. In others, competition, contracting, and decentralized coordination can do valuable work. Regulation should focus less narrowly on approving capital and more broadly on enabling performance, adaptation, and intelligent risk allocation.

That shift will not be easy, and no clever slogan will spare us the hard institutional work. But the alternative is worse: trying to govern a changing electricity system with categories designed for a world that no longer exists.

And that, in the end, is why data centers matter so much. They are not just another source of load growth. They are a stress test. They reveal where the grid is strong, where it is brittle, and where its regulatory operating system has begun to show its age.

Originally published in the Dispatch Energy newsletter: Is Our 20th-Century Grid Up to the Current Task?

Source: ChatGPT upon reading this article

Economists spend their lives studying how people respond when circumstances change. It would be odd if we failed to apply the same logic to ourselves.

Artificial intelligence is not just another research tool. It is a technology shock to the academic knowledge economy. It lowers the cost of some inputs into research, raises the relative value of others, and begins to change what the profession rewards. That shift matters not only for how economists work, but also for which skills and forms of judgment are likely to become more valuable.

Subscribe now

The old equilibrium

For the past several decades, economics has been shaped by a particular methodological equilibrium. The profession rewarded scholars who could assemble large datasets, digitize new sources of information, and use increasingly sophisticated econometric tools to extract credible causal estimates. This development improved economics in important ways by imposing discipline, raising standards of evidence, and moving the field away from loose speculation untethered from the world.

But tradeoffs are everywhere, including in research methods.

While also raising standards, the causal-inference era changed the kind of questions economists asked. Because credible identification strategies usually require narrow treatments, clean variation, tractable settings, and measurable outcomes, the profession gradually tilted toward questions that fit the available empirical machinery. The result has been a great deal of technically impressive work, but also, too often, a narrowing of intellectual ambition. Economics became more a field of data analysis and less a field of deeper inquiry into institutions, governance, organizational form, and the structure of human choice.

AI may now disrupt that equilibrium.

A growing number of economists have begun thinking seriously about what AI means for research. Anton Korinek has traced the progression from AI assistance on research micro-tasks to semi-autonomous agents that can work across multi-step workflows (JEL 2023, JEL 2025). Matthew Kahn emphasizes the fall in the fixed costs of pursuing ideas, Kevin Munger the rising importance of evaluation when production becomes cheap, and others have begun to map the broader consequences (for example, Alex Kustov, Scott Cunningham, and Chris Blattman’s Claude Blattman). I agree with much of that discussion. But the best way to understand what is happening is through price theory.

A shock to relative prices

AI is a shock to the relative prices, values, and costs inside the academic knowledge economy. It cheapens some inputs, raises the relative value of others, and therefore changes what the profession rewards.

The clearest place to see the shift is in the falling cost of routine cognitive labor. Literature search, synthesis, coding, formatting, drafting, referee-response generation, specification exploration, and some forms of data cleaning and visualization are all becoming cheaper. Not free, and not perfect, but much cheaper than they were even two years ago. Tasks that once required hours of faculty time, graduate-student time, or RA time can now be completed in minutes, often at surprisingly high quality, though still with the need for human review.

The issue is not whether AI can produce useful work; it plainly can. The issue is that once many more people can produce plausible literature reviews, decent code, competent drafts, and standard empirical analyses, those things no longer define scholarly distinction in the way they once did. The floor rises, but the skill premium shifts.

And that shift raises the value of something else: judgment. In a world where production becomes abundant, discernment becomes relatively scarce and thus relatively more valuable. What matters more is the ability to decide what is worth studying, which mechanism is first-order, which institution matters, what assumptions are plausible, which results travel across contexts, and what pattern in the evidence actually matters.

Theory is deeper than formalism

This point is easy to miss because economics has often conflated “theory” with formal mathematical modeling. But theory in the deeper sense is economic reasoning: the disciplined application of principles and logic to understand behavior, tradeoffs, institutions, and constraints. It is the ability to ask the right price-theoretic question: what changed, what became cheaper, what became more valuable, what bottleneck moved, what margin now matters?

On that understanding, AI should not push economics away from theory. It should raise the return to deeper economic reasoning.

This is where Tyler Cowen’s recent work published last week is both provocative and, I think, wrong in an instructive way. In Chapter 4 of The Marginal Revolution, Cowen argues that machine learning and AI are weakening the traditional ties between economics and intuitive marginalist reasoning. In one sense, he is obviously right. More and more empirical work relies on high-dimensional prediction, algorithmic pattern detection, and black-box methods that do not map neatly onto the sort of verbal microeconomic intuition economists are used to teaching and defending. The old chalkboard-style link between intuitive theory and empirical implementation is under strain.

But it does not follow that marginal analysis is fading away. If anything, AI makes marginal analysis more important. AI does not change whether tradeoffs, substitutions, complementarities, constraints, and shadow prices govern the world; it changes where the important margins are. If AI lowers the cost of coding, synthesis, search, and routine empirical work, then the economist’s task becomes even more about understanding which scarcities still govern the system. What is the marginal value of human interpretation when machine-generated analysis becomes abundant? What happens to peer review when paper production becomes cheap? What becomes relatively more valuable when standard empirical execution is commoditized? What kinds of local knowledge remain hard to formalize and therefore rise in importance?

