← Analyses

Lineage forecast

For each of the 8 detected lineages, project where the next entry is likely to land based on the cadence of past releases. Exploratory — watch list, not prediction.

Read this first. Most lineages on this page have only 3–5 dated members. A cadence computed from 2 or 3 inter-arrival gaps is informative directionally (is this family active or stalled?) but is not a statistical prediction. The "next expected" quarter is a watch trigger — a date to check the source feeds — not a forecast to defend. If a family dries up, gets absorbed, or accelerates by a single release, the projection moves by a year.
How the cadence is computed
  • For each lineage we list its members in descending order by created date, parsed to quarter granularity (year × 4 + quarter).
  • The cadence is the mean of the inter-arrival gaps in quarters, with a floor of 0.5 quarters per gap so that same-quarter clusters don't crash the average to zero.
  • The prediction interval is the population standard deviation of those gaps. We surface it as "Q3 2026 ± 2 quarters" — read this as "if the cadence holds, the next entry will probably land within this window."
  • Discount the projection when (a) the lineage has fewer than 5 members (small-sample badge), (b) the cadence is built from fewer than 3 gaps (insufficient-data badge — we hide the date entirely), or (c) the gap spread is large relative to the mean.
So what? A short-cadence lineage means the architectural idea is still earning new contributions — reading the next 2–3 arXiv updates in that lineage is high-leverage. A long-cadence lineage means the idea is either mature (settled, low marginal returns) or stalled (researchers moved on). Adjacent lineages are where cross-pollination shows up first: work in one family routinely informs predictions for its neighbours.

8 of 8 lineages shown · 0 small-sample (N<5) · 0 insufficient data (cadence intervals <3)

Kind
Provenance
Min members
N ≥ 3