Recently, the people behind the re3 project (public decompilation of the games GTA III and GTA Vice City) were served a civol lawsuit by Take Two, the owning company of the games.

What the group did was run the binaries through decompilers, such as IDA, and making a recreation of equivalent high-level C++ source code.

From my underdstanding of law and jurisprudence, this would not breach copyright law. My reasoning is the following:

Computer programs are only covered by copyright because their source code is, therefore so is the resulting object code. However, the processes and sequences of operations that make the program do what it does aren't copyrightable (a patent would be better suited for those things).

Taking that into account, by decompiling the object code into new, re-arranged source code, the only elements that would remain are the specific sequences and processes that are represented by it. No copyrighted works remain, neither the object code nor the original source code the engineers wrote and later compiled.

I have what I believe to be a good analogy to explain it better.

Suppose there is a recording (object code/binary) of an orchestral performance (original source code) of a song that is in the public domain (processes/sequences of operations) Wouldn't the decompilation project be the equivalent of me running the recording through a program which would recognize the individual notes and throw together a music sheet? The recording and performance are copyrighted, but the core concept in which they are based on is not (in this case because the song is on the public domain, however keep in mind I only used this to have an example of an "non-copyrighted" thing. processes/sequences of operations can never enjoy copyright protections at all)

While of course noting that we cannot predict the future: could this argument hold up? If not, what are its flaws?

  • 4
    "What the group did was run the binaries through decompilers, such as IDA, and making a recreation of equivalent high-level C++ source code." That is the exact opposite of clean-room reverse engineering, so I'm not sure why you mention that in your title? Nov 30 '21 at 4:37
  • 1
    Isn't that like arguing that I can freely copy the Harry Potter novels if I OCR them? I only reproduced the order of letters on the page, and didn't duplicate J. K. Rowling's handwriting. Nov 30 '21 at 4:58
  • 1
    Or, if I translate them into French: by replacing "wizard" with "sorcier" I have only copied the idea of a wizard, and not the actual word - and the idea of a wizard is not copyrightable. Yet in fact an French translation of Harry Potter is the clearest possible example of a derivative work. Copyright law is not so reductive as you make it out to be. Nov 30 '21 at 5:02
  • 1
    @ENZOLU: Doesn't the binary likewise express only the operations in sequence, without any of the variable names or any of that other stuff? Yet you seem to accept that the binary is properly copyrighted as a derivative work of the source. I don't see how you can argue that copyrightability is preserved through one step of translation, yet somehow destroyed when there are two steps. Nov 30 '21 at 5:37
  • 1
    Regarding your remark about clean-room engineering (and how you recognize that decompilation is not an example of it), you may want to edit this into your post, as it seems to be misleading people as it stands. Nov 30 '21 at 6:03

I think that both your examples would be considered, if not outright copies, then at the very least derivative works of the originals. Under US law, the copyright holder of a work has the exclusive right to prepare derivative works. So anyone who prepares such a work without the authorization of the copyright holder is infringing their copyright and will be liable for such infringment.

The US Copyright Office, in their Circular 14, explicitly includes "a new version of an existing computer program" as an example of a derivative work. That would certainly seem to cover the GTA folks. The concept that derivative works generally include "translations" might capture them, too.

Your example of transcribing a musical performance isn't explicitly mentioned, but I think your sheet music would likewise be a derivative work of the performance: it includes all the editorial and arrangement choices made by the orchestra. Of course, both the performance and your transcription are derivative works of the original composition, but since the composition is out of copyright, there's no legal problem with that aspect.

  • What I'm arguing is that the parts of the work that were copied do not constitute copyrightable elements. Wouldn't only the specific expressions of the processes be under copyright (that particular source code and binaries) instead of the actual instructions and processes (and their particular order) that make the program work?
    – ENZOLU
    Nov 30 '21 at 4:39
  • 4
    @ENZOLU: The decompilation is derivative of those specific expressions, since it is a translation of them into another language. Nov 30 '21 at 4:59
  • So then how did clean room reversing earn its right in court? While the people who wrote the new code didn't see the original, not only did it achieve the same results, but it did the same things as well
    – ENZOLU
    Nov 30 '21 at 5:36
  • 4
    @ENZOLU: Because the clean-room engineers didn't copy, or derive from, the specific expressions of the algorithms that are contained in the original source and binary. Nov 30 '21 at 5:38
  • 2
    @ENZOLU Look up Clean-room reverse engineering and look into both the technical details and the legal argument on why it is allowed. You're completely misunderstanding what it is and thinks that your question is clean-room reverse engineering (it's not). Once you know what clean-room reverse engineering actually is, you'll know what is wrong with your question.
    – Nelson
    Nov 30 '21 at 5:58


Translations are covered by copyright.

The original high-level code is translated to binary by an automated process and then translated back into another (or the same) high-level language also by an automated process.

This is equivalent to running an English novel through Google Translate to French and then running the French through again into Spanish (or back to English).

Both are copyright violations.

Although the results from the latter are much funnier. Specifically, these automatic translations aren't copyright violations because the transformative nature and artistic endeavour involved make it Fair Use.

This is also explicitly not clean-room reverse engineering. CRRE is looking at the input and output, writing software specifications from that without ever decompiling the code and using those to develop the software.

  • Just to b e clear, "clean-room reverse engineering" in which you produce a work independently of another similar work in a manner that makes it manifestly provable that you developed it independently because the risk of someone assuming that it was derived from someone else's work when it actually was an independent invention, is allowed. The example in the question simply isn't such a case because they used another author's work as the basis of their own work. The classic fictional "clean room" story is "Paycheck" by Philip K. Dick. en.wikipedia.org/wiki/Paycheck_(novelette)
    – ohwilleke
    Dec 1 '21 at 0:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.