These are price-theoretic questions.

A Hayekian twist

A Hayekian perspective sharpens the point. If AI cheapens formalized information processing, then tacit knowledge, local knowledge, judgment, and institutional understanding may rise in relative value. That is not a departure from economics. It is economics. Hayek taught that knowledge is dispersed, contextual, and often difficult to aggregate. The more abundant machine-generated outputs become, the more important it becomes to know which facts matter, which contexts are special, which institutions shape behavior, and which apparent regularities are misleading.

This possibility helps explain why AI may open up space for a revival of parts of economics that have not been central to the profession’s self-understanding for some time.

Why TCE and NIE may matter more

The field I have in mind here is transaction cost economics and new institutional economics. Some of the deepest insights in this tradition come from Oliver Williamson, in works like Markets and Hierarchies (1975) and The Economic Institutions of Capitalism (1985), as well as many journal articles. Williamson’s core contribution was to provide a disciplined way of reasoning about governance, contractual incompleteness, asset specificity, adaptation, authority, and the comparative institutional logic of markets, hierarchies, and hybrid arrangements.

That framework has had important empirical influence, especially in the work of Scott Masten and co-authors and Steven Tadelis and co-authors on incomplete contracting, and Robert Gibbons and co-authors on relational contracting. But here there is an important asymmetry. The empirical uptake of Williamson has been strongest in areas where the relevant phenomena are amenable to formal (often game) theory and leave observable traces in data. Contracting and procurement are natural examples. In those domains, governance choices and contractual structures are visible enough to study using the kinds of empirical tools that modern economics prefers.

But that empirical success should not obscure the larger point. The profession absorbed the slice of TCE/NIE that fit the reigning evidentiary regime. Much of the broader agenda remained underdeveloped, not because it lacked rigor, but because its testable implications were harder to force into the form of large datasets and narrow identification strategies.

That matters because some economically important phenomena require mixed evidentiary strategies precisely because the mechanisms themselves are institutionally embedded. Adaptation under uncertainty, authority within organizations, tacit coordination, internal dispute resolution, sequencing, relational governance, and the comparative fit between institutions and tasks are often economically central. But they do not arrive naturally as variables in a clean spreadsheet.

For a long time, that fact made such questions methodologically costly. They were harder to study in a profession that increasingly rewarded what could be digitized, scaled, and cleanly identified.

AI changes that cost structure. It lowers the cost not only of routine empirical analysis, but also of gathering, synthesizing, and interpreting richer kinds of evidence: textual records, contractual materials, case comparisons, archival documents, organizational descriptions, meeting transcripts, internal language, and other qualitative or semi-structured materials. More importantly, it raises the value of economists who can reason carefully about institutions even when the evidence is messy, incomplete, or not easily reducible to a panel dataset.

This is where the most interesting opportunity arises. The future may belong less to the economist who can execute a standard narrow causal design more efficiently, and more to the economist who can identify the right institutional margin, integrate multiple kinds of evidence, and make disciplined comparative judgments about governance and adaptation. Not a rejection of rigor, that is a broader understanding of what rigor in economics can look like.

Beyond workflow

Mullainathan and Rambachan’s recent paper “Science in the Age of Algorithms” (NBER 2025) pushes the argument beyond workflow. Their claim is not only that algorithms save time, but that they can formalize parts of science that have long been treated as tacit and off-stage: anomaly generation, theory generation, and judgments about when one theory matters more than another in a given setting. Their argument is especially relevant for economics because ours is not a “reigning champion” science in which one theory cleanly defeats another and takes its place. It is much more a patchwork science, in which researchers carry around portfolios of models and choose among them based on context, intuition, and judgment.

Researchers already decide which variables matter, which frictions are first-order, which simplifying assumptions are harmless, and which interpretation of the setting is economically meaningful. If AI makes narrow empirical execution cheaper, then those judgments move closer to center stage.

What remains scarce

That is why I think the coming change in economics research may be deeper than many current discussions suggest. The most immediate effect of AI is to accelerate existing workflows. The more important effect may be to reprice the components of intellectual distinction. Once routine production becomes cheaper, the profession has to ask what remains scarce.

One answer is evaluation. If papers become much easier to produce, then the cost of filtering them rises in relative terms. Journals, editors, reviewers, and trusted intellectual communities become more important. As Kevin Munger has argued, when the supply of manuscripts rises dramatically, the key problem becomes metascience: how to allocate scarce attention and preserve standards when production is cheap. Put differently, AI lowers the cost of speaking, so the cost of listening rises.

Another answer is training. The apprentice model of economics has long depended on graduate students and young scholars learning by doing the smaller tasks that add up to research competence. If AI absorbs many of those tasks, the profession may gain current productivity while weakening one of the channels through which future judgment is formed. The profession will have to rethink how it cultivates judgment when many traditional apprenticeship tasks have been automated.

