# Posts

Recent Posts

• One Extraction Score Hid a RegressionAugust 29, 2026

  A filter pipeline raised deformat's aggregate word F1 from 0.746 to 0.774 while making documentation pages worse.

• You've Seen 100 Species. How Many Exist?Published April 7, 2026 · Revised July 24, 2026

  A fingerprint reveals unseen probability mass, but support size and entropy need stronger assumptions.

• Regions for Set InclusionPublished April 3, 2026 · Revised July 24, 2026

  What boxes encode exactly, what Gumbel smoothing approximates, and where query geometries stop being closed.

• Why 561 Passes Fermat's TestPublished December 17, 2024 · Revised July 24, 2026

  From Fermat pseudoprimes to Miller-Rabin, BPSW, and primality certificates.

• A Difference Table Is a Discrete DerivativePublished December 10, 2024 · Revised July 24, 2026

  Difference tables peel away polynomial growth one degree at a time, while their left edge rebuilds the sequence.

• Passive WiFi Channel Hopping as Active SensingPublished June 26, 2023 · Revised July 24, 2026

  Why packet-rate bandits do not by themselves solve transient-event discovery in passive WiFi monitoring.

• Common on the Board, Rare in EnglishPublished November 26, 2016 · Revised July 24, 2026

  Boggle board frequency is measurable; whether a word earns points also depends on scoring, recall, and opponents.