Independent reproduction will determine whether the Qwen 3.8-assisted reconstruction of the 1994 game Rats! recovers source code with practically useful correctness and completeness.
state: expiredheat: lowuncertainty: highknownscott: lowcoding-agents open-models software-reconstructionneuromancerQwen
What is this?
The case describes a repository attributed to “neuromancer” that reportedly used Qwen 3.8 to reconstruct the source code of the 1994 game Rats!, but none of the supplied search snippets documents that repository, its method, or its results. The snippets establish only that Qwen-family models and tooling are evaluated for coding and code correctness, with one Qwen3.8-Max page describing a separate long-running autonomous coding harness. Despite the web answer’s assertion, the supplied evidence does not establish any independent reproduction or practically useful correctness and completeness for the Rats! reconstruction.
Why it matters to Scott
Scott’s AI Legacy Takeover and Characterisation Testing pages already hold the load-bearing position that reconstructed software earns trust through independent, measurable behavioural verification rather than generated-source claims. With no supplied evidence establishing the repository, method, tests, or results, this is currently only another unverified example of a pattern already tracked in the Cursor SQLite and COBOL-to-Java reconstruction cases.
ip:framework.ai-legacy-takeoverip:concept.characterisation-testingip:concept.measurable-convergenceip:concept.evaluation-driven-developmentradar:cursor-sqlite-doc-reconstructionradar:cobol-java-agent-migration-validationradar:concept.verification
queries asked of Scott's wikis
- software reconstruction with coding agents
- independent verification of agent-generated code
- behavioral equivalence versus source-code fidelity
- coding-agent harnesses for legacy codebases
- open-model coding reliability and evaluation
- test oracles for reconstructed software
Measured heat
no measured readings yet — the hourly heat pass fills this in
How the heat travelled
no chain yet — the hourly chain pass fills this in
Evidence (2) — ⭐ canonical anchor
Interpretation history
2026-08-18T14:47:41Z
The artifact has attracted no independent reproduction, methodology, or behavioral validation within the observation horizon. It remains an unverified instance of an already tracked reconstruction pattern, with no reason to keep the episode active.
2026-08-16T14:32:10Z
No new validation, implementation detail, or independent reproduction has appeared; the case remains an unverified artifact rather than evidence of practically useful software reconstruction.
2026-08-16T14:27:39Z
grounded: known/low — Scott’s AI Legacy Takeover and Characterisation Testing pages already hold the load-bearing position that reconstructed software earns trust through independent
2026-08-16T14:24:48Z
case created — The public repository is a concrete, reproducible artifact testing an unusual coding-model capability on legacy software.
Decision trace
- 08-19 00:47expireThe artifact has attracted no independent reproduction, methodology, or behavioral validation within the observation horizon. It remains an unverified instance of an already tracked reconstruction pat
- 08-19 00:47alert_silentThe only change is negligible engagement movement without comments or new evidence; it does not alter the capability assessment or merit attention outside normal discovery.
- 08-19 00:47alert_routeThe only change is negligible engagement movement without comments or new evidence; it does not alter the capability assessment or merit attention outside normal discovery.
- 08-17 00:32repriceNo new validation, implementation detail, or independent reproduction has appeared; the case remains an unverified artifact rather than evidence of practically useful software reconstruction.
- 08-17 00:32alert_silentThe reobservation is unchanged and adds no consequential delta; wait for an independent reproduction, behavioral-equivalence results, or documented test methodology.
- 08-17 00:32alert_routeThe reobservation is unchanged and adds no consequential delta; wait for an independent reproduction, behavioral-equivalence results, or documented test methodology.
- 08-17 00:30alert_silentA repository appears to exist, but the supplied evidence provides no method, test results, behavioural-equivalence measurements, or independent reproduction. The claimed reconstruction therefore adds
- 08-17 00:30alert_routeA repository appears to exist, but the supplied evidence provides no method, test results, behavioural-equivalence measurements, or independent reproduction. The claimed reconstruction therefore adds
- 08-17 00:27groundScott’s AI Legacy Takeover and Characterisation Testing pages already hold the load-bearing position that reconstructed software earns trust through independent, measurable behavioural verification ra
- 08-17 00:24createThe public repository is a concrete, reproducible artifact testing an unusual coding-model capability on legacy software.