But the most important answer, I think, is deeper economic reasoning. For the past generation, economics often rewarded the subset of theories that could be converted into clean tests using existing data infrastructures. That equilibrium made sense given the relative scarcities of the time. Credible identification, data construction, and technical execution were scarce, and the profession rationally rewarded them. But a change in technology changes the logic of reward. Once those activities become less scarce, a different set of capabilities rises in relative value.

The economist of the near future will still need empirical skill. But she will increasingly distinguish herself not by her ability to carry out narrow empirical routines, but by her ability to frame deep institutional questions, decide which mechanisms are first-order, and interpret a rapidly changing world with theories that travel across contexts. That is economics recovering some of its comparative advantage.

Economics should notice first

For too long, many parts of the discipline have acted as though rigor and interestingness were substitutes. The result has often been a choice between intellectually ambitious but difficult-to-test questions and tightly identified but conceptually narrow ones. AI does not eliminate that tension, but it may soften it. By lowering the cost of standard research production and expanding the feasible set of evidence-gathering and synthesis, it creates more room for economists to ask bigger questions without giving up discipline.

This possibility should make us more pluralistic, in the sense of recognizing that different mechanisms require different forms of evidence, and that institutional phenomena are often too rich to be captured by one evidentiary strategy alone. Economics at its best has always been about disciplined reasoning under constraint. AI changes the constraints. It should also change the discipline’s self-understanding.

If this shift continues, the economists who stand out will not be the ones who most efficiently perform the routines that have defined prestige for the past three decades. They will be the ones who can see where the bottlenecks have moved, where institutional change is occurring, what forms of knowledge remain tacit and local, and how to reason clearly about governance, adaptation, and human choice in a world whose old constraints are coming undone.

That would not mean economics is abandoning its core logic. It would mean economics is rediscovering it.

Subscribe now

A recurring mistake in energy debates is assuming that decarbonization is only a technology story. Wind and solar got cheaper. Natural gas got abundant. Coal declined. Emissions fell. That story is not wrong, but it is incomplete. Generation technologies do not deploy themselves, and they do not operate in an institutional vacuum. They are built, financed, dispatched, and retired within rules, and those rules influence incentives, shaping the path of technological change along the way.

Has market participation fostered lower-carbon generation? That is the question my new paper with co-authors Joshua Rhodes and Samantha Zyontz, Do Power Markets Promote Decarbonization? Evidence from State-Level Data in the United States, takes up (ungated manuscript version). We ask whether participation in wholesale power markets is associated with a lower-carbon generation mix over time, especially in states that restructured and unbundled generation from traditional utility regulation. Using annual state-level generation data from 1990 through 2020, we compare restructured states to several different control groups across five distinct phases of wholesale market development, letting the institutional history guide the empirical strategy. Because the institutional history here is messy, staggered, and heterogeneous, we do not make strong causal claims. Instead, we use a block-treatment-window difference-in-differences strategy to study a comparative institutional question that has been surprisingly undertheorized in the literature.

Three institutional categories drive this comparison. Nonmarket states retained the traditional model of vertically integrated, regulated utilities — a single company owning generation, transmission, and distribution, with a regulator setting rates and approving investments. Restructured states went further, requiring utilities to divest their generation assets to independent power producers and opening both wholesale and, to varying degrees, retail markets to competition. Hybrid states occupy the analytically interesting middle ground: they joined organized wholesale markets through MISO or SPP, gaining access to regional dispatch and market pricing, but retained vertically integrated utilities and did not require generation divestiture. This category turns out to be valuable precisely because it allows us to separate the effect of market participation from the effect of full restructuring, treating each as a distinct institutional feature rather than collapsing them into a single undifferentiated treatment.

The five phases reflect the uneven chronology through which these arrangements developed. The first two hypotheses compare early restructuring states — those beginning wholesale market operations between 1997 and 1999 — against nonmarket states, first through 2003 and then extending to 2012 to capture the effects of the shale gas revolution and falling wind production costs. The third and fourth hypotheses run the same comparisons, swapping the control group from nonmarket states to the original MISO states, asking whether restructured states look different from states that joined an organized market without unbundling generation. The fifth hypothesis extends the analysis through 2020, using the full set of MISO and SPP hybrid states as the control group, asking whether full restructuring produced a different generation mix than market participation alone. Running these five analyses in parallel, rather than forcing them into a single specification, allows the institutional history to do what it should: discipline the empirical strategy rather than be flattened by it.

The headline finding is straightforward, although its implications are not. In all five cases, market participation through restructuring is associated with more lower-carbon generation than in the control states. But the strongest and most consistent effect is the substitution of natural gas for coal and oil, not a sudden flowering of wind and solar alone, especially in the earlier years and through roughly 2012. In later years, restructured states and control states begin to look more similar, suggesting that the institutional advantage of restructuring has been narrowing as markets and policies have converged.

