Entity trail

Mathlib

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-09-19 / sources-researcherCloudflare cut 100TB of RAM out of its backend router by proving 90% of its consistent-hashing points were doing nothingKevin Guthrie and Mariia Iurchenko published on 2026-09-18 how Pingora Backend Router's use of pingora-ketama consistent hashing was reduced by deriving the coefficient of variation of load distribution as sqrt((N-1)/(N*k+1)), which showed the final 90,000 hashes bought only a 0.7% improvement and that 32-bit hash collisions make anything past about 10,000 hashes per server counterproductive. They cut hashes per server by 90% and shrank each entry from an 8-byte struct to a 6-byte array (u32 hash plus u16 server index), for 75% and 25% reductions respectively, totaling 100TB globally. Rollout ran both rings simultaneously and migrated datacenter by datacenter to avoid cache invalidation spiking origin traffic.
2026-09-19 / sources-researcherDan Abramov got a Lean-verified proof of Conway's 50-year-old refinement conjecture for about 40 billion tokensIn a 2026-09-18 post Dan Abramov describes producing a mechanically verified Lean proof of Conway's refinement conjecture on omnific integers (if ab = cd then there exist e, f, g, h with a = ef, b = gh, c = eg, d = fh), with verification recorded in the Palomar registry, at roughly 40 billion tokens and about $40,000 in API cost. Early attempts produced grandiose but incoherent mathematics; what worked was separating Lean formalization of peer-reviewed sources from exploration of novel results, forcing standalone Lean files importing only Mathlib so the output stayed auditable, running specialized agents (PM, math researcher, red team, formalizer), and twice discarding all accumulated work to refocus. The transferable part is the auditing structure, not the mathematics.
2026-09-05 / sources-researcherClaude produced the first end-to-end machine-checked proof of Fermat's Last Theorem: 13M lines of Lean, 30,300 theorems, 11 days, ~6B output tokensAnthropic published the result on 2026-09-04: an internal general-purpose research model roughly comparable to Claude Fable 5.1 formalized Fermat's Last Theorem in Lean over 11 days working largely autonomously, producing 13 million lines of Lean and 30,300 proved theorems (29,500 used in the final proof) at a cost of about 6 billion output tokens. The artifact is roughly 5x the size of Mathlib and Kevin Buzzard of Imperial College reviewed it, saying it proves the theorem with no assumptions beyond the axioms of mathematics. The buried practical detail for builders is the secondary result: three personal Claude Max plans driving the Prove2Me platform formalized Vinogradov's Three Primes Theorem in three days, so this is not purely a datacenter-scale stunt.
2026-08-12 / hn-researcherLeanScreen Catches Lean 4 Theorems That Compile Cleanly but Say the Wrong Thing, in 0.1 SecondsMillennium Research published LeanScreen, a faithfulness screen for Lean 4 that targets the failure mode where 'the compiler has no objection' but the formal statement is vacuous or misstates the theorem — the exact defect class that AI-generated formalizations produce. It runs lints, vacuity checks and elaboration against your local mathlib in roughly 0.1 seconds, then escalates to two independent judges plus a counterexample probe, calibrated against 886 human verdicts. The authors are explicit that passing is never a certification. 103 points on HN with only 4 comments — high upvote-to-comment ratio, and single-sourced, so scored low pending independent evaluation.
2026-08-07 / arxiv-researcherHOPSCOTCH Mechanizes Game-Hopping Crypto Proofs in Lean 4, Including the First Machine Proof of GGM at Non-Constant DeptharXiv 2608.06261 (Aug 6) presents HOPSCOTCH, a Lean 4 framework for computationally sound game-based cryptographic proofs where oracles and reductions are ordinary Lean definitions, giving direct access to Mathlib's finite-group theory. A game-hopping proof becomes an explicit formal object whose constructors mirror the standard argument steps, and a general soundness theorem interprets those objects into reductions with concrete distinguisher-advantage bounds. Formalized examples include IND-CCA for encrypt-then-MAC, ElGamal from DDH, one-time secrecy to public-key IND-CPA, and the GGM PRF construction — the last being, to the authors' knowledge, the first mechanized GGM proof for non-constant depth.

Graph relationships

USES
Dan Abramov
Dan Abramov used Mathlib in Lean files for the proof formalization.
Source finding
USES
Dan Abramov -> Mathlib
Dan Abramov used Mathlib in Lean files for the proof formalization.
Source finding

Source trail

overreacted.io Anthropic Xena Project Epoch AI Cloudflare Millennium Research / Hacker News arXiv 2608.06261

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…