Good morning.

Comparing like with like — each claim type against its own passed-over group — 27% of what The Trench Desk surfaced played out, versus 20% of what it passed over — 6 points better, across 17,686 claims.

Adding the rows above together instead makes it look 8 points worse, but that sum is misleading: The Trench Desk deliberately ranks earlier-stage narratives above later-stage ones, so the two groups are not made up of the same kinds of claim. Later-stage narratives migrate more often whoever picks them, which flatters the passed-over column. The headline above compares each type only against its own passed-over group, which removes that effect. 555 surging claims are left out of this comparison entirely, because The Trench Desk has never passed one over — there is no group to compare them against.

Every claim type above is retired. On 2026-08-12 the narrative rows were removed from the queue: measured across 156,812 launches, a token launched under a ticker ten other creators had just piled into completed its bonding curve 0.46% of the time, against 2.35% for a ticker nobody else had touched. This table is the record of a capability that no longer runs — kept visible because deleting the history of something we got wrong is the opposite of a track record.

“Passed over” are narratives that met the same evidence bar but ranked below the queue. They are the comparison that makes the first column meaningful. A rate is only shown once 20 claims of that type have resolved. 4567 claims could not be judged either way because The Trench Desk's feed was interrupted during their window — excluded rather than counted as misses.