Repository navigation
rosetta[go]: cross-language consistency — 3 red / 0 amber metrics, 5 ledgered shapes #2575
Description
Part of #2560 — per-language tracking for the keyword-rosetta control corpus (built for #1096). Planted intent is identical in every language, so every deviation below is measured bias vs. the 46-language median, validated in the corpus's deviation ledger. 🔴 = beyond ±50% of the cross-language median, 🟡 = ±25–50%.
Triage order (per the epic): structure detection first, then signal/keyword rules, then risk-score consequences — §3's numbers are downstream of §1–§2 and re-baseline as those fix. Cross-cutting engine issues (e.g. #2535/#2546) fix many languages at once; this issue tracks this language's residual after each cross-cutting fix lands, and closes when the language sits inside the ±25% band on every comparable metric (or every remaining deviation is ledgered as intended morphology).
1. Structure detection (fix first)
In band — no structure-level deviations.
2. Signal / keyword deviations
- 🔴
safety_bypasses: 5 vs median 2 (+150%) - 🔴
state_mutation: 11 vs median 6 (+83%)
3. Risk-exposure consequences (downstream)
- 🔴
risk_api_exposure: 1.89168 vs median 5.31104 (-64%)
Ledgered shapes touching go
| defect | type | issue | evidence | summary |
|---|---|---|---|---|
census-requires-git-tracked |
engine-semantic | — | — | GalaxyScope's census enumerates git-tracked files only; an untracked folder scans as '0 files mapped' with no per-file warning |
state-flux-branch-weighting |
engine-semantic | #2546 | b.go | state_mutation is recorded x3 when branch context exists in the file ('flux' semantics, matching the state_flux column name) |
visibility-morphology-encapsulation |
engine-semantic | — | — | Languages whose visibility IS a naming/declaration convention count it wholesale: go's encapsulation rule counts every lowercase top-level symbol (7-9 per corpus file), … |
go-import-counts-bypass |
upstream-question | #2542 | (notes) | go's safety_bypasses rule's dot-import alternation (\bimport\s+(?:.[ \t]+)?") matches EVERY quoted import, not just dot-imports: +1 bypasses per importing file corpus-w… |
string-literal-selective-shielding |
upstream-question | #2535 | b.go | Keywords inside one string literal ('if eval fails, try open') count toward branch (+2), safety (+1), io (+1) but eval does NOT count toward high_risk_execution (routes … |
Decoy outcomes (file-level evidence)
- b.go ·
note := "if eval fails, try open"(string): if counts branch +1; the := is a raw state hit. state 11 = probe_state's 3 raw x3 (flux, phantom branch in that function) + probe_bypass's 2 '_ =' unweighted — the flux weighting is PER-FUNCTION, not per-file. Ledger: string-literal-selective-shielding, state-flux-branch-weighting - c.go ·
(no code reverse decoy — go file carries comment decoys only)(comment): comment HACK/TODO count exactly 1 each
Reproduce / evidence
- Corpus folder:
data/go/—expected_signals.jsoncarries the shape-by-shape accounting - Gate:
GALAXYSCOPE_BIN=<venv>/bin/galaxyscope python tools/verify_language.py goin the corpus repo - Full context: findings_by_language.md#go · bias_report.md · bias chart
Activity
squid-protocol commentedon Sep 3, 2026
Status: 1🔴 / 1🟡 unexplained, 5 further out-of-band cell(s) accounted for
Generated from
docs/bias_data.json+deviation_ledger.jsonbytools/issue_status.py, using the same verdict machinerybias_report.py --gateruns on (gitgalaxy#2669 E.1), so this comment and the gate cannot disagree about what is left.Still unexplained — the work this issue tracks
risk_api_exposurerisk_tech_debtOut of band, but accounted for
ledgered — a validated ledger entry names this language and this metric
cog_raw— composite-nonlinearity-amplifies-in-band-inputsrisk_cognitive_load— path-and-extension-modifiersrisk_documentation— path-and-extension-modifiersrisk_state_flux— path-and-extension-modifiersstate_mutation— string-literal-selective-shielding, state-flux-branch-weightingClose criterion (gitgalaxy#2669 Batch E.4)
python tools/language_deviations.py goexits 1: 2 unexplained cell(s) remain. Each needs a fix, or a validated ledger entry saying why it is correct, before this issue can close.squid-protocol commentedon Sep 6, 2026
Closed 2026-09-06 as superseded by #2812 (
docs/contract_roadmap.md, decision D2). Every out-of-band cell for go is accounted for (the gate reads 0 unexplained) and is now tracked by cause family, not by language: the ledger entry that explains it, its cause category in keyword-rosetta'sdocs/bias_data.json["cell_categories"](echo / inherency / extraction / correlation / scoring), and the family issue #2812 lists for that cause. Live standing at any time:python tools/language_deviations.py goin keyword-rosetta, ordocs/findings_by_language.md#go.This is a change of axis, not a claim that go is green. The §10 capstone in
docs/language_status/go.mdis written when the language reads clean bylanguage_deviations.py, by whichever family PR gets it there — no longer tied to this issue.