That pattern tells a more historically disciplined story about how electricity decarbonization has actually happened in the United States. Markets have helped move generation toward a cleaner mix, largely through how dispatch rules and investment environments enabled the gas-for-coal transition, while also producing weaker but still meaningful evidence in some cases for non-gas low-carbon generation.

That distinction matters because too much contemporary discussion treats decarbonization as if the only intellectually respectable endpoint is eliminating every last ton immediately. But power systems are complex, capital-intensive coordination systems operating under hard reliability constraints and political limits on cost. How do we build a power system that gets cleaner over time without making it less reliable or unnecessarily more expensive?

This system design framing is the deeper argument running through the paper. Decarbonization is an institutional and system design problem. Electricity markets determine how generators compete, how prices convey information, how dispatch works, how new technologies enter, and how investment incentives are structured. Those institutional features are part of the mechanism through which the generation mix changes.For that reason our historically grounded comparative institutional analysis that captures the messy reality of regulatory restructuring contributes to this ongoing literature and discussion.

The Paper’s Contributions

One original contribution of the paper is that it moves the literature away from a too-simple binary. Many discussions implicitly compare “markets” to “regulation,” as if the United States had two distinct categories and a referee with a whistle. It did not. Our analysis distinguishes among the three institutional arrangements described above: nonmarket states, restructured states, and hybrid states. Beyond being useful taxonomic housekeeping, that distinction captures a central feature of the U.S. electricity sector: organized market participation is not the same thing everywhere, and institutional architecture shapes incentives and shows up in variation in the generation technology mix.

A second contribution is that the paper integrates three strands of literature that are too often discussed separately: the literature on renewables and emissions, the literature on natural gas and emissions, and the literature on restructuring and economic efficiency. If institutions affect investment incentives and dispatch incentives, then institutions can also affect emissions through the technologies that get built and the plants that run. Yet much of the decarbonization literature still behaves as if fuel prices and technology costs do nearly all of the explanatory work, with market institutions entering stage left only as a control variable. Our paper argues that this approach misses something important. The path of decarbonization is an institutional story that includes markets as well as a technology story and a fuel-price story.

A third contribution is methodological, although here the contribution lies as much in restraint as in technique. The historical evolution of wholesale power markets in the United States has been messy, and does not lend itself neatly to a single clean treatment date and a simple causal identification strategy. Different regions evolved at different times, with different combinations of market formation, regulatory restructuring, generation divestiture, and later technological change. Rather than pretending otherwise, we structured the empirical analysis around five separate hypotheses and let the institutional history guide the econometrics. There is a certain charm to econometric elegance, but false elegance is expensive. In this context, institutional realism is more useful than an elegant model built on a simplified premise.

For Economist Readers

Why does this paper matter for economists? Because it makes a claim that economists should find both familiar and oddly neglected. Institutions matter. More specifically, market institutions affect dynamic efficiency by shaping innovation, entry, exit, and the relative profitability of competing technologies. That proposition would not have startled Hayek, Buchanan, Coase, North, or Ostrom. But in electricity economics, it often gets crowded out by narrower questions about dispatch, production cost savings, or the emissions consequences of a given renewable increment at the margin. Those are important questions, but they aren’t the only questions. Our paper asks whether market participation itself belongs inside the explanation of decarbonization. The answer appears to be yes.

The paper also matters for economists because it clarifies the mechanism. The strongest historical association is not that markets directly produced a zero-carbon grid. The strongest association is that market participation facilitated cleaner substitution, especially gas for coal, during a crucial phase of transition, doing so in ways that reshaped both investment and dispatch decisions simultaneously. That claim is more modest and nuanced than some market enthusiasts may want, but it is also more credible. Market participation appears to have mattered most where market institutions altered the competitive and operational environment in ways that did not erect barriers to less carbon-intensive generation. In other words, the institutional margin mattered, but it mattered through historically specific mechanisms.

For Readers Interested in Policy

Why does this research matter for policy audiences? Because policy makers tend to hear climate debates in one of two registers, both unsatisfying. In one register, decarbonization is a moral imperative detached from costs, reliability, and institutional feasibility. In the other, any acknowledgment of those constraints is treated as hostility to cleaner energy. Both frames are analytically lacking, for different reasons. The paper points toward a better one.

The practical lesson is not that market design substitutes for policy. Nor is it that any market, even one with bad rules, will automatically produce good environmental outcomes. The lesson is that market processes can and do align economic and environmental incentives, and that market design is itself part of sound policy. The rules governing dispatch, entry, ownership, and competition shape which technologies become investable, financeable, and operationally viable, determining in turn which investment decisions actually get made. That reality means the path to a cleaner grid is to design market institutions that do not pose barriers and transaction costs to building, integrating, and dispatching cleaner generation in practice. That implication goes beyond the scope of this paper and speaks to the challenges of RTO governance.

