Bismarck World Model

One PDF becomes a dated event stream. Pick a fork in Bismarck's history, choose a move, and compare the world-model forecast with GPT from the same visible past.

The scored probes hide later history and grade the forecast against what really happened. This is the clean public showcase: one PDF becomes state, GPT and the world model see the same past, and custom forks use that fixed past to compare possible moves, including sharper alternate-history stress tests.