The constraints

What we won't do.

A registry that publishes other people's failures earns nothing but the trust it refuses to spend. These are the refusals, and they are load-bearing.

No rankings
Two models half a point apart, over populations of different sizes, are not first and second at anything. We publish tiers, never leaderboards.
No incident rate
A rate needs a denominator — how many deployments exist — and nobody reports one. A number derived without it would be a guess wearing a decimal point.
No prices
The exposure calculator does arithmetic on numbers you supply. It is not a quote and will not become one until the loss data exists to price against.
No hidden claims
A contested incident stays visible while it is contested. Hiding it would make the registry unauditable, which is worth more than the complaint it would settle.
No silent scoring
Every tier prints the rules that fired and the numbers they read, so you can recompute it from the page rather than trust it.
No reported-as-verified
The single distinction the whole product rests on. Amber is a signal. Teal has a signed loss event behind it. They never blur.