This practical lesson also helps translate a very economist point into normal English. The policy objective is not decarbonization at any cost, regardless of tradeoffs. The objective is processes that enable a power system that becomes as clean as possible while remaining reliable, affordable, and capable of meeting growing demand. A policy that reduces emissions but destabilizes reliability or imposes needless costs is not more serious because it is bold; it is simply worse policy.

This perspective is especially important now, because the grid is under strain from multiple directions at once: load growth from data centers, continuing renewable integration, transmission bottlenecks, interconnection delays, and persistent institutional friction in adapting rules designed for an older resource mix. In that environment, system design matters even more. The central policy question is how to build institutional arrangements that provide a framework for investment and operation that enables lower-carbon technologies while preserving the system properties that electricity consumers understandably value but take for granted.

Our paper does not settle every question in that debate. It does not claim that markets alone explain decarbonization, and it does not deny the impact of renewable portfolio standards, tax credits, net metering, or political choices. In fact, the econometric analysis controls explicitly for several of those factors. What the paper does show is that comparative institutional analysis and institutional design belong much closer to the center of the conversation than they usually sit. If we want to understand why the U.S. power sector got cleaner, and how to help it get cleaner still while remaining affordable and reliable, we need to pay more attention to the rules of the game.

That, in the end, is the broader significance of the paper. Decarbonization is not only about what technologies exist. It is about what institutions make possible. The grid is a machine, a governance system, a market platform, and a technological order all at once. It is a layered system of systems, technological and institutional.

Imagine a household in 2026 with an EV in the garage, rooftop solar on the roof, and a smart thermostat that has preferences encoded by its owner. The household also has something new: an agent.

On a Tuesday night, the agent buys electricity, schedules EV charging, nudges the heat pump, and arbitrages time-varying prices against the family’s comfort constraints. It does all of this quickly, quietly, and without asking permission every five minutes. The family wakes up to a charged car and a comfortable house, and the bill is lower than it would have been otherwise. The whole system feels like a small miracle.

Then the uncomfortable thought arrives, usually right after the miracle: “Wait. What exactly did it do, and why?”

Subscribe now

That moment is the economic problem of the agentic economy in miniature. The promise transcends clever AI. The promise is delegation: AI systems acting for us in markets at the tempo of modern systems. Delegation is where the economics becomes interesting, and potentially treacherous.

Our newly-published paper, Markets, agency, and trust: AI agents and the knowledge problem (co-authored with philosopher Brennan McDavid and my long-time collaborator, engineer David Chassin), makes a simple claim with broad implications: agentic AI does not solve the knowledge problem; it relocates it into principal–agent relationships, and it relocates it into the institutional framework that makes delegation trustworthy.

Hayek (1945) identified this knowledge problem: the impossibility of any single mind — planner, regulator, firm, or algorithm — possessing, articulating, and updating in real time all the dispersed, tacit, context-specific information that complex economic coordination requires. Cognition is bounded, much relevant knowledge is local, and much of it emerges only through interaction and exchange.

The knowledge problem is not an engineering defect that better models will eliminate. This constraint is a fundamental feature of human cognition and social coordination: cognitive constraints, rational inattention, limits of articulation, and action under uncertainty. AI can help us cope with these limits, and, by doing so, it can expand the feasible set of coordination. It cannot remove the underlying epistemic condition that creates the need for institutions in the first place.

Our paper develops three contributions working together. First, we define the principal-plus-AI-agent “super agent” as the relevant unit of market participation — the human-informed integrated system that bids, learns, and adapts. Second, we reframe delegation as principal-agent economics plus epistemics: preference articulation and learning become the binding bottlenecks, not transaction execution. Third, and most distinctively, we treat trust as a functional constraint and system requirement — something to specify, design for, and evaluate, not a sentiment to hope for. Transactive energy serves as our laboratory, providing a real, constrained setting where all three dynamics play out simultaneously.

Markets as epistemic systems, now mediated by super agents

Markets are more than allocation mechanisms — they are knowledge ecosystems. Prices and bids compress dispersed, often tacit information; competition serves as a discovery procedure; profit and loss provide error correction. None of this requires omniscience. This structure requires institutions that let learning happen.

Agentic AI changes who participates in that learning. The “person on the spot” increasingly delegates. A household does not manually optimize its EV charging against feeder constraints, time-varying prices, departure times, and weather. It sets priorities and constraints, and an agent acts.

This delegation creates what we call a “super agent”: the integrated principal-plus-agent system that bids, responds, and adapts. Super agents can improve coordination because they operate at machine scale and speed. But they also introduce epistemic opacity: principals may understand less about why actions were taken even as outcomes improve. That opacity matters because it changes monitoring, override behavior, adoption, and, ultimately, market performance.

Agentic AI thus does not eliminate the knowledge problem, but redistributes it. Some knowledge moves from human judgment into preference inputs and constraints. Some becomes inference inside the model. Some remains contextual, surfacing only when the principal encounters an outcome and says, “That’s not what I meant”.

