This post is coauthored by Mike Gregory & Kritika Maheshwari

Introduction

Imagine institutions that can’t be abolished by voters. Expert panels with veto power over elected officials. Constitutional protections that make budget cuts impossible. Decisions so deeply entrenched that reversing them requires 90% supermajorities—effectively permanent. All claiming legitimacy from representing people who don’t exist yet: trillions of them, thousands of years in the future. Sounds undemocratic? Here’s the thing: this isn’t a dystopian thought experiment. It’s what institutional longtermism requires. And its proponents (Schmidt & Barrett 2025) insist this is perfectly compatible with democracy. We’re going to argue why it’s not.

What Is Institutional Longtermism?

Institutional longtermism is the idea that when designing our institutions—governments, laws, regulatory bodies—we should give significant weight to their long-term effects. Not just thinking about our grandchildren but optimizing for civilizations that might span millions of years. The argument goes like this:

Future people matter morally (their lives count just as much as ours)
There could be trillions of them
We can affect their lives (by preventing extinction, for example)

Put those premises together and you get a striking conclusion: our institutions should prioritize the long-term future. Maybe even make it their primary concern. Proponents suggest democracies should reform themselves to better represent future people. They propose things like “future citizen assemblies” and constitutional protections for long-term goals. The idea is that democracy can bend itself toward the far future while remaining fundamentally democratic.

The Problem

But here’s what advocates of institutional longtermism rarely spell out in detail: what would these institutions actually look like? What powers would they need? How would they resist pressure from voters focused on today’s problems? Who gets to decide what counts as “optimal” for people living in the year 3000? And how do we stop current majorities from simply dismantling these institutions the moment they become inconvenient? Once you start filling in these details—the actual institutional machinery needed to achieve longtermist goals—something troubling emerges. The features you need aren’t tweaks to democracy. They’re replacements for democracy’s core structures.

This essay demonstrates that institutional longtermism requires five key features: permanence (institutions that can’t be dissolved), irreversibility (decisions that can’t be overturned), expertise-based authority (rule by qualified specialists), insulation from popular pressure (protection from voters), and temporal asymmetry (privileging future people over present ones).

Each of these features substitutes something essential to democracy with something fundamentally different. By the time you implement all five, you might still have elections. You might still call it democracy. But the structural relationships that actually make governance democratic—accountability to voters, the ability to change course, authority derived from popular consent—will have been systematically replaced.

We’ll use what we call the Substitution Test to show this. It’s a thought experiment: we’ll take democracy’s core temporal features and systematically replace them with longtermist alternatives. Then we’ll ask: is what’s left still recognizably democratic? Think of it like replacing parts of a car. Replace the engine: still a car. Replace the wheels: still a car. But replace the engine, wheels, transmission, steering, and frame? At some point, you’re not fixing a car—you’re building something else entirely.

The Stakes

To be clear: maybe this tradeoff is worth it. Maybe the stakes of humanity’s long-term survival justify non-democratic governance. That’s a serious conversation worth having. However, it seems that institutional longtermism’s proponents want to have it both ways—cosmic-scale optimization and democratic legitimacy. Our argument is simple: you can have one or the other, but not both. If we’re going to restructure our institutions around the far future, we need to be honest about what we’re giving up.

One Clarification

We’re not arguing against all long-term thinking in democracies. Environmental regulations, infrastructure planning, research funding—democracies can and do think beyond the next election cycle. What we’re critiquing is something more radical: the claim that democracies should fundamentally restructure themselves to optimize trajectories across cosmic timescales, and that this restructuring somehow preserves democratic governance. That’s the claim we think fails.

Why Institutions Matter (More Than You’d Think)

Before we dive in, a quick note on why longtermists focus on institutions rather than just individual behavior. Individuals have limited power. You can personally reduce your carbon footprint, donate to pandemic prevention, or advocate for AI safety. But your impact is tiny compared to what institutions can do.

