The Narrow Corridor
The Narrow Corridor: States, Societies, and the Fate of Liberty (2019) presents liberty as neither the absence of state power nor the abundance of state power, but as a dynamic equilibrium between a capable state and an organized society strong enough to constrain it. Too little state capacity yields the absent Leviathan — lawlessness, predation by private power. Too much state capacity uncontested by society yields the despotic Leviathan — tyranny. The corridor between them is narrow, entry is historically rare, and maintenance requires continuous effort from both sides. AI, Acemoglu and Robinson argue, threatens to force societies out of the corridor by strengthening state and corporate surveillance capacity while weakening the societal organization that would balance it.
The Corridor Already Closed — Contrarian ^ Opus
There is a parallel reading in which the corridor framework mistakes a temporary historical anomaly for a sustainable equilibrium. The mid-20th century balance between state capacity and organized labor wasn't the natural state of industrial democracy—it was the byproduct of specific material conditions (mass factory production requiring concentrated workforces, broadcast media creating shared public spheres, Cold War pressures forcing concessions) that have already dissolved. AI didn't narrow the corridor; it revealed that the corridor was always an illusion sustained by technological constraints on coordination and control.
From this view, what Acemoglu calls "corridor maintenance" was actually the expensive overhead costs of limited information processing. Unions weren't expressions of inherent societal capacity—they were coordination solutions to the problem of dispersed workers under conditions where management couldn't monitor individuals. Investigative journalism wasn't democratic accountability—it was the information arbitrage available when states and corporations couldn't process their own data. The "Red Queen race" wasn't a healthy dynamic equilibrium—it was friction masquerading as freedom. AI doesn't threaten the balance; it eliminates the inefficiencies that made balance necessary. The real question isn't how to preserve corridor conditions in an AI world, but whether democratic governance was ever more than a temporary adaptation to information scarcity, now being correctly superceded by more efficient forms of coordination and control.
In the AI Story
The corridor framework refines the inclusive-extractive distinction by adding a temporal and dynamic dimension. Inclusive institutions are not a stable endpoint but an ongoing achievement. They require societal capacity — unions, press, civic associations, political parties — that matches and constrains state capacity. When state capacity grows faster than societal capacity (the modern surveillance state, now AI-enabled), the corridor narrows. When societal capacity collapses (deunionization, media decline, atomization), the corridor narrows from the other side.
AI's threat to the corridor operates on multiple fronts simultaneously. State surveillance capabilities expand dramatically with AI-enabled pattern recognition, predictive policing, and behavioral modeling. Corporate behavioral modification capabilities expand through engagement optimization and personalized persuasion. Meanwhile, the organized societal forces that would constrain both — labor unions, local press, civic associations — are being further weakened by AI-driven labor market disruption and the attention economy's degradation of shared public space.
The Red Queen effect — the mutual running between state and society that keeps both capable — operates through specific institutional channels that AI is disrupting. Collective bargaining generates the information and coordination that lets workers constrain firms; AI-mediated individualized work contracts bypass it. Investigative journalism generates the accountability that constrains governments; AI-disrupted media economics threaten it. Each disruption weakens one side of the balance.
Applied concretely, the framework reframes the AI governance question. The goal is not to regulate AI per se but to preserve the corridor conditions — capable state, organized society — within which AI's benefits can be realized without its dangers dominating. This is a different policy agenda than traditional AI safety: it focuses on labor organization, press sustainability, civic infrastructure, and democratic accountability as much as on model capabilities or deployment restrictions.
Origin
The book appeared in 2019, before ChatGPT made AI's political economy urgent. The framework has been updated in Acemoglu's post-2022 writings to incorporate AI specifically, including his 2024 Economic Policy piece and various op-eds emphasizing that the corridor is narrowing under AI pressure.
Key Ideas
Liberty is a balance, not a state. Freedom requires both capable states and organized societies — neither absent nor dominant.
The corridor is historically rare. Most societies have not entered it; most that have entered have been pushed out by the collapse of either pole.
AI threatens both sides of the balance. It strengthens state and corporate capacity while weakening the societal organization that would constrain them.
Preserving the corridor is the primary AI governance task. This is a different agenda than either laissez-faire or state-centric regulation.
Debates & Critiques
Political scientists including Francis Fukuyama have argued the corridor framework understates the role of national political culture in sustaining liberty. Acemoglu's response is that culture matters but operates through the institutional balance the framework describes, rather than as an independent variable.
Appears in the Orange Pill Cycle
Friction as Constitutional Necessity — Arbitrator ^ Opus
The right framing depends on whether you're asking a historical-diagnostic question or a normative-strategic one. Historically, the contrarian view carries weight (65/35): the mid-century corridor conditions *were* partly artifacts of technological limitations, and many "natural" checks on power were actually coordination costs in disguise. Acemoglu's framework does romanticize what may have been contingent friction. But this historical reading doesn't settle the normative question it purports to answer.
Strategically—if we take liberty as a value worth preserving—Acemoglu's framework is approximately correct (75/25). Even if the historical corridor was artifactual, the functional requirement it names remains: liberty requires contestation, and contestation requires organized capacity on both sides. Whether that organization emerged from "natural" societal strength or from technological constraints is less relevant than whether it can be deliberately reconstructed. The policy agenda Acemoglu identifies (labor organization, press sustainability, civic infrastructure) isn't nostalgia—it's the minimal institutional requirements for democratic governance under any information regime.
The synthesis the topic benefits from reframes "friction" not as inefficiency to be eliminated but as constitutional necessity. Some frictions (transaction costs, coordination limits) are indeed waste. Others (checks and balances, due process, deliberation time) are features. The AI question is which frictions encoded democratic values and how to preserve those values when the frictions themselves become technologically optional. The corridor framework, properly understood, is a vocabulary for this distinction—not a claim that all historical arrangements were optimal, but a structural analysis of what arrangements liberty requires.
Further reading
- Acemoglu and Robinson, The Narrow Corridor: States, Societies, and the Fate of Liberty (Penguin Press, 2019)
- Acemoglu, 'The Simple Macroeconomics of AI,' Economic Policy (2024)
- Francis Fukuyama review in The American Interest (2019)
- Acemoglu, various Project Syndicate columns 2023–2025