Hacker News Reader

Comment by gmiller123456 | Hacker News Reader

Not a lawyer, but I'm pretty sure that purely algorithmically generated music isn't copyrightable. A copyright can only cover "significantly creative" works. One example is David Slater's "Monkey Selfie", the fact that the photo merely exists doesn't make it copyrightable, it matters how it was created. Had he taken the photo himself, it would be copyrightable. In this specific case, it was a computer that created the music, not a human.

View on HN

Lists of facts, where there is no creativity or originality in the arrangement of those facts, are not copyrightable. (See, e.g., Feist v. Rural, U.S. 1991 [0][1] (which found that a telephone directory "white pages" that listed names and phone numbers alphabetically was not copyrightable).

[0] https://en.wikipedia.org/wiki/Feist_Publications,_Inc.,_v._R....

[1] http://cdn.loc.gov/service/ll/usrep/usrep499/usrep499340/usr...

Regarding David Slater's "Monkey Selfies", as far as I can tell, no court decided the question of whether the photo was copyrightable or whether the copyright belonged to Slater, only that it could not belong to the monkey.

Slater never filed a suit against anyone, so no court ever ruled on it. But he did request it removed from Wikimedia, and lots of experts weighed in. The WikiPedia page has summaries and links to lots of sources: https://en.wikipedia.org/wiki/Monkey_selfie_copyright_disput...

>Lists of facts, where there is no creativity or originality in the arrangement of those facts, are not copyrightable. (See, e.g., Feist v. Rural, U.S. 1991 [0][1] (which found that a telephone directory "white pages" that listed names and phone numbers alphabetically was not copyrightable).

Was there a reason you mentioned this? I don't see the relevance to the surrounding discussion.

A listing of all the possible melodies of a certain length is a list of facts. Sorry if that wasn’t clear.

In don’t think anyone mentioned a listing of all possible melodies either, and if you think something they did mention is functionally equivalent, it would help if you made that explicit.

Algorithmically generated music is not such a list, and does not involve publishing such a list, as no one would want that. They typically involve a lot of selection and choices by the author. They’re only algorithm-generated on the sense that the Sistine Chapel’s ceiling was “brush-generated”.

My comment was in the context of OP's video. The Youtube description of that video states, "I sat down to talk with Damien Riehl and Noah Rubin about their project to copyright every possible melody." (emphasis added)

The video shows that the authors wrote a program to enumerate every possible melody up to a certain size, possibly narrowed by some criteria. A list like that should be as entitled to copyright protection as a list of the first million prime numbers.

Where an algorithm is used as a tool for creativity, I agree that the resulting work should be copyrightable because an algorithm can be like a brush in the hands of an artist. Here, however, that is not the case because the algorithm used simply lists all melodic possibilities and there is no creative input whatsoever.

>My comment was in the context of OP's video. The Youtube description of that video states, "I sat down to talk with Damien Riehl and Noah Rubin about their project to copyright every possible melody." (emphasis added)

Your comment was replying to this one:

https://news.ycombinator.com/item?id=22301633

which makes no mention of any of those points, so I’m not sure I understand why you thought that was the right place to put your comment, or why the connection would have been clear. I think that would have been better as a top level reply, with the relevant part quoted, as you’ve done here.

FWIW, since it was my comment he was replying to, I understood the relevance and connection just fine. He was supplying additional reasons as to why it was not copyrightable.

Good to hear! Not having that context, it just looked like the all-too-common "I'm going to show off my knowledge of the topic despite its dubious relevance".

And yes, I know, "you should watch the video/read the article before commenting and then you'd know the context", but usually it's much easier on everyone for the commenter to just take three seconds to indicate what they're addressing, rather than expect everyone else to blow the ten minutes of their life just for a chance to guess at it.

(For that matter, the title would have been better as "Copyrighting every possible melody...".)

In any case, it was especially confusing since it seemingly had nothing to do with your comment, about algorithmically generated music, which generally refers to something else besides publishing an entire domain (rather than the composer-selected best elements).

>Not a lawyer, but I'm pretty sure that purely algorithmically generated music isn't copyrightable.

As far as I can tell, that's not always true. For now it seems largely context-specific, provided that a human wrote the software generating the output. There's a relatively well-known legal case of someone writing numerological software to generate eschatological texts, where the texts themselves were found to be copyrightable.

The USPTO is currently soliciting comments on whether and under what circumstances AI-generated outputs should be copyrightable [1].

[1]: https://www.govinfo.gov/content/pkg/FR-2019-10-30/pdf/2019-2...

On the other hand, most other photos are copyright by the person who took the photo. However, in those cases, it was a camera that took the photo, not a human. The camera contained a digital computer that executed code (not even written by the photographer in this case!) in order to produce a series of values stored on digital media. All the human did was push a button to tell it to go.

There is also a long history of generative art and as far as I know, the author of the code retains full copyright to the works produced by the program.

I think you were trying to be a bit disingenuous, but that is actually true. Just because a human pressed a button on a camera does not mean the image is copyrightable. Example case [1]

1: https://law.justia.com/cases/federal/appellate-courts/ca11/1...

Generally, the human applies creativity in choosing the location, angle, exposure settings, and timing of "pressing the button"

Many game engine studios are claiming the graphic image outputs of their game engines are copyrighted. Most definitely, they are computer generated.

What about CGI in movie scenes? Also algorithmically generated.

The "creative" aspect is what's important. Designing the sprites, backgrounds, art styles, etc of a videogames is creative.

It's harder to argue that brute force generating every single possible sequence of musical notes is creative.

