Yawn experience
Say what a Leaderboard means: it shows what competes for a Yawn's attention and why, instead of telling you what matters.
Stated by Dave · 2026-09-27
Show, don't tell. It never tells you what matters; it shows what is competing for a Yawn's attention, in order, with the reason for every place.
Each place carries its basis: Dave's authored order, the frozen collective cognition, a live request, or a recorded meaning.
A Yawn with no basis is not ranked. It is listed as having no meaning yet, with a request to supply one.
Example · proposed by Claude
An emergency-department triage board. Everyone on it competes for the same attention, and the board shows who is next with the recorded reason: severity first, then time waiting. Nobody has to be told to trust the order.
Many emergency departments assign one of five triage levels from recorded signs and the resources a patient is expected to need, not from who argues hardest.
What carries over: the order is computed from what was observed, and the observation sits on the board beside the name.
Where the example stops: a triage level is a clinical judgment with a validated protocol behind it. A Yawn's weights are attributed salience with no such validation, and they say so.
Implemented
Every place is computed from recorded inputs you can open: Dave's stated order first, then the frozen collective cognition, then live requests, then recorded meanings. Nothing is scored that was not recorded, and what was left out is listed.
Collective cognition is a snapshot, not a feed: it says when it was frozen, who took part, and what it did not count. The pulse reports its age at yawn.bot/status and calls it stale after seven days.
Weights are core, high or supporting: attributed salience, never a probability or a measurement.
A board computed from what everyone can see invites gaming it; data-science competitions keep a hidden test set for that reason. Freezing and dating the snapshot is the same guard at a smaller scale: the order can be argued with because it holds still.
Each face shows a state in the mascot model's words: Focused while a live request is in flight, Curious while the collective's question is on it, Resting when only a meaning is recorded.
Connections · proposed by Claude
Relationship first: a Leaderboard is the relationship between a Yawn and what is inside it, made inspectable. Its pull is gravity in the house sense: attributed salience and action pull, never truth, command or permission.
Observation is operationally first: nothing is ranked before it is observed.
Only evidence upgrades a claim. A higher place is not evidence.
Question first: the board opens with the questions the Yawn is asking, and each contender says which one it serves.
Stack ranking and the coverage gauge under Relationships are the same idea at other scales.
