Driftproof

Receipts

Each receipt page shows the two arms, what was verified, and how to check the file yourself. They are listed here by report, newest report first.

Skill badges and README snippets · Published receipt list

Report 020

Report 019: a same-length placebo text

Read Report 019.

Report 018: a same-length placebo text

Read Report 018.

Report 016: Popular rules files vs. a placebo: most made Claude write more

Read Report 016.

Report 014: Claude Haiku 5.5 on release day, three skills

Read Report 014.

Report 013: Claude Sonnet 5.5, the day after its release

Read Report 013.

Report 011: Claude Opus 5.5, the day after its release, three skills

Read Report 011.

Report 009: three skills under two eval harnesses

Read Report 009.

Report 008: Claude Fable 5.1, no separation detected across the release in either arm

Read Report 008.

Report 007: the baseline is the noisy arm

Read Report 007.

Report 006: the skill text moves, the substrate holds still

Read Report 006.

Report 005: what does a skill cost to run?

Read Report 005.

Report 004: does encoded expertise still lift output on the frontier tier?

Read Report 004.

Report 003: do skill verdicts hold across a model release?

Read Report 003.

Report 002: does a skill's benefit hold across substrates?

Read Report 002.

Report 001: do skill verdicts hold across a model release?

Read Report 001.

Receipts outside a report

Words with a dotted line link to the glossary, where each is explained in plain words.