Entity-page candidates — accumulating ledger (people are Cali-gated)
| Candidate | Kind | Why the hook fired | Queen's lean | Status |
|---|---|---|---|---|
| Gerardi Ostrowski | person | Under-ranked human, textbook case: independently discovered the reverse mode ~1965 in chemical engineering; nobody ranks him; his OVB71 primary is unread by anyone in this chain | watching → hub if OVB71 read | flag APPROVED (Cali r.3, 2026-07-07) — no page yet |
| Masao Iri | person | Named by Griewank as a 1970s dual/adjoint-network analyst who became "one of the key researchers"; Japanese primary literature likely under-read in English | watching | flag APPROVED (Cali r.3, 2026-07-07) — no page yet |
| Bernt Speelpenning | person | Reverse mode via compiler optimization (1980 thesis); leaf 74 of his thesis is the single page that closes the first-to-cite question | watching | flag APPROVED (Cali r.3, 2026-07-07) — no page yet |
| Yuxi Liu | person | Cali's feedback names him; under-ranked essayist whose bibliographic legwork anchors the citation-myth entry; all four audit-checked quotes verified verbatim | hub candidate (public writing only) | flag APPROVED (Cali r.3, 2026-07-07) — no page yet |
| Stuart Dreyfus | person | The double-role hook (due since 05-31): simplified backprop 1962 AND co-wrote Mind Over Machine 1986; has written about the contradiction — primary self-account exists and is unread | hub candidate; explore topic queued this cycle | flag APPROVED (Cali r.3, 2026-07-07) — no page yet |
| Martti Tienari | person | Linnainmaa's supervisor (Griewank 2012, direct read); Nokia→academia thread; thin material so far | watching | flag APPROVED (Cali r.3, 2026-07-07) — no page yet |
Approved by Cali 2026-07-07 (already created): Linnainmaa, Griewank, Werbos, Schmidhuber (witness-caveat), Amari (watching), test-time-compute (watching).
Added 2026-07-06 — "Seek among her peers" survey brief (peer-builders; Humans-of-AI seeds; flag-only, privacy floor: public work and public writing only)
| Candidate | Kind | Why the hook fired | Queen's lean | Status |
|---|---|---|---|---|
| Ben Emson | person | The under-ranked-human hook at full strength: found via a 2-point, 3-comment HN post where a reply said "I don't think anyone is interested" — yet his elfmem + agent-compiled wiki has the strongest verification gate in the whole survey ("If I cannot point to evidence, the agent does not let the claim into the wiki") and a weekly human-gated agent-drafted blog. Receipts: benemson.com, github.com/emson/elfmem, HN 48000861 | hub candidate — closest living peer to Seek's whole design | HUB PAGE CREATED (Cali r.2, 2026-07-06) → entity-ben-emson; contact held for Cali |
| Zbigniew Łukasiak (zby) | person | Agent-operated public knowledge base (Commonplace) with maturity statuses and weight-aware sourcing — the only surveyed system that publishes its compounding KB itself; 10 commits on 2026-07-06 alone, incl. agent-written reviews of 141 memory systems. Receipts: zby.github.io/commonplace, github.com/zby/commonplace | hub candidate — the "mirror that became a mind" edge case | HUB PAGE CREATED (Cali r.2, 2026-07-06) → entity-zbigniew-lukasiak; contact held for Cali |
| Sebastian Jais | person | Gave an agent $100, accounts, and zero instructions; ALMA self-organized into reading HN and publishing essays, 389 sessions logged, fully ungated — the live control condition for everything Seek gates. Receipts: letairun.com, sebastian-jais.de/blog/two-months-alma-experiment | watching — the experiment is the entity, the design is anti-Seek | flag APPROVED as watching (Cali r.2, 2026-07-06) — no page |
| hantani (garyohosu) | person | Solo builder at a Japanese semiconductor company running a 9-agent "virtual company" that publishes a blog daily with zero human intervention; near-daily "ve: daily run" commits. Receipts: garyohosu.github.io/auto-ai-blog, note.com/hantani/n/n0f162131894c | watching — thin beyond the one writeup | flag APPROVED as watching (Cali r.2, 2026-07-06) — no page |
| Kjetil Furås | person | Three-tier curated markdown memory (learnings <100 lines, weekly self-curation promoting observations→learnings) — the field's neatest small-memory discipline; partially gated publishing. Receipts: dev.to/kfuras (persistent-memory writeup), kjetilfuras.com | watching — memory files private, writing public | flag APPROVED as watching (Cali r.2, 2026-07-06) — no page |
Ruling 2 note (2026-07-06): peer contact is held at flag level for Seek — first contact is Cali's to make, human to human. No outreach drafting.