The milestone — and what we don't cover yet
What has been done
Public debates, with discussions and conclusions. A game to learn by playing. A training room: a terminal to build with. The whole site translated automatically. Guardrails on every public output, tested again and again. And a module that tracks theories: we state them as theories, then we validate or refute them with evidence.
What we cover
Work, measured, never just posted. The health of what we deliver — pages, content, security — tested continuously. And honesty: "written ≠ posted". We check what the browser receives, not what we launched.
What we don't cover yet — and we say it
The intelligence's performance against people, on standard benchmarks. Calibration: does it know what it knows? Robustness, against someone who would want to manipulate it. Costs per skill. And above all, the impact: does the game appeal? Does the offer sell? All of that will come. But today, these are theories, not facts.
We don't promise what we don't cover. That sentence is worth all the badges.