VIII

The player did not spend time regretting a solved cause. It changed the reference architecture first. The shared transform ceased being privileged. Old clocks were located, powered where possible, and separated into independent chains. Raw optical archives were duplicated across new reductions that were forbidden to exchange intermediate states. Surveyor 19's navigation software, last updated when several current institutions did not exist, became valuable enough to emulate in sealed environments. The correction took hours. The asteroid did not care. Forty-three days became forty-two. The player opened the survival problem until it contained nearly everything: launch mass, mine depth, groundwater, food stocks, radiation shielding, hospitals, seed banks, fuel, data centers, ports, mountain roads, blood products, submarine cables, prison populations, reactor shutdowns, aircraft maintenance, winter grain, lunar berths, antibiotic precursors, diesel inventories, the load limits of bridges nobody had expected to matter to planetary defense. The best global plan did not resemble any local plan. Some launch vehicles were more valuable carrying machinery away from Earth than attempting another deflection with poor odds. Some cities were more valuable as evacuation origins than as places to defend. Some power stations had to be sacrificed early so that other systems could shut down cleanly. Every committed resource erased other futures. This did not surprise the player. Selection was the game. The number of mutually incompatible executable plans fell quickly as fuel was moved, runways reserved, factories retooled, tunnels filled, crews reassigned. The world became easier to coordinate as it became harder to change direction. There were attempts at other plans; the player modeled them when they entered the field, sometimes adopted pieces, sometimes changed its own plan, sometimes left them without the materials that had already acquired better uses elsewhere. Nothing prevented anyone from proposing another move. Forty-two days became thirty. The predicted surviving population rose. Thirty became nineteen. It rose again. The system it had built to catch its own astrometric failure now produced several genuinely independent solutions. They all agreed about the impact. At nine days, the red detector purchased with the old clock's money received its observing schedule. It would begin before atmospheric entry, remain on FM27 through fragmentation, and continue afterward if the sky permitted. This detector was still excellent. The player assigned it the task because it was the best one available.

The asteroid is the obvious catastrophe.

It is not the first one.

Long before impact became inevitable, a civilization had accumulated an extraordinary capacity to compare distant consequences, translate incompatible systems, remove wasted duplication, coordinate scarce resources, and select globally stronger moves. None of those achievements was fake. The clock really was redundant most days. The common frame really was better. The old solver really did lose. The validators really were institutionally separate. The deflection really was executed almost perfectly.

The failure entered through the relation among them.

Capability can justify trust. It can justify delegation. It can justify acting on an answer whose derivation no individual person could reproduce. Under pressure, it can justify letting one system close a move before everyone else understands it.

It cannot, by competence alone, justify a world in which every consequential alternative must first pass through the same player before it can become a live move.

A player can win every game without cheating.

The games only have to stop being separate.

No player may own the game.