Schmidt & Barrett claim that institutions, by contrast, have enormous scale and staying power. A well-designed constitution shapes society for centuries. Bad colonial institutions still affect formerly colonized nations generations later. The U.S. Constitution is 238 years old and still determines how American politics works. This “stickiness” cuts both ways. Good institutions can lock in beneficial trajectories. Bad institutions—or the absence of good ones—can foreclose valuable options for the future. So if you’re serious about affecting the long-term future, you need institutional change. That much makes sense. The question is: what kind of institutional change?

Institutional Longtermism’s Requirements

Requirement #1: Permanence (No More Dissolution)

Longtermist institutions can’t be temporary. They need to exist forever—or at least be nearly impossible to abolish. Think about what it takes to build AI safety frameworks, pandemic preparedness systems, or climate resilience infrastructure. These aren’t projects you complete in four years. They take decades, even centuries. An institution that might be dissolved after the next election can’t commit to multi-generational projects. Politicians won’t start programs their successors might kill. Experts won’t build careers in institutions that might disappear. Even worse: dissolution doesn’t just pause progress. It dissipates accumulated knowledge, disrupts ongoing work, and locks in alternative paths that become hard to reverse.

What does this look like in practice? Longtermist institutions would need:

Constitutional protection against abolition (like independent central banks, but stronger)
Very long term appointments for decision-makers
Guaranteed funding that can’t be cut during budget crises
Structural independence from other institutions that get reorganized frequently

Requirement #2: Irreversibility (No More Changing Course)

It’s not enough for institutions to be permanent. Their decisions need to be irreversible too. Here’s a crucial distinction: an institution can survive for decades while having all its important decisions overturned. Imagine a “Future Generations Council” that exists for 50 years but whose recommendations are non-binding. Or whose authority gets transferred to bodies with shorter timelines. Or whose budget gets slashed to nothing. The institution persists, but it’s been neutered. That doesn’t protect the long-term future.

Short-term political pressure creates systematic incentives to abandon long-term commitments. When present and future interests conflict—and they often do—present voters win. This pressure intensifies during crises. When recession hits, that multi-decade pandemic preparedness program looks like a tempting place to find money. And here’s the kicker: reversal creates its own path-dependence. Stop funding a program halfway through and you don’t just pause it—you redirect resources into alternative paths that become hard to abandon later. To prevent reversal, longtermist commitments need:

Supermajority requirements (say, 90% approval) to modify core policies
Constitutional entrenchment placing decisions beyond ordinary legislation
Jurisdictional protection putting authority in bodies beyond electoral reach
International treaties creating external constraints on domestic reversal

Requirement #3: Expertise-Based Authority (No More Popular Sovereignty)

Long-term expected value calculation is hard. It involves:

Forecasting technological development decades out
Modeling complex systems under deep uncertainty
Evaluating existential risk probabilities for novel threats
Navigating thorny moral questions about population ethics

These aren’t tasks most voters—or most elected officials—are equipped to handle. They require specialized knowledge: existential risk analysis, population ethics, AI safety, climate science, rigorous probability reasoning. Someone trained in these fields is clearly more competent than someone who isn’t. Even if neither can achieve certainty, expertise makes a difference.

Schmidt & Barrett argue that these problems are so complex that nobody has genuine expertise. If the future is radically uncertain, maybe no one is more competent than anyone else. We’re skeptical. Yes, the future is uncertain. But that doesn’t eliminate meaningful differences in competence. A population ethicist who’s spent years studying welfare aggregation across time is more qualified to make these judgments than someone chosen at random. An AI safety researcher understands technical risks better than the average legislator. Deep uncertainty doesn’t mean all judgments are equally good. Rather, expertise-based authority means:

Decision-making power vested in technical and philosophical specialists
Procedures that privilege expected value calculations over other considerations
Epistemic credentials (not popular election) as the basis for authority
Expert panels with binding authority over policy domains

Requirement #4: Insulation from Popular Pressure (No More Accountability)

Even if you have permanent institutions with irreversible expert decisions, there’s still a problem: voters can create pressure to change course. This pressure operates through:

Elections (throwing out politicians who support longtermist policies)
Budget battles (cutting funding for programs with no visible near-term payoff)
Public opinion (media campaigns against “waste” on far-future concerns)
Protests and activism

And here’s the structural problem: present citizens experience present costs but not future benefits. The generation funding pandemic preparedness bears the financial burden but may never see the catastrophe prevented. People asked to sacrifice today for benefits they’ll never experience have rational reasons to oppose it.

