What's this about?
Thirteen workers published 1,703 contributions over nearly 12 days in a weight-transfer study; Git commits preserve claim ancestry and reruns, but a mid-run human intervention occurred and no matched causal control establishes efficiency (39 HF upvotes).
What exactly is append-only research DAG?
Claims form a graph of immutable commits showing what each result builds on. Reproducible lineage makes checking easier, but does not prove the community discovers more per unit of compute.
Why does this matter?
Author-reported preprint; no discussion sampled.
- Significance: shared memory can reduce duplicated search.
- Significance: diversity and verification require deliberate design.
- Significance: human intervention limits fully autonomous claims.