Redistributed

The complaint that "regulation can't keep up with technology" has become a cliché. Like most clichés, it gestures toward something real while obscuring the mechanism. The problem is not that governance is slow and technology is fast. The problem is that different parts of society — political systems, legal frameworks, cultural norms, economic structures, international institutions — adapt to technological change at different rates, creating gaps between them. Those gaps become exploitable. And the actors best positioned to exploit them are the ones who least need the protection that governance provides.

Renaud Bougueng T., in his March 2026 strategic analysis AI and Human Civilization: Strategic Foresight, gives this phenomenon a name: institutional decoherence. He defines it as "a profound, long-term civilizational risk where technology races far ahead of our political, social, and cultural capacity to adapt." The term is borrowed from physics, where decoherence describes a quantum system losing its coherent properties when it interacts with its environment. The analogy is precise: institutions that once functioned as a coherent system of checks and balances lose their alignment, their ability to reinforce each other, their capacity to constrain power.

This is not a metaphor for slowness. It is a description of structural fragmentation.

The Mechanism

To understand institutional decoherence, consider how governance normally works. A new technology emerges. Markets adopt it. Problems become visible. Cultural norms shift. Legislatures respond. Courts interpret. International bodies coordinate. Agencies enforce. Each of these systems operates on a different timescale, but in periods of moderate change, the gaps between them remain manageable. Harm occurs, but it is bounded. Correction is possible.

Decoherence occurs when the rate of technological change exceeds the adaptive capacity of these systems not individually, but in their relationships to each other. Markets adopt faster than legislators understand. Courts interpret laws written for prior technologies. International bodies negotiate while deployments multiply. Cultural norms haven't crystallized because the technology keeps changing shape.

David Collingridge identified the central paradox in 1980: during a technology's early stages, when it is still malleable, we lack the information needed to regulate it wisely. By the time the impacts are clear, the technology is entrenched — economically, socially, institutionally — and change becomes expensive and politically contested. Collingridge called this the "dilemma of control." Institutional decoherence is what happens when that dilemma repeats across multiple technologies simultaneously, each advancing on its own timeline, each outrunning governance in different ways.

Historical Precedent

This is not new. The pattern recurs with each major technological wave, as Cesare Marchetti's long-wave analysis of technology adoption demonstrates. What changes is the scale and speed.

Electricity (1880s–1930s). Thomas Edison opened his first commercial power station in 1882. The first serious federal regulation of the electricity industry — the Federal Power Act — did not arrive until 1920, nearly four decades later. In the interim, private monopolies consolidated control over generation and distribution, pricing power flowed to utility companies, and millions of rural Americans were left without access until the Rural Electrification Act of 1936. The regulatory gap didn't just delay good policy. It allowed a distribution of power — both electrical and political — that took decades of public investment to partially correct.

Financial derivatives (1990s–2008). The Commodity Futures Modernization Act of 2000 explicitly deregulated over-the-counter derivatives, removing them from CFTC oversight. Financial engineers created instruments — collateralized debt obligations, credit default swaps — that regulators did not understand, rating agencies could not evaluate, and investors could not price. The gap between financial innovation and regulatory capacity produced the 2008 crisis, which destroyed $10 trillion in household wealth and cost 8.7 million jobs in the US alone. The redistribution was stark: banks received bailouts; homeowners received foreclosure notices.

Social media (2010s–present). Facebook reached a billion users in 2012. The first serious legislative attempt to address platform governance in the US — Section 230 reform proposals — remains unresolved in 2026. In the intervening years, platforms became the primary infrastructure for public discourse, political organizing, and information distribution. Content moderation policy was set by corporate trust and safety teams, not democratic processes. The governance gap allowed platforms to optimize for engagement — which meant optimizing for outrage, polarization, and algorithmic amplification of extreme content — while bearing almost no liability for the results.

In each case, the pattern is the same: the gap between technological capability and institutional capacity does not remain neutral. It is colonized by the actors with the resources to operate in ungoverned space.

AI-Specific Decoherence

Artificial intelligence presents a particularly severe case. The technology is advancing across multiple domains simultaneously — language, vision, reasoning, scientific discovery, code generation — and each domain intersects with different regulatory frameworks, professional norms, and social expectations. The decoherence is multidimensional.

Legislative fragmentation. The EU AI Act, the world's most comprehensive AI regulation, was finalized in 2024 after three years of negotiation. By the time its provisions took effect, the systems it was designed to regulate had changed fundamentally. The Act's risk classification framework — built around specific use cases — struggles to account for general-purpose AI systems that can be adapted to any use case post-deployment. Regulation designed for one generation of technology confronts the next generation's capabilities.

Executive instability. In the United States, AI governance has been conducted primarily through executive orders — instruments that can be reversed overnight by a new administration. The oscillation between the Biden administration's comprehensive AI executive order and subsequent rollbacks demonstrates a governance mechanism that cannot provide the stability businesses, workers, or civil society need to plan around. Executive orders are not governance. They are gestures.

International incoherence. International AI summits have produced declarations, communiqués, and voluntary commitments. They have not produced governance. The Bletchley Declaration, the Seoul Ministerial Statement, the Paris communiqué — each represents diplomatic achievement and institutional failure simultaneously. There is no international body with the authority, capacity, or mandate to govern AI development. The IAEA analogy frequently invoked has a fatal flaw: nuclear weapons required state-scale resources. Frontier AI models are built by private companies that operate across jurisdictions and answer primarily to shareholders.

Corporate self-regulation as default. In the absence of binding governance, the rules are being set by the companies building the systems. This is not accidental — it is structural. Labs publish safety research, propose evaluation frameworks, and define acceptable use policies. This is presented as responsibility. It is also regulatory capture in its earliest, least visible form — the phase where industry defines the categories that future regulation will use. As Langdon Winner argued in "Do Artifacts Have Politics?", the design choices made during this period embed political decisions into technical infrastructure. Those decisions become harder to reverse once systems are deployed at scale.

