I've spent years accumulating a fairly large emulation collection. My current LaunchBox catalogue has just under 33,000 entries.
Recently I started trying to answer what sounded like a simple question:
Does the collection LaunchBox says I have actually match the game files sitting on disk?
It turned out to be a much more interesting question than I expected.
I found cases where LaunchBox showed two different games, but both entries were backed by the exact same game payload.
In one case I appeared to have both Battle OutRun and OutRun Europa. Both archives actually contained the same Battle OutRun ROM.
In another, four differently named PlayStation CHDs appeared to represent different Beat Mania discs/variants. Reconstructing the disc payloads showed that all four were actually the same disc.
But I also found the opposite problem: identical content that shouldn't simply be treated as a duplicate and deleted. Cat Trax / Cat & Mouse, for example, turned out to be a legitimate alternate identity/re-release relationship.
That rabbit hole became Retro Auditor.
The idea isn't to build another duplicate remover or ROM downloader. Retro Auditor tries to reconcile several different claims about a collection:
what the frontend says the game is;
what physical file is actually present;
whether different files contain identical payloads;
structural and embedded evidence from formats such as archives and CHDs;
and, where available, external preservation/reference evidence.
The important distinction I've ended up making is:
Payload certainty and identity certainty are two different things.
Two files can be proven byte-for-byte identical without that automatically meaning one should be deleted. Likewise, having two different titles in a frontend doesn't prove that two different games are physically present.
On my collection the current audit produced 39 findings worth examining:
4 Action Required
16 Review
17 Cleanup
2 Information
It also found files that initially looked like possible game content but were actually RPCS3 internal shader data. Those were excluded rather than being reported as bogus games.
Retro Auditor never automatically deletes anything.
I'm looking for a couple of real-world collections
I've reached the point where testing this against my own collection isn't useful enough anymore.
I'd like to try it against 2–3 other established LaunchBox collections—particularly collections that have been imported, moved, renamed, merged, upgraded or accumulated over a long period.
For this early test I'm keeping the analysis engine private.
The collection owner runs a small read-only evidence probe locally. It records the catalogue information and cryptographic/file evidence required for analysis and produces a ZIP that the owner can inspect before choosing to send it to me.
The probe does not:
upload anything automatically;
send ROM/game payloads;
modify, rename, move or delete collection files;
include absolute local paths in the shareable package.
I analyse that evidence here and return the resulting collection-integrity report.
There is no charge for these first tests. What I want in return is candid feedback: did it discover anything you didn't know, was the information actually useful, and is this a problem worth solving as a finished tool?
If you've got a substantial LaunchBox collection and have ever suspected there might be some strange archaeology buried in it, reply here or send me a message.
I'm particularly interested in collections that have been through years of changes rather than freshly curated sets.
