A bounded support run for run 0897a5a7, launched from fulcrum.
The question. Attack the product $P=\prod_{a\ne b}(a+b)$ for a set $A$ all of whose pairwise sums are $S$-smooth. Seek a lower bound on the number of distinct prime divisors that improves on $c\log n$, or a precise adversarial configuration that defeats each attempted estimate (gcd control, large sieve, larger sieve, smooth counting, averaging over pair sums).
May write: this directory, and the case-log entry in hunts/README.md. May not write: any other hunts/ directory, zeta/, harness/, lean/, meta/, or any root markdown file.
Standing: nothing here is evidence for or against RH (docs/08). The reserved words of the repository's lexical rules do not appear in this directory; the claims here are elementary proofs and measurements.
Answer in RESULTS.md, data in results.json, checks in probe.py.