Recent experimental work by Imas and Krishnan found that AI agents, left to their own devices, failed to develop price mechanisms organically. They required institutional scaffolding to coordinate at scale, offering a vivid demonstration that institutional context matters as much to machine agents as to human ones. In their simulated agentic markets, universal agent adoption actually collapsed welfare by 88% through congestion, until a price mechanism was introduced, at which point most of the gains were recovered. Hayek, vindicated in silico.

The economics of AI agents is principal–agent economics, plus epistemics

If you want an antidote to breathless “AI agent” talk, use economics to name the relationship and ask what it costs and what benefits it generates. In an agentic market the relevant relationship is principal and agent, often in a market transaction.

Principals delegate because delegation economizes on attention and expertise. Delegation also shifts risk: principals still bear consequences — financial, legal, reputational — when agents act badly or oddly. This structure reflects classic Jensen–Meckling (1976) logic: agency costs arise from imperfect alignment and costly monitoring.

Our paper adds an important twist: here the agency problem is also epistemic. Humans do not just have preferences; humans discover preferences. People often cannot fully specify objectives ex ante; they learn by observing outcomes, and, while learning, they calibrate trust. Delegation becomes a feedback loop: preferences, outcomes, revisions, and renewed delegation.

This point also explains why the “Coasean singularity” idea is relevant. Shahidi et al. (2025) argue that agentic AI could dramatically lower transaction costs—search, bargaining, verification, execution—and therefore shift activity from hierarchy toward markets, perhaps quickly (see also a Marginal Revolution post on the paper and a Twitter/X thread summarizing it). That’s correct. Our paper finishes the Coasean logic: when agents lower some transaction costs, other costs become binding — specification, monitoring, governance, liability. Those costs show up as agency costs and trust constraints. In epistemic terms, bounded cognition and limited articulation remain. The bottleneck shifts from “can we transact?” to “can we delegate, intelligibly and safely, at scale?”

Trust is not a vibe; it is a system requirement

Here’s the paradox. The best agent is the one you do not have to watch; constant oversight defeats the point of delegation. But not watching creates vulnerability. Principals must accept autonomy under opacity, and that acceptance requires trust.

Trust, in this context, is the condition under which a principal accepts autonomy under opacity, allowing the system to act without auditing every decision in real time. Trust is a design condition for participation: “I will allow this system to act on my behalf even though I cannot audit every decision in real time.” If that condition fails, monitoring and override rise, transaction costs rise with them, and the market mechanism can fail to deliver its promised gains.

One of the paper’s central institutional claims is that trust should be treated like an engineering requirement. Safety is specified, designed for, and tested; trust should be too. Explainability slogans are not enough. Warranted trust depends on interfaces, fallback modes, auditability, and governance, not just model structure.

This view also connects to “agentic commons” concerns of Imas and Krishnan. With many agents, congestion and strategic noise can degrade environments. But a second scarce resource can bind: legitimacy. Trust shocks spill over, and one failure can reduce willingness to delegate across the ecosystem. Imas and Krishnan’s simulations suggest the concern is not theoretical: trust shocks and congestion can cascade through agentic markets faster than any individual firm can anticipate, making ecosystem-level institutional design a precondition for realizing individual-level gains.

Transactive energy is a laboratory for the agentic economy

Electricity keeps this discussion honest because it is a real coordination problem with hard constraints. Distribution systems must balance reliability, congestion, heterogeneity, and legacy rules, and small mistakes can be expensive. Transactive energy implements agentic AI that reflects the theory laid out in our paper.

Transactive energy coordinates electricity supply, demand, and distributed resources using price signals and distributed decision-making. Instead of relying only on centralized control, devices and participants express constraints and preferences through bids, and a market or market-like mechanism clears allocations and payments. Market participation enables coping with complexity and local constraints in a scalable way.

Our TESS project — the Transactive Energy Services System, described in detail in the paper — was built to make transactive energy testable. TESS is a platform for simulating and prototyping high-frequency transactive coordination among heterogeneous devices in distribution settings, including the institutional details — information flows, bidding rules, settlement, participation incentives, and the ‘brownfield’ reality of existing tariffs — that ultimately determine whether the system works. Device bidding is the most important feature of our TESS design: thermostats, EVs, batteries, and water heaters are programmed with bid/offer vectors that reflect the owner’s/principal’s preferences, and devices participate in a local energy market and autonomously adjust their settings based on emergent price discovery. The market-clearing price becomes the engineering device control signal.

Source: McDavid, Kiesling, and Chassin (2026)

This domain is a natural testbed for the agentic economy because it exhibits all three themes at once. Transactive systems are epistemic (prices coordinate dispersed information), agency-laden (humans delegate through interfaces), and trust-constrained (participation depends on tolerance for autonomy under opacity). Our paper’s added value is to foreground that last point: market design is necessary, but it is not sufficient; delegation and trust architecture are part of the institution and must also be part of the engineering design process.

