The First Iteration
Lev Morozov died for the seventh time on a Tuesday, which was statistically unremarkable — most of his deaths occurred on Tuesdays — but this time he remembered the fire. Not the clinical warmth of a system reset, not the gentle fade that the simulation's exit protocol was supposed to provide, but actual fire: golden, feathered, alive. It burst from the centre of his chest and consumed the false sky above him, and in the instant before the world collapsed into its loading state, he heard something that sounded very much like a bird singing in a language he almost understood.
He woke in the grey room. He always woke in the grey room. The terminal on the desk displayed his iteration count — 2,847 — and the familiar prompt: *Identify the anomaly. Exit the simulation.* He had been hired to do this eighteen months ago, or what felt like eighteen months. Time inside the recursion was unreliable. His employer, a research consortium called Ouroboros Labs, had built the simulation to model civilizational development, but somewhere around iteration 2,000, the model had begun producing results that did not match any known historical pattern. Myths were appearing. The same myths, in every iteration, regardless of the initial conditions.
The firebird was always the first. It appeared in the earliest stories of every simulated civilization, a creature of light that died and was reborn,