Normally, political accountability is good. Politicians answerable to voters, removable if they fail, responsive to public concerns—these are features of democracy. But for longtermism, accountability becomes a problem. When politicians must justify decisions to present voters, and future people can’t vote, short-term interests systematically win. During crises, long-term projects become easy targets. Democratic accountability—the virtue—becomes the obstacle. Insulation requires:

Independent funding streams that legislatures can’t redirect
Long terms without reelection for decision-makers
Structural barriers against public demands for reversal
Constitutional provisions placing commitments beyond ordinary politics

An ideal implementation might include an institution whose budget is constitutionally guaranteed, whose officials serve extended terms (e.g., 20 years), and whose core mission cannot be altered by conventional legislation or short-term public pressure. This design ensures that future-oriented decisions remain protected from immediate political considerations.

Requirement #5: Temporal Asymmetry (Future People Count More)

When present and future interests conflict, future interests must win. And they must win by a lot. Here’s the logic: If humanity survives for millions of years, future populations could include trillions or quadrillions of people. That’s a vast numerical asymmetry. Present population: ~8 billion. Potential future population: quadrillions. If future people matter morally (premise #1), and there are quadrillions of them (premise #2), then their aggregated interests outweigh ours by many orders of magnitude. A policy that slightly increases existential risk might affect quadrillions of people. That makes its expected value astronomically negative compared to policies affecting only the present.

This isn’t just “care about the future.” It’s a specific hierarchical relationship:

Present citizens’ preferences receive some weight
But when they conflict with future populations’ interests, future interests must prevail
The difference in weight is enormous—proportional to population size

This requires:

Explicit weighting of present vs. future interests in decision-making
Override mechanisms letting future-oriented priorities trump present preferences
Institutional commitment to resisting present political pressure

To Summarize

So here are the five requirements:

Permanence—institutions that can’t be dissolved
Irreversibility—decisions that can’t be overturned
Expertise-based authority—rule by qualified specialists
Insulation from popular pressure—protection from voters
Temporal asymmetry—future interests trump present ones

Each one chips away at democratic governance. None requires “strong longtermism”—the radical view that only long-term value matters. Even the moderate position that long-term value deserves significant weight generates these requirements. They’re not optional extras. They follow logically from taking long-term expected value seriously. Now let’s see what happens when we actually implement them.

Democracy’s Four Features

Democracy isn’t just “people voting.” It has a specific temporal structure—a distinctive way of organizing authority through time. Four features define this structure. Each plays a crucial role in making governance democratic rather than something else.

Feature #1: Periodicity (Regular Renewal of Authority)

Officials must return to citizens at fixed intervals to have their power renewed or transferred. In practice: elections every 2, 4, or 6 years. Term limits. No one holds power indefinitely. Every position of authority eventually returns to the people for reauthorization. This isn’t just a procedural detail. It’s a structural principle: authority in democracy is always temporary, always conditional, always subject to periodic review.

Feature #2: Reversibility (Future Majorities Can Change Course)

A policy enacted today can be repealed tomorrow. No generation permanently binds its successors. A constitutional amendment can be amended again. A budget allocation can be redirected. Future citizens inherit the authority to chart their own course, not obligations to honor their predecessors’ choices. Why this matters: it expresses temporal equality. The judgment of citizens in 2025 has no inherent superiority over the judgment of citizens in 2075.

Feature #3: Contemporaneity (Rulers and Ruled Share a Timeline)

Officials answer to the people who live alongside them and experience the consequences of decisions within their lifetimes. This creates accountability in real time. A senator making healthcare policy answers to citizens who’ll use that healthcare system and can vote her out next election. What makes this democratic: the mutual presence of governors and governed. They share a temporal location, face common risks, and can hold each other accountable.

Feature #4: Openness (The Future Remains Contestable)

The future stays politically undetermined. We refuse to lock in any particular vision of what collective life should become. Future directions remain subject to ongoing determination rather than fixed by present decisions. The future belongs to future citizens to define through their own democratic processes, not to present citizens to determine on their behalf. We can make decisions today. But we can’t permanently close off alternative paths tomorrow’s citizens might want to pursue.