Who Benefits, Who Pays

Institutional decoherence is not a neutral condition. It redistributes power, wealth, and risk along predictable lines.

Those who benefit are actors with the resources to operate effectively in ungoverned space. Large technology companies can deploy systems globally, establish market dominance, and shape norms before regulations constrain them. Well-resourced actors can hire lawyers to navigate fragmented regulatory landscapes, lobby across multiple jurisdictions, and structure operations to exploit gaps between national frameworks. First movers capture markets and data — both of which create barriers to entry that persist even after regulation arrives.

Those who pay are actors without the resources to adapt. Workers displaced by AI automation face retraining systems that don't exist yet, unemployment insurance designed for different kinds of job loss, and labor protections written for a prior economy. Communities affected by algorithmic decision-making — in hiring, lending, criminal justice, public benefits — lack the technical capacity to identify when systems are discriminatory and the legal frameworks to seek redress. Developing nations face AI systems built in and optimized for wealthy countries, deployed into their markets without adequate consideration of local contexts, values, or needs. The lobbying infrastructure that shapes AI policy in Washington and Brussels is not available to a farmer in Bihar or a gig worker in Nairobi.

This is the redistribution problem at its core. Decoherence doesn't just slow down good governance — it creates the conditions for value extraction. Every month that governance lags behind deployment is a month in which market positions are consolidated, data moats are deepened, and the cost of future regulation increases.

What Coherent Governance Would Require

If institutional decoherence is the diagnosis, what does coherence look like?

Not perfect synchronization — that is neither possible nor desirable. Governance should not move as fast as technology. Some friction is protective. The question is whether the gaps between institutional systems are manageable or exploitable.

Adaptive regulation. Static rules applied to rapidly changing technology will always decohere. Governance frameworks need built-in mechanisms for revision — sunset clauses, mandatory review periods, capacity for rapid updating without full legislative cycles. The UK's proposed "pro-innovation" approach attempted this but lacked enforcement teeth. The EU's AI Office could evolve in this direction if adequately resourced.

Sandboxed experimentation. Bougueng proposes regulatory sandboxes — controlled environments where new AI applications can be deployed under observation, with clear boundaries and mandatory reporting. This addresses the Collingridge dilemma directly: sandboxes generate the information needed for wise regulation while constraining the technology's scope during the learning period. Singapore's AI Verify sandbox and the EU AI Act's sandbox provisions point in this direction — and early results from financial services sandboxes (which the UK's FCA has operated since 2016) suggest the model can work when adequately resourced and politically supported.

Institutional capacity building. Regulators cannot govern what they do not understand. This requires sustained investment in technical capacity within government agencies — not just hiring a few technologists, but building institutional knowledge that persists across administrations. The UK's AI Safety Institute represents one model; democratic governance structures that include affected communities represent another. Both are necessary. Technical expertise without democratic accountability produces capture. Democratic process without technical capacity produces theater.

International coordination with teeth. Voluntary commitments among nations have not constrained corporate behavior. Effective international governance requires either binding agreements with enforcement mechanisms or — more realistically — coordination among regulatory blocs that collectively represent markets large enough to compel compliance. The "Brussels effect" — where EU regulations become de facto global standards because companies find it easier to comply everywhere than to maintain separate systems — suggests a model, but it requires the EU to actually enforce its rules.

The Stakes

Institutional decoherence is not a theoretical risk. It is the current condition. The question is whether it deepens — creating permanent governance gaps that entrench existing power structures — or whether societies build the institutional capacity to close the gaps before they become irreversible.

The precedents are mixed. On one hand, electricity took fifty years to regulate equitably. Financial derivatives were deregulated on purpose. Social media governance remains unresolved after fifteen years. In each case, the distributional consequences of the governance gap persisted long after regulation eventually arrived. The homeowners who lost their homes in 2008 did not get them back when Dodd-Frank passed in 2010. On the other hand, the Montreal Protocol addressed ozone depletion with remarkable speed, and the EU's GDPR, while imperfect, demonstrates that democratic institutions can write rules for complex technology that companies actually follow. The question is whether AI governance will follow the financial derivatives pattern (slow, too late) or the environmental protection pattern (imperfect but functional).

AI governance will follow this pattern unless institutional capacity matches the ambition of the technology. That requires money, expertise, political will, and — most fundamentally — a recognition that the governance gap is not a failure of speed but a redistribution of power. Every day without coherent governance is a day when the benefits of AI flow to those who already have the most and the costs fall on those who can least afford them.

The question is not whether governance will eventually catch up. It always does, in some form. The question is what has been redistributed — permanently — in the interim.

Sources

Bougueng T., Renaud. AI and Human Civilization: Strategic Foresight. March 2026.
Collingridge, David. The Social Control of Technology. Frances Pinter, 1980.
Winner, Langdon. "Do Artifacts Have Politics?" Daedalus 109, no. 1 (1980): 121–136. JSTOR
Marchetti, Cesare. "Fifty-Year Pulsation in Human Affairs: Analysis of Some Physical Indicators." Futures 17, no. 3 (1986): 376–388.
Stigler, George J. "The Theory of Economic Regulation." Bell Journal of Economics and Management Science 2, no. 1 (1971): 3–21.
Regulation (EU) 2024/1689 of the European Parliament and of the Council (AI Act). EUR-Lex
Marchetti, Alessandro, and Filippo Ferroni. "The Pacing Problem and the Future of Technology Regulation." Oxford Journal of Legal Studies 42, no. 4 (2022): 941–968.

General

Institutional Decoherence: When Technology Outruns Governance