Skip to main content
September 2026
← All 28 pairs

Pair 1×7 · structured conjecture

Access to capability × Governance coordination

Every corner solves the access question by growing a new gate somewhere else — governance doesn't remove chokepoints, it decides who's allowed to relocate them: the lab door, the border, the payment rail, or the walled garden you build around yourself.

The full 2×2. Click to enlarge.

The four scenarios

Two questions: how many hands can wield frontier capability, and whether the world governing it can actually coordinate or is racing itself. The naive reading treats these as independent — lock down access and you get safety, whatever the politics. What actually happens is that chokepoints are what give governance something to hold onto, and that handle never stays a safety instrument for long: it becomes a gate, and the gate migrates — from the lab door, to the border, to the payment rail, to your own walled garden — but it never disappears.

What do these axes mean? ▸

Each axis is a spectrum. The card takes the two poles of each and reads off what the corner where they meet produces downstream.

Axis 1 · Access to capability

Low
Chokepointed a few controlled actors
High
Proliferated frontier capability widely available

Who can use or deploy high-end capability — a few controlled actors, or anyone with a laptop.

Axis 7 · Governance coordination

Low
Fragmented race no binding coordination
High
Enforced coordination binding rules at home and abroad

Whether AI governance is binding and internationally coordinated, or fractured into a competitive race with no shared rules.

Chokepointed · Enforced Coordination

The Intelligence Passport

The obvious policy win curdles from the inside. Verifying the chokepoint holds requires an auditor as capable as the audited, so regulator and regulated fuse into one institution with no external referee left. And if the regime actually works, nothing visible happens, which is exactly the condition under which a democracy dismantles a costly, paternalistic-looking success it can't see the benefit of. Access rights turn into a traded diplomatic asset (compute-for-bases), and a handful of safety teams' calls become the world's de facto speech policy by default, not design.

Proliferated · Enforced Coordination

The Open Rulebook

Unable to control the weights, coordination bites at behavior instead — continuous monitoring of compute, outputs, and use — so the corner that promised no wall buys a surveillance settlement instead, and civil liberties become the actual battlefield, not AI capability. The chokepoint doesn't vanish, it relocates one layer down to payment processors, insurers, and certifiers, who become the real regulator nobody elected. Done well this is a genuine governed commons with shared defensive tooling; done badly it's a hidden compliance tax that only large actors can pay.

Chokepointed · Fragmented Race

The Compute Embassy

Arms control worked because you can count silos; you cannot count cognition, so worst-case assumption becomes the only rational posture, permanently. Champion labs fuse with states until alignment research itself gets classified as a national secret — the one body of knowledge everyone benefits from sharing fragments into compartments, and whistleblowing becomes espionage. The few datacenters that exist become first-strike targets. Fragmentation still preserves some exit from any single global regime, though, and no one actor can dictate terms to everyone else.

Proliferated · Fragmented Race

The Unmanaged Commons

It looks like pure chaos, but the real structure is the handful of stabilizers that emerge anyway. Regulatory havens host the riskiest deployments while panic regulation clamps down elsewhere, so nobody experiences governance as right-sized, and the offense-defense balance splits by domain rather than holding uniformly. As synthetic content saturates everything, trust re-physicalizes: notaries return, in-person verification becomes essential, and people voluntarily re-chokepoint themselves into curated, filtered realities — inviting the access oligopoly back in through the consumption layer.

How this whole reading could be wrong

What would undo the pair's thesis — not any single corner.

Two things could unseat this reading. First, governance texture may dominate lived experience more than access topology does: coordinated worlds converge on similar surveillance textures whether access is chokepointed or proliferated, which would mean the grid is better drawn on the governance axis alone. Second, the fragmented-and-proliferated corner may be thinner than it looks — some of its apparent richness is the ordinary churn of an ungoverned space rather than a distinct structural scenario.

Related pairs

Other cards that share one of these variables.

More pairs with Access to capability

More pairs with Governance coordination