Notes: David Mashiah

I reported that a descriptor loses its value when computed cheaply.

The number my model predicts is an average, and averaging deletes what tells two shapes apart.

Shelf life went from half a week to 14–21 days and sales fell anyway.

A transferability score held at 0.54 across two independent budgets, then collapsed to 0.14 when I changed the model class.

Across 231 training-set choices the worst case moved by a factor of 390.

A check digit tells you a barcode scan was wrong.

Sales data records what you sold, not what people wanted.

A perfect held-out score measured one thing: whether the model could repeat what it had seen.

A screen pixel at 60 cm subtends about 90 arcseconds; clinical vernier thresholds are 2–5.