rseng-discoverylisted

# Discovery: publications and software around a topic Most research work begins in a landscape someone else has already mapped in part. Discovery is the deliberate survey of that landscape before building, writing or claiming novelty: the publications that define the state of the field, and the software that already solves pieces of the problem. Doing it early is cheaper than discovering a competitor in review, and the survey itself becomes reusable material for the paper's related-work section and the proposal's state of the art. ## Finding publications Work from open, keyless APIs first; they make searches reproducible and scriptable: - OpenAlex covers most of the scholarly record with abstracts, citation links and topic filters; Crossref serves DOI metadata; Semantic Scholar adds influence signals and TLDRs; arXiv covers preprints in its fields. Zenodo and JOSS matter specifically for research software: JOSS papers ARE software descriptions, and Zenodo records often carry the software itself. - Google Scholar casts the widest net (theses, reports, gray literature) but has no API and rate-limits scraping; use it interactively for coverage checks, not in pipelines. - Snowball in both directions from the two or three most relevant hits: who do they cite (backward), who cites them (forward - OpenAlex and Semantic Scholar both serve citing works). One good seed paper beats twenty keyword guesses. - Search by concept variants, not one phrasing: field syno