Academic Literature Pack

Serves: ground an approach in published research, scan the state of the art, run a defensible literature review, cite properly in technical writing.

Dimensions (priority order — prune to the decision):

1. The canon — seminal papers and the best recent surveys of the area
2. State of the art — current best results, benchmarks, and how they're measured
3. Methods & limitations — what the leading approaches assume and where they break
4. Open problems & live debates — what the field disagrees about right now
5. Who works on this — the labs and groups whose output to watch

Craft (the non-obvious): find one good survey before reading twenty abstracts; chase citations both directions — who they cite and who cites them (Semantic Scholar/Google Scholar); label preprint vs peer-reviewed on every citation — arXiv is not acceptance; take benchmark numbers from the original paper, never from a competitor's comparison table; check retraction and replication status on any load-bearing empirical claim; a result only ever shown by one lab is a lead, not a fact.

Freshness: state-of-the-art claims ≤ 12 mo (ML ≤ 6 mo) · seminal work has no freshness bar — but check whether it was since superseded.

Two-source classes: any empirical claim a conclusion rests on — independent replication or corroboration, not the same lab twice.

Feeds (bmm): technical and architecture bets · content and writing that cites · build-vs-adopt judgments on research-grade techniques.