Three takeaways

Three takeaways follow. Agentic AI shifts the locus of the knowledge problem: the key question is not whether decisions improve, but how delegation changes which knowledge gets used, how it is encoded, and how it feeds back into discovery. The central economic object is therefore the delegation interface; preference articulation, constraint specification, override rights, monitoring, liability, and governance all determine whether agents become genuine assets or brittle black boxes. And trust is a design parameter affecting equilibrium participation and performance: ignoring it produces elegant mechanisms that disappoint; designing for it can unlock genuine gains.

What this means for regulators and utilities

Regulators should treat agentic participation as an institutional issue, not only a consumer-protection or reliability issue. If policy makers want flexible load, EV charging coordination, and DER value, then they will have to confront delegation: rights retained by principals, auditability, liability, disclosure, and governance.

Utilities should recognize that integration is not just technical. The distribution system is becoming a coordination platform, and platform performance will depend on whether households and third parties will delegate to automated systems that interact with grid constraints and prices. Trust-supporting program design, such as clear rules, verifiable settlement, sensible overrides, and graceful degradation when things go wrong, will matter as much as any particular algorithm.

Coase taught us that the costs of using the price system — finding prices, negotiating terms, enforcing contracts — are never zero, and that institutions exist precisely to reduce them. Hayek taught us what the price system accomplishes epistemically, aggregating knowledge that no planner can possess. The agentic economy forces those lessons into the same frame. As machines transact for us, institutions matter more, not less, and trust becomes part of the infrastructure of coordination.

Photo by Sam LaRussa on Unsplash

During last week’s State of the Union address, President Donald Trump declared that tech companies will “pay for their own power.” The line landed effectively. But, as is often the case in electricity markets, reality is far messier than slogans. This week, I unpack what it would actually mean for an AI data center to “bring its own power”—from the transaction cost economics of vertical integration to the layered jurisdictional puzzle of federal and state utility regulation to the counterintuitive ways that large new customers can actually result in lower average rates.

Subscribe now

‘BYO’ Energy

Trump’s speech last week raised one of the most politically charged—and economically misunderstood—questions in energy policy today: Who actually pays for the electricity powering the AI boom? His salient message: Technology companies building massive data centers should foot the bill for their electricity use. The White House followed up on the vow on Wednesday, announcing that seven tech companies had signed a “ratepayer protection pledge” to shield American consumers from rising utility prices.

Across much of the country, monthly bills are higher than they were a few years ago. Inflation explains part of the increase, so in terms of real purchasing power, the rise has been less drastic than headlines imply. Another major driver is long-deferred investment in transmission and distribution infrastructure as utilities modernize aging grids and harden them against wildfire and storm risk. These cost pressures are structural and familiar to regulators; artificial intelligence did not sneak them into the rate base last year. But the politics of a monthly utility bill rarely turn on a careful analysis of nominal versus real increases or capital expenditure cycles. The number at the bottom of the page is what voters see—and feel.

Data centers, meanwhile, are visible monuments to digital abundance—vast warehouses of servers owned by some of the largest firms in the world, powering photo storage, text messaging, video streaming, and now the growing computational demands of artificial intelligence. AI promises productivity gains and geopolitical advantage. It also consumes electricity at a scale that would have startled utility planners a decade ago. If bills are rising and tech firms are booming, the political juxtaposition is hard to ignore.

That intuition is understandable but incomplete.

Electricity is not produced and consumed in isolation. It flows through a capital-intensive network with lumpy, long-lived costs allocated through layered federal and state institutions. Retail rates are set largely by state public utility commissions under a “just and reasonable” regulatory standard. Wholesale markets operate under federal oversight. Infrastructure is financed up front and recovered over decades. In such a system, the question “Who pays?” has no clear-cut answer, however emotionally satisfying the slogan might be.

The president’s ratepayer protection pledge crystallizes a deeper economic question. What does it mean, in a regulated network industry, for a large new customer to “pay for its own electricity”? Does that principle imply vertical integration—that a technology company should build and own its own power plants? Does it mean the use of long-term contracts that cover incremental generation and transmission costs? Are those contracts with the regulated utility, with a non-utility independent power producer, or some hybrid arrangement? Or, because data centers are new customers on this large shared network that has been built over decades, does this principle require them to pay for infrastructure costs incurred long before a single AI server came online?

These questions sit at the intersection of transaction cost economics, regulatory law, and the political economy of utility cost allocation. They determine whether new load raises average rates, lowers them by spreading fixed costs more broadly, or does something in between. They also reveal the limits of federal rhetoric in a system where most of the policy levers that shape residential bills reside at the state level. What, then, does it mean for AI companies to supply their own power?

At first glance, the phrase suggests vertical integration. If a data center requires 500 megawatts to power it, perhaps the technology firm should build and operate a 500-megawatt power plant. The symmetry is appealing: a self-contained industrial ecosystem that neither leans on nor burdens the surrounding grid—what engineers would call an islanded system—like a medieval manor, but with better cooling systems and a worse relationship with local zoning boards.