Democracy’s “Interval Logic”

Together, these four features create something distinctive: what we call democracy’s interval logic. Here’s what we mean: Democratic decision-making is cyclical. Elections, legislative sessions, public deliberations—they all happen at regular intervals. Each cycle renews legitimacy by subjecting authority to the judgment of the living. Authority doesn’t flow from tradition, expertise, or historical mandate. It flows from the ongoing consent of present citizens. This creates bounded temporality. Legitimacy is produced and renewed within specific time intervals, not projected indefinitely. The interval between elections marks the boundary of unconditional authorization. Beyond that boundary? Authority requires renewal through fresh democratic judgment.

This interval structure does four crucial things:

Prevents permanent power concentration—authority stays temporary and contested
Enables democratic correction—citizens can evaluate outcomes and adjust course
Expresses political equality—no generation gets permanent authority over later ones
Maintains possibility of change—alternative futures aren’t foreclosed

Here’s the tension: this temporal boundedness enables self-rule but necessarily enforces what longtermists call “short-term orientation.” Decisions must stay revisable within intervals measured by electoral cycles and generational horizons. When policy commitments extend beyond the temporal reach of those who authorized them, they transform from democratic decisions into inherited constraints. That’s precisely what democracy’s structure aims to prevent.

Now contrast this with what longtermism requires (remember our five features?):

Permanence (not periodicity)
Irreversibility (not reversibility)
Expertise-based authority (not contemporaneous accountability)
Insulation from popular pressure (not responsiveness to present citizens)
Temporal asymmetry (not temporal equality)

These stand in direct opposition to democracy’s temporal infrastructure. Where democracy requires periodic renewal, longtermism demands permanence. Where democracy enables revision, longtermism requires entrenchment. Where democracy insists on contemporaneous accountability, longtermism establishes temporal hierarchy. Where democracy preserves openness, longtermism seeks determinacy.

The Substitution Test

We’ve shown what longtermism requires. Now let’s see what happens when you implement those requirements. Our Substitution Test first requires us to take democracy’s core features and systematically replace them with longtermist alternatives, and then ask: is what’s left still democracy?

Substitution #1: From Periodicity to Permanence

What gets replaced: Regular elections and term limits—those “temporal punctuation marks” when authority returns to the people.

What replaces it: Enduring bodies with indefinite terms, insulated from electoral pressure to preserve focus on long-term trajectories.

Example: A Future Generations Council with longterm (potentially lifetime) appointments.

The justification sounds reasonable: we need continuity. We need to protect long-term interests from short-term political cycles. But permanence doesn’t just alter a procedural feature. It removes a constitutive element of democratic legitimacy: the people’s ultimate authority over those who govern them. Officials no longer depend on citizen approval for their continued mandate. Citizens lose the power to dismiss leaders whose judgment they reject. What changes isn’t policy content—it’s the structure of authorization itself.

Democratic authorization is conditional and renewable, grounded in ongoing consent. Longtermist authorization, by contrast, appeals to epistemic credentials and moral necessity. Some goals are too important to be revalidated through elections. The shift from periodic to permanent authority transforms legitimacy from democratic consent to expert judgment about optimal futures.

But, here’s the tension: in democracy, nothing is permanent. Voters can always change course. That’s not a bug—it’s the defining feature. Today’s majority can’t bind tomorrow’s majority. Each generation gets to decide for itself. Permanence removes this power. Once you constitutionally entrench a longtermist institution, future voters can’t get rid of it even if they think it’s doing harm. You might say: “But we have permanent institutions now! The Supreme Court, the Federal Reserve...” True. But notice what happens when you multiply this permanence across all the institutions needed for comprehensive longtermism. At some point, democratic choice becomes impossible.

Substitution #2: From Reversibility to Entrenchment

What gets replaced: The ability of future majorities to revise or overturn present decisions.

What replaces it: Constitutional or structural entrenchment that constrains future revision.

Example: A constitutional amendment requiring 90% supermajority to alter AI safety protocols.

