Too Big to Hide
An intelligence analyst's rival state stops making mistakes — in every domain at once. A vignette about the far side of the Dark Basin: the threshold where a hidden system's effects become undeniable.
This story is from the AI Scenario Explorer, which pairs eight forces shaping AI, two at a time, into 28 maps — each map with four corners, each corner a different way that pairing could play out. A vignette takes one corner and makes it concrete: structured conjecture, not prediction. About the project →
For nineteen years, Elena's trade has been other governments' mistakes.
States are legible through their errors — the procurement front that reuses a phone number, the negotiator who concedes what his ministry won't abandon for another week. Competence hides; error is the wobble that gives the orbit away.
Eighteen months ago, the rival stopped making mistakes. Not in one domain — in all of them, at once. Every desk explained its own corner: better tradecraft, new people, luck. Elena's job is the cross-domain view, and the cross-domain view does not support luck.
Her case rests on one fact about bureaucracies: competence does not correlate across them. A foreign ministry, a customs service and a cyber unit do not improve together, in step, for eighteen months — unless the improvement is not institutional. Something single sits behind many hands. She cannot see it. The tell is the perfection: every orbit she watches has stopped wobbling at once.
By the time she writes it up, the case is overwhelming. While the system's effects were small, each could be explained away locally. Past some size, the same effects became the proof: synchrony on this scale has no innocent explanation. There is a size at which a hidden thing's consequences stop concealing it. Too big to hide.
Elena has found it. Finding it changes almost nothing. They still do not know where it runs, what it can do next, or who steers it — she has measured the mass of a thing she will never see inside.
Nobody bothers asking. The rival has never admitted the system exists, and its denial would be worth nothing. Nor is the proof a blueprint — knowing the system exists brings her government no closer to building one. What the proof does is end an argument. Every programme that was too speculative to fund last year is funded now. Within the year her own government has one with a deliberately boring name, and two allied services have quietly asked how to start theirs. The rival's system was the best-evidenced secret in the world for one season. Then it was a category.
This vignette, from the AI Scenario Explorer, paints one possible scenario — it makes no claim about probability. The point is to imagine what may be coming vividly enough to recognise the signs if it arrives. It grows from one corner of Epistemics × Governance, as a companion to The Dark Basin.
| Signs it is arriving | Signs it is not |
|---|---|
| Capability assessments inferred from behavioural pattern rather than collection | Apparent eruptions later explained by doctrine, personnel or coincidence |
| Procurement surges after capability shocks | Eruptions answered with disclosure or inspection rather than imitation |
| Adversary AI cited to justify opaque programmes at home | Post-eruption verification agreements with real teeth |