Well that's the question, isn't it? Is downloading those particular URLs infringement if the content is legally allowed to be hosted on YouTube in the first place? If so, was the use of those URLs specifically intended to demonstrate or verify the ability of the tool to infringe copyright? Alternatively, can the tool be said to circumvent a protection device under DMCA section 1201?
(For the record, my answer to all of the above is an emphatic "no".)
And would a "derivative work" that happens to use "some" open source code from youtube-dl, but that never included a URL to a Taylor Swift song anywhere in the code, be considered to have a different intent?
I think this is actually an interesting question. Someone could fork youtube-dl to create "public-___domain-dl" which has multiple references in the README to the fact that it is only intended for downloading public ___domain videos from YouTube (and elsewhere).
Instead of a unit test which successfully downloads a copyright-protected video, it would have a unit test which attempts to download such videos but returns an error message to the user based on a hardcoded blacklist of videos IDs, stored in a separate config file. (It would also have a unit test for successfully downloading a public ___domain video from the developers' own channel, of course).
The developers could also make clear that they are happy to receive DMCA requests to add specific video IDs to their blacklist. As for what happens if the user deletes the blacklist config file, maybe that won't have a unit test.
If such a fork was successful in fending off court cases, though, it would start to raise interesting First Amendment questions. The difference between the two pieces of software, in practice, is just the text in the README telling people not to break the law (plus a single "Delete" button press after installation, which the user would have to learn about from another source). This would mean that the difference between developers complying with the law and breaking the law is including that text in the README, which effectively seems like compelled speech.
> would a "derivative work" ... be considered to have a different intent?
Probably not. If intent were to be demonstrated in court, a derivative work would almost certainly be tainted in turn unless those authors had a _really_ good justification for their actions. (IANAL though; I'm just guessing based on other copyright cases that came up in the media before.)
For example, imagine forking Napster back in the day, redoing the UI, and rewriting half of the internals. Do you really think a court would let that fly? Judges are hardly idiots.
That being said, why not just mirror the original to a host running as a tor hidden service and continue all development efforts there?
My guess is that the answers to the questions is likely "no", "it'd be a hard case for youtube-dl since their lawyers are probably not that great" and "no", even though I wish the answers were "yes", "irrelevant", "irrelevant".
Well that's the question, isn't it? Is downloading those particular URLs infringement if the content is legally allowed to be hosted on YouTube in the first place? If so, was the use of those URLs specifically intended to demonstrate or verify the ability of the tool to infringe copyright? Alternatively, can the tool be said to circumvent a protection device under DMCA section 1201?
(For the record, my answer to all of the above is an emphatic "no".)