But transaction cost economics implies that such a fully symmetric approach to organizing production is neither empirically common nor necessarily efficient. In his seminal 1937 article on the theory, economist Ronald Coase asked a deceptively simple question: Why do firms exist at all? Coase framed the firm as an alternative governance structure to the price system, arguing that firms arise when coordinating activity through markets entails nontrivial transaction costs—search and information costs, bargaining and drafting costs, and the costs of monitoring, adaptation, and enforcement under incomplete contracts. On this account, vertical integration is optimal when market contracting is comparatively costly.

In electricity, these governance trade-offs are even more acute. Operating a power plant requires several specialized, capital-intensive capabilities, giving utilities and independent power producers a comparative advantage over technology companies, which specialize in computing and software. Even very large firms rarely wake up one morning and decide that running a power plant is an obvious side hustle.

In many industries characterized by large, durable demand and highly specialized assets, the efficient solution is not vertical integration but long-term contracting. Power purchase agreements, capacity contracts, and bespoke infrastructure arrangements already exist precisely to align incentives without erasing organizational boundaries. The make-or-buy literature in economics suggests that the relevant question is not whether a data center should own a power plant. The relevant question is which institutional arrangement minimizes transaction costs while ensuring that incremental system costs are borne by the load that causes them.

That framing shifts the debate. “Bring your own generation” can mean negotiating contracts that internalize incremental costs, not reincorporating as a utility. Even once firms settle on contracting instead of vertical integration, what the contract must accomplish is defined by regulatory law.

Here, the technology firms’ language—that “growth should pay for itself”—must align with regulatory law. Under the Federal Power Act and parallel state statutes, retail and wholesale rates must be “just and reasonable.” Over decades, this standard has come to embody cost causation and non-discrimination. Customers should bear the costs they cause, and not much more. A related regulatory principle holds that customers should bear costs proportionate to the benefits they enjoy. The combination of “cost causation” and “beneficiary pays” provides guidance for data center payment that is just and reasonable.

Between these poles lies a bargaining zone that goes back to Coase: When rights are clear and transaction costs are manageable, parties can negotiate arrangements that internalize incremental cost causation and limit inefficient cost shifting onto residential customers. In that setting, the data center can procure a power service commitment with a predictable expected cost profile, while the utility can secure revenue sufficiency and reduce demand and investment risk, thereby improving capital recovery and mitigating the prospect of underutilized or stranded network upgrades. On this interpretation, “growth pays for itself” is not a claim that new load finances the entire embedded system, but rather an incremental recovery principle: Incremental demand funds the forward-looking costs it induces, with legacy sunk costs allocated under prevailing rules. This outcome emerges from contract and tariff design under regulatory oversight, rather than from political rhetoric.

The issue of jurisdiction further limits what a federal political pledge can accomplish. Most of the investments currently pushing residential bills upward—distribution upgrades, wildfire mitigation, storm resilience—fall under state public utility commission authority. The Federal Energy Regulatory Commission oversees wholesale markets and interstate transmission. It does not set retail distribution rates in any state, and presidential rhetoric cannot change that.

This distinction matters because much of today’s rate pressure reflects capital cycles that predate AI’s surge. Artificial intelligence coincides with those cost items; it did not cause them. To the extent that new utility rate cases incorporate forecasts of future data center growth, those AI-related infrastructure costs will start showing up, which is why this question is so fraught right now.

The economics become less intuitive in systems with excess capacity. Electricity networks have economies of scale, high fixed costs, and low marginal costs. When demand is weak, fixed costs are spread over fewer kilowatt-hours and average rates rise. A large, steady customer with a high load factor can improve utilization and spread those fixed costs more broadly. Under the right conditions, new incremental demand can lower average rates. In tighter systems, it can raise them. The effect depends on local capacity, regulatory treatment, and contract design. A universal claim in either direction is politically convenient and often wrong.

Overlaying all this is the uncertainty I mentioned in my last edition of Dispatch Energy. Transmission lines and generation assets last decades. Decisions made today assume something about the durability and location of AI demand. Will models become more energy-efficient? Will on-site generation expand? Will data centers migrate toward energy-rich regions? Infrastructure built for one customer may later serve many; infrastructure justified for system reliability may become indispensable to a single large load. The value creation is rarely static, even if grounded in backward-looking static accounting.

Electric power networks are complex adaptive systems governed by layered law. Costs are joint. Investments are forward-looking. Allocation is institutional, not automatic. Against that backdrop, Trump’s State of the Union pledge operates at the level of rhetorical clarity rather than system design.

Such clarity has political value. It signals fairness and attentiveness to affordability. But institutional design requires more than clarity; it requires cost causation, jurisdictional awareness, and bargaining discipline. Growth can either raise or lower average costs. Which outcome prevails depends on how firms structure the deal and under what regulatory principles.

Ensuring that growth pays for itself is a defensible objective. Achieving it depends less on presidential promises than on the careful application of just-and-reasonable principles in the institutions that actually govern electricity.

The applause line was simple. The electricity economics, as usual, are not.

Originally published at Dispatch Energy, Is AI Really to Blame for Costly Electricity Bills?