This article describes a scheme to have a computer generate every possible "8-note, 12-beat melody combo" and release it under a Creative Commons license so that anyone can use it (not exactly public domain, but close enough for practical purposes). The idea is that from now on it will not be possible to copyright a basic melodic theme because it will already have been written in a tangible form.

Will this really forestall copyright infringement suits?

Edit: to clarify. Suppose I write a melody, and someone else claims that they wrote the same melody and hence have copyright on it. Can I point to the existence of the melody in this generated corpus as a defence?

  • 2
    It will cause infringement suits, files by authors whose already-existing tunes are infringed by this scheme. – user6726 Feb 28 '20 at 16:32

Seems unlikely that it will "forestall copyright infringement suits".

  1. Some jurisdictions, e.g the USA, say that "Works produced by mechanical processes or random selection without any contribution by a human author are not registrable". On the face of it, in such a jurisdiction copyright can't exist in a randomly generated work. Which the TED talk doesn't mention. https://www.youtube.com/watch?v=sJtm0MoOgiU

  2. Let's imagine a case in a jurisdiction where copyright can exist in such a work. There is a dispute between two artists or labels. The plaintiff produced a well known tune and accuses the defendant of copying this work. The defendant says the plaintiff didn't have copyright in that work because it wasn't original in the first place, there is a 1200GB TAR file (compressed file) on GitHub that contains all possible single octave, 8-note, 12-beat melody combos, which were produced before the plaintiff's work. The plaintiff says, "like the majority of the population I never heard of GitHub, let alone downloaded, uncompressed a 1200GB file and listened to every melody."

  3. That's all aside from plaintiffs or lawyers deciding they have a case or believing the mere threat of civil proceedings will cause the alleged infringer to acquiesce to their demands.

I think they are making a point about the law rather than a realistic means of thwarting copyright disputes. It's reasonable of the creators to say there is a finite set of melodies and the likelihood of inadvertently 'creating' the same melody as someone else may be smaller than we think, maybe copyright law has led to some unjust outcomes and led to a chilling effect on music-making.


The "release under a Creative Commons license" part fails because computer-generated content is not subject to copyright anyway.

It will also not prevent lawsuits, because there's no precedence yet that this scheme works. Someone who feels their copyright is violated will sue you, you can be pretty sure of that.

It might affect the result of the lawsuits, though, but that's uncertain. Copyright claims generally need to show that it's reasonably likely that the violator copied the actual copyrighted work, and didn't independently achieve the same result. It's unlikely that the courts will accept this project as the original source for these beats, unless the defendant in the copyright claim can show to have used this project as a source work. I.e. if the defendant releases a CD and the CD liner credits the source, then it becomes harder to dispute that source. Post-hoc claims are much less likely to succeed.

  • "computer-generated content" - what does this mean though? I know you don't mean any music made with a computer is not copywritable, but there is a lot of automation that goes on in modern music production, and not just the non-creative bits. What if I wrote a program to generate songs and only published the good songs it generated, and never told anyone they were generated by a program I wrote? – Dai Dec 23 '20 at 0:52
  • @Dai: It refers to the process in the question - a fully automated process. Your proposed manual filtering of "good" songs is the creative process. – MSalters Dec 31 '20 at 11:30

You get copyright for a creative work written down in fixed form.

Filling a ten terabyte hard drive with automatically generated sequences of musical notes doesn't involve any creativity. So if there are 100 note sequences that would be nice melodies and could have been created with creativity, there was still no creativity involved. Therefore there is no copyright.

If 100 million monkeys on 100 million typewriters by coincidence typed something like "War and Peace" there would still be no copyright.

However, if you managed to program an Artificial Intelligence that is capable of writing something that could have been Beethoven's 10th symphony, I'd be reasonable sure that you could get a copyright for that. So it's not human written vs. machine written, it's creative vs. not creative.

  • You get the copyright for the AI and the 10th is a derivative of that? – Trish Dec 23 '20 at 8:33

Even if all those melodies could be copyrighted somewhere, a defense for copyright violation is independent creation. If the composer has not listened to a computer generated tune, they can’t have copied it.


And, as user6726 points out in a comment, if it was all published or played, every single preexisting 8-note, 12-beat melody combo would be infringed by this project.

  • How would a defendant establish that they had not listened to a given tune? – user6726 Feb 29 '20 at 5:25
  • 1
    @user6726 it would be fairly easy to establish probable doubt about any musician sitting down and listening to this particular collection, so therefore they almost certainly wouldn’t have copied from it. In other cases, the copying, unconscious or not, was on the basis of the earlier work being something the later musician was likely to have listened to. It’s not going to get played on the radio, it’s not going to be something they hear by chance elsewhere etc etc. – Moo Mar 1 '20 at 8:47

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.