I think the whole idea of "clean-room reverse engineering" is essentially predicated on the idea that copyright would indeed apply to a scenario where you take a copyrighted work and incrementally replace all the individual pieces
Back in the IBM BIOS days, they were writing machine code. Basically a series of operation codes (ie, numbers).
Two people trying to implement the same complex system were almost guaranteed to produce the same code for several components. A clean-room procedure offers a verifiable defense that the duplication was not a result of copying.
Today we use much higher level languages so it’s trivial to write the same code a hundred different ways. I’m not sure a clean room is worth the effort other than to force the engineer to not be lazy and copy.
All that said, there’s also copyright rules covering derivative works. I’m sure there’s plenty of precedent in literature to cover “copying without actually copying” but I wouldn’t know how that works.
That's funny - I think the opposite! Clean-room reverse engineering would give you the same indivisible line of code (background-color: #xyz;) so it can't be required... or can it?
I'm thinking of stuff like the re-implementation of the IBM bios where the re-implementation had to effectively match a certain undocumented specification but not necessarily follow the actual implementation from the original bios
but in some sense clean-room reverse engineering does necessarily entail that certain things will be identical, and I guess the scenario in the article could be describing a situation where none of the final product matches the original template at all, so maybe it is slightly different, and in that case I guess there wouldn't be a copyright issue at all?
It would be more like how when movies are made they tend to use existing music as a placeholder until the final music is made, and it doesn't seem like anyone considers the final movie to require a license from the creator of the placeholder song even though the direction of the final work is often strongly influenced by its pacing
I've actually wondered why projects like WINE and ReactOS didn't just start with Windows and replace one DLL at a time until there was no Windows code left. Then again, it seems weird that we hyper focus on the value of the source code and not all the engineering effort that went into the architecture. As in I can make a command-for-command knockoff of Unix and that's fair game, but if I reuse "a = b + c" that's a violation.
