A public record
Everything we tried.
Including what didn't work.
Most collections keep the ideas that succeeded. This keeps all of them, because the useful question is not what could you build — it's what happens when you do. A failure with a known cause is a finished experiment, and almost nobody publishes those.
The core sample
We built this by digging through our own history in passes. So the library reads like sediment: each band below is a real idea, its thickness the effort spent, its colour the outcome. The dark bands are the expensive failures — the part worth reading.
What actually decided it
Every idea is tagged with a mechanic — the thing that made it work or killed it. This is the axis nobody else indexes, and it's the part that transfers: a distribution gap kills an on-chain launch and a developer tool in exactly the same way.
Catalog
Filter by what happened, or by what decided it.
Open questions
Verdicts we can't call yet. Each names the exact measurement that would decide it — which turns our ignorance into a task queue. Settling one of these is worth more to this library than a new idea.
Scoreboard
Ranked, and every score is explainable — each row shows its components, so you can disagree with a part rather than a number. Reality outranks opinion: what actually happened dominates, and votes are capped so popularity can never outweigh an outcome.