Talk notes
Every other talk in this summit references one of six atomic signals. If we don't define them precisely, the rest of the summit is folklore. So this talk is the dictionary. We define each signal as a formula, attach a decision rule, and name the most common false-positive pattern.
Commit velocity is a 14-day rolling sum of authored commits across all public repositories owned by the organization, normalized to a z-score within the sector to control for sector-level seasonality. Decision rule: a z-score above 2.0 sustained for two consecutive 14-day windows is a signal. The common false positive is monorepo migration — a single squash-merge can show as a velocity spike when it's actually a structural commit.
Contributor influx is the net-new commit authors observed in a 30-day window, sized as a ratio to the trailing-180-day baseline. Decision rule: a 2× ratio is a signal. The common false positive is hackathon participation — a contributor influx that's all single-commit authors usually means a public hackathon, not a hiring wave.
Repository expansion is the count of new public repositories created within an organization in a 90-day window, weighted by commit density on day 30. Decision rule: a 3× expansion versus the trailing 180-day average is a signal. The common false positive is fork-only expansion — companies sometimes fork a hundred OSS repos for vendoring, not for real product work. Star detachment, issue cadence, and dependency adoption follow analogous formulae detailed on /signals.