The seed graph maps extreme-precision radial velocity (EPRV) instrumentation: the techniques that let a spectrograph measure a star’s line-of-sight velocity to below a metre per second, the regime needed to detect small planets by the tiny reflex wobble they induce on their host star.
A high-resolution spectrograph disperses starlight and measures how far spectral lines shift. At the precision EPRV needs, the limiting factors are rarely the idea and almost always the systematics:
EPRV is a field where progress is gated by instrument systematics as much as by ideas, so the gap between interesting and reachable is unusually sharp. The techniques are tightly coupled — calibration touches stability touches data-reduction — which makes a graph the natural representation. And the frontier moves in identifiable steps: a better calibrator, a quieter enclosure, a sharper activity diagnostic. That is exactly the shape the tool is built to read.
The seed is intentionally small and legible: a couple of calibrators, a couple of stability methods, two data-reduction methods, and one diagnostic. It is enough to make the encoding and the scoring meaningful without pretending to be a survey. A real snapshot, built from the ingestion pipeline, is wider and changes over time; the seed is the fixed, readable starting point.
The node summaries are factual and deliberately plain. The scores on the seed
are driven by illustrative activity numbers, not measurements — see
docs/data-sources.md. Nothing here should be cited as a statement about the
real state of any of these techniques.