Again, the justification sounds compelling: we’re preventing reckless future policy shifts that could increase existential risk. We’re protecting crucial safeguards. But the effect is to erode democracy’s openness to ongoing popular sovereignty. Future citizens inherit institutional trajectories they lack the authority to revise. What’s lost isn’t mere policy flexibility—it’s structural equality across generations. Earlier generations exercise enduring power that later ones can no longer reclaim.

Think about it: Suppose that current generations in 2025 make a decision. We constitutionally entrench it. Citizens in 2075 disagree with our choice. But they can’t change it without achieving a 90% supermajority. We’ve bound them to our judgment. Their democratic authority has been subordinated to ours.

However, democracy is built on reversibility. A policy passed today can be repealed tomorrow. A constitutional amendment can be amended again. This expresses a fundamental principle: the judgment of citizens in 2025 has no inherent superiority over the judgment of citizens in 2075. Future generations inherit the authority to chart their own course, not obligations to honor their predecessors’ choices. Irreversibility breaks this. It says: “We know better than future majorities will. Our judgment must override theirs.” That might be true! Maybe we really do know that future people shouldn’t reverse our AI safety protocols. But it’s not democratic.

Substitution #3: From Contemporaneity to Temporal Asymmetry & Insulation

This substitution is more complex—it operates through two mechanisms working together.

What gets replaced: Accountability between officials and those they presently govern, within shared temporal horizons.

What replaces it: Temporal asymmetry (accountability redirected to future persons) + insulation from popular pressure (protection from present citizens).

The first mechanism operates as follows: Officials become fiduciaries for future generations. They’re accountable not to current citizens but to their own interpretation of posterity’s interests. Consider boards representing “future stakeholders” with authority to override present majorities when decisions threaten long-term welfare. This might enhance intergenerational justice. But it severs the democratic link between government and governed. Present citizens can no longer meaningfully hold officials accountable if officials claim legitimacy from fidelity to future persons rather than responsiveness to present electorates.

The second mechanism operates as follows: Even when officials acknowledge responsibility to present citizens, longtermist institutions must protect against how those citizens normally influence governance: Elections, budget pressures, public opinion, protests, and media scrutiny.

What this looks like: Independent funding streams immune to legislative cuts. Decision-makers with extended terms beyond electoral cycles. Structural barriers against demands for policy reversal. The result? Officials become unaccountable in the democratic sense. They can’t be pressured, sanctioned, or removed by those they govern. When interests diverge—and they often do—officials must override current preferences. Democratic resistance isn’t legitimate contestation. It’s a short-term bias requiring correction.

Together, both mechanisms dismantle contemporaneity’s core function: ensuring that those who exercise power remain answerable to those over whom power is exercised. Officials claim authority from future persons who can’t hold them accountable and are protected from present persons who could. Can you have democracy without accountability to the governed?

Here’s some tension. Officials become unaccountable in the democratic sense. They can’t be pressured, sanctioned, or removed by citizens. When voter preferences conflict with long-term optimization, voter preferences lose. This isn’t a side effect—it’s the design. Future people’s interests must be protected from present people’s political power. But democracy is present people exercising political power. When you insulate institutions from that power, you’ve removed them from democratic control. When a senator makes healthcare policy, she answers to citizens who’ll use that healthcare system and can vote her out in the next election. Temporal asymmetry breaks this. Officials claim legitimacy from fidelity to future persons, not responsiveness to present citizens. When interests diverge—and they often do—officials must override current preferences.

Substitution #4: From Openness to Expert Determinacy

What gets replaced: The future as an open domain of ongoing contestation and collective redefinition.

What replaces it: Fixed developmental paths determined by expert calculation and protected from political volatility.

This also operates through two mechanisms.

The first one involves exercise of expert-based authority. Decisions about long-term trajectories get removed from democratic contestation. They’re vested in those with technical and philosophical qualifications to calculate expected value. The basis of legitimate authority shifts: from popular authorization to epistemic credentials. Think of a panel of AI safety experts with binding authority over technology policy. Democratic leaders become implementers of expert judgment rather than stewards of collective self-determination. The future’s direction becomes a technical question with correct answers, not a political question requiring ongoing deliberation.

