Methodology
whosrc ranks proof artifacts — the merged PR, the accepted answer, the RFC debate. The author is the ranking signal, never the result. A result is only as good as the provenance behind it.
What we rank on
- Topic match between your query and artifact title/topics/excerpt
- Proof quality, computed per artifact (merged into others'repos > accepted tag-specific answers > accepted RFCs > sustained commits > posts)
- Specificity (rare technical tokens beat generic "software")
- Recency (domain half-life)
- Author track record on this topic — repeated proofs boost an artifact; they never create a leaderboard
Every result shows whyit ranks in one line. If we can't explain the rank, we don't show the result.
What we never rank on
- Follower or connection count
- Job title or company name
- Self-written skill tags without artifacts
- Endorsements / "Top Voice" theater
- Raw stars on the author's own repos (weak prior only)
- Paid placement
Snippets, not mirrors
We store metadata rows and bounded excerpts (a few lines of a diff, the first sentences of an answer) — always attributed, always deep-linked. The full content lives at its source. Stack Overflow excerpts are CC BY-SA and credited; GitHub data comes via the official API under their ToS.
Data sources (v0)
This demo ships with a curated Rust-wedge seed so search works offline. Production ingest uses official GitHub and Stack Exchange APIs — not LinkedIn scrape, not cold email.
Authors
Author pages are artifact indexes built from public signals and are display only. We do not message people who have not claimed and opted in. No leaderboards, no "top experts" pages — a person emerges from authorship or not at all.