Stage 20: checker-family crossing harness on banked compositions

↗ view on GitHub · Eli Ziff · 2026-07-31 · fc820b65

The program's founding premise -- "the harness carries grounding
decisions, not the model family" (H5) -- rests on 8 cells over one item
pair, with a single false accept as the decisive datum. A replication was
registered twice (experiment log ~1299-1302 and ~1573) and never run.

This runs it on BANKED compositions: no composer call, only the checker
varies. Three conditions per sampled row -- the archived verdict, a
same-model re-check (the run-to-run stochasticity floor at THIS n), and
the other family's anchor model.

- legalGroundingBenchmarks.ts: the CSLB/CLERC/HousingQA case builders
  lifted out of the experiment script unchanged. Run receipts store the
  answer and evidence but not the question, so a re-check has to rebuild
  the item; a second copy of the builders would drift from the runs it
  re-checks.
- holisticVerificationPass exported so the crossing reuses the exact
  system prompt, tool and payload the original runs used.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01H9ToHYJVDxfeJcJwdzrP2H
Repository eliziff/Beaver
Author Eli Ziff <eliasziff@gmail.com>
Authored
Parents b80f05ec
Stats 4 files changed , +881 , -259
Part of Legal grounding and retrieval research

Capture this commit into my fork

Download a Markdown prompt that tells Claude how to port this exact commit into your working tree. Run it via claude -p < capture-commit-fc820b65.md from inside the repo you want the change in.

⬇ Download capture-commit-fc820b65.md