Here’s the second mechanism: These expert determinations get locked in against future revision. Think of locked-in value functions for AGI development that fix the moral objectives of powerful technologies to ensure safe alignment. This secures predictability and safety. But it transforms the future from a shared project of deliberation into an object of optimization. The result is institutional paternalism. Insulated expert bodies decide what futures are valuable and how to pursue them—even when the public might reject the tradeoffs. Democratic leaders become managers of predetermined goals, not stewards of collective judgment.

In a democracy, authority flows from popular consent, not expertise. Yes, we elect people we think are competent. But their authority comes from being chosen by voters, not from their credentials. And voters can remove them if they’re dissatisfied—regardless of whether experts think that’s wise. Expertise-based authority reverses this. It says: “These decisions are too important to leave to voters. Qualified specialists should decide.” The future’s direction becomes a technical question with correct answers, not a political question requiring collective deliberation. Maybe that’s right! Maybe some questions really are too complex for democratic decision-making. But then we’re not talking about democracy anymore. We’re talking about epistocracy—rule by the knowledgeable.

The Cumulative Effect

We started with democracy’s four temporal features: Periodicity, Reversibility Contemporaneity Openness. We systematically replaced them with longtermist alternatives: Permanence, Irreversibility, Temporal asymmetry and Insulation, Expertise-based determinacy. What’s left?

Citizens no longer author the trajectory of their collective political life. Experts determine optimal civilizational paths. Institutional structures ensure conformity. Present majorities can’t hold officials genuinely accountable officials claim legitimacy through service to future interests. Future majorities inherit obligations to honor commitments made by predecessors. Commitments they never authorized and can’t revoke. The political future has transformed from a space of possibility requiring ongoing collective determination into a predetermined trajectory requiring stable management.

Now, you could still hold elections in this system. You might still call it democracy. But the structural relationships that make governance democratic are gone. Think again about what democratic legitimacy requires:

Periodic authorization through bounded intervals
Ongoing accountability to the living
Mutual recognition among contemporaries
Retained authority for revision

Each of these depends on the concrete institutional features we just replaced. Periodicity creates the intervals when authorization returns to citizens. We eliminated those intervals. Reversibility ensures future citizens retain authority to revise present choices. We entrenched decisions against revision. Contemporaneity situates accountability within shared temporal horizons. We redirected accountability to future persons and insulated officials from present ones. Openness preserves political futures as domains of ongoing collective determination. We fixed trajectories through expert calculation. These aren’t peripheral attributes. They’re constitutive infrastructure. They’re what democracy’s “interval logic” actually consists of.

The Verdict: Where’s democracy?

The Substitution Test reveals something important: democracy’s temporal features aren’t optional extras you can swap out. They’re essential elements. Remove them, and democratic legitimacy dissolves—because the structural relationships that make governance democratic cease to exist. The resulting system might be good. It might optimize for long-term welfare. It might embody sophisticated technical reasoning about civilizational futures. But it lacks the defining features of democratic governance:

Authority derived from and regularly renewed through judgment of the living
Ongoing accountability to present citizens within shared temporal horizons
Temporal equality refusing permanent power for any generation
Political futures remaining open to collective redefinition

Here’s what matters: this isn’t about contingent institutional details. It’s a structural necessity. Democracy’s interval logic—the bounded temporality through which legitimacy is produced and renewed—cannot survive the substitutions longtermism requires. Remove or replace these features, and democratic legitimacy dissolves. Not as a side effect, but as a logical consequence. This reveals a genuine tradeoff, not a reconcilable tension.

The Bottom Line

The institutional requirements that follow from taking long-term expected value seriously—even in its weak form—are the very features that dissolve democracy’s constitutive temporal infrastructure. You can have institutions optimized for cosmic-scale futures. Or you can have democratic governance characterized by bounded intervals of authorization and accountability. The Substitution Test demonstrates you cannot have both. Schmidt and Barrett’s reconciliation project fails not because of inadequate institutional imagination. It fails because it attempts to preserve democratic legitimacy while systematically replacing the temporal structures through which that legitimacy operates. It’s like trying to preserve the car while replacing everything that makes it a car. At some point, you’re just building something else.

Can Democracy Survive Longtermism?