Conversation
…t write 340: `max_vacuous` exists to stop vacuous completeness theorems growing. There are 114 in 250 models; it counts 44. Not because its marks are wrong -- all 44 are correct -- but because it reads the MISMATCH ledger, and a model reaches that ledger only if it also disagrees with the Rust classifier. Being vacuous has nothing to do with the classifier disagreeing; the two were joined because one file held both answers. Worse than no ratchet: with none the question is open, with this one it has an answer that is wrong by 70. The check for any ratchet is what population it walks, and whether that is the population its name says. 341: last pass this document recommended writing real Lean models for four of them as "the only change that lowers this number honestly", and this pass did not. `lean`, `lake` and `elan` are not installed here. And a FAITHFUL model may make its theorem false -- `t27c gen-rust` on those specs exits 1 -- which would be the correct outcome and a real finding. So a red `lake build` would mean either "the theorem was false all along" or "you transcribed it wrong", and with no local build there is no way to choose. Pushing an artefact you cannot check to see what CI says is guessing with a longer feedback loop, and here the artefact is a proof. The honest substitute is the count, plus saying which of the two you did. Refs #2747
gHashTag
enabled auto-merge (squash)
August 29, 2026 21:39
Contributor
|
📓 NotebookLM Notebook linked to this PR
This notebook contains session context, decisions, and artifacts for this work. |
Contributor
PR DashboardGenerated at: 2026-08-29 21:39:29 UTC
Summary
Seal Status
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Two sections from #2890.
340 — a ratchet measuring a subset of its own subject.
max_vacuousexists to stop vacuous completeness theorems growing. There are 114 in 250 models; it counts 44.Not because its marks are wrong — all 44 are correct, and no entry is marked
model_emptythat is not. It counts 44 because it reads the mismatch ledger, and a model reaches that ledger only if it also disagrees with the Rust classifier. Being vacuous has nothing to do with whether the classifier happens to disagree; the two questions were joined because one file happened to hold both answers.This is worse than having no ratchet. With none, "how many vacuous theorems are there" is an open question. With this one it has an answer, the answer is 44, and it is wrong by 70. The check to run on any ratchet: what population does it walk, and is that the population its name describes?
341 — I did not write the proof I said I would, and the reason is the point. Last pass this document recommended writing real Lean models for four of them as "the only change that lowers this number honestly".
lean,lakeandelanare not installed here — the instrument exists in CI, not where I am. And a faithful model may make its theorem false:t27c gen-ruston those four specs exits 1 and emits nothing, so an honest transcription might be unprovable, which would be the correct outcome and a real finding rather than a bug.Which means I could not tell the two apart. A red
lake buildwould mean either the theorem was false all along or you transcribed it wrong, and with no local build there is no way to choose.Pushing an artefact you cannot check to see what CI says is not delegation — it is guessing with a longer feedback loop, and here the artefact is a proof. The honest substitute is the count, and saying plainly which of the two you did.
Refs #2747