I guess you can make the argument that coding up the solution that does the generating of the music is the creative part here, just as with games where the creative process is creating the environment for example.

I guess music that is made by programming is just as copyrightable as "normal" music. With Ableton/Max Live and similar, programming and music production is getting closer and closer to each other anyways.

"I guess you can make the argument that coding up the solution that does the generating of the music is the creative part here"

They unambiguously have a copyright to whatever software they wrote to create this output. You do not in general get copyright to the output of software, though. (You may in specific. I'm not getting too detailed here. But you definitely do not in general.)

What is or isn't "significantly creative" will ultimately be left for a jury to decide. Even though the code might have involved significantly creative elements, that does not automatically imply that the program's output also includes significantly creative elements.

A list of all possible permutations of anything isn't creative, regardless of how it was generated.

What about syth arpeggiators? Essentially it allows you to push one key on a keyboard and the computer plays notes and chords according to an algorithmic sequence. These are used heavily in pop and EDM music.

It seems pretty clear that creative input is required: https://youtu.be/vnOXHUK9LK0?t=487

But can a person acquire copyright on a melody they wrote that was also previously algorithmically generated?

You can generally only copyright the parts you change. Sort of like how Disney can copyright their version of Rapunzel, but they don’t own the fairy tale.

(Edit: that’s assuming you’re asking about a derivative work, which I may have misunderstood)

I guess my question is more focused on whether a person can copyright something they created that was created by something else previously. So, if a musician creates a melody and then later finds that same melody was previously recorded on an album from the 1990's can they still copyright their melody? Does it matter whether the 1990's recording was machine-generated melodies or human-generated ones?

This is discussed in the video somewhat. Copyright allows parallel invention. E.g. 2 people can copyright the same thing as long as neither one copied the other. In the video, they claim this is usually difficult to prove in court if the first work had at least moderate notoriety before the second was produced.

I think the sticky bit comes when someone composes one of these songs. They might have put creative work into it, but the computer program put the song on the Internet first. So who owns it?

I think where we're going here is that mere melodies are no longer copyrightable. I also think that was the point, because there was some high profile lawsuit not too long ago where some popular musician sued another for copyright infringement on a centuries old melody. And won.

"Monkey Selfie" is not a good analogy for algorithmically generated music. A better analogy would be satellite images, which of course are copyrightable.

I think it's crystal clear under US copyright law that on-nadir satellite images are not copyrightable; the elements of creative expression that normally give copyrightability to a photograph (composition, lighting, framing, choice of angle) are entirely absent. Feist, Hearn, and Bridgeman seem to provide a clear-cut answer in such cases. What jurisprudence are you thinking of?

Here are some sources that say they are copyrightable:

https://ieeexplore.ieee.org/abstract/document/1027166

https://scholarlycommons.law.northwestern.edu/cgi/viewconten...

https://ieeexplore.ieee.org/abstract/document/1292646

Any GIS tool with satellite images (Google earth/maps, Bing maps, etc...) will show a copyright logo on the screen specific to the satellite images.

From your second link https://scholarlycommons.law.northwestern.edu/cgi/viewconten..., which is a law review article, not precedent or statute: "Copyright protection is thus unavailable to unenhanced data. ... Predictably, EOSAT does not attempt to copyright unenhanced Landsat data." However, it's from 1990, prior to Feist (1991) and Bridgeman (1999), though not Hearn (1987). Even so, it doesn't attempt to make the argument that you're making, that raw satellite photos themselves are subject to copyright: "The statutory definition of a copyright certainly suggests that...data which has been enhanced through human creativity qualifies for protection." (Emphasis mine.)

Law review articles, like legal briefs, are primarily attempts to advocate a certain point of view by presenting the best available arguments for it, in the hopes of influencing future jurisprudence or legislation. The positions they advocate are thus not reliable guides to reality, although they are more persuasive to the extent that their arguments are based on reality and thus not easily demolished.

The other two https://ieeexplore.ieee.org/abstract/document/1027166 and https://ieeexplore.ieee.org/abstract/document/1292646 not only aren't jurisprudence, they aren't even law review articles, nor do they reference any jurisprudence or make any legal arguments. Also, they are both from Singapore. Actually they are papers about steganography, not copyright.

Did you... did you just google [satellite image copyright] and paste the first few links you saw without reading them, hoping you could waste a bunch of my time for a minimal amount of effort on your part?

"Copyright protection is thus unavailable to unenhanced data ... Predictably, EOSAT does not attempt to copyright unenhanced Landsat data" - That's a reference to data from a public satellite.

Bridgeman is specifically about imaging artworks. You're taking it way out of context. Here's the summary from wiki https://en.wikipedia.org/wiki/Bridgeman_Art_Library_v._Corel...:

"...exact photographic copies of public domain images could not be protected by copyright in the United States because the copies lack originality."

That has nothing to do with satellite images. If you disagree I would ask you to cite jurisprudence or legislation.

You haven't addressed the fact that google/bing/any other GIS tool specifically put copyrights on their satellite images.

From the wiki https://en.wikipedia.org/wiki/Google_Earth:

"Every image created from Google Earth using satellite data provided by Google Earth is a copyrighted map. Any derivative from Google Earth is made from copyrighted data which, under United States Copyright Law, may not be used except under the licenses Google provides"

If this is not correct you should edit it accordingly.

I said, "I think it's crystal clear under US copyright law that on-nadir satellite images are not copyrightable." Google's claim to own copyright in the output of Google Maps or Google Earth may or may not be correct, but either way it's irrelevant, because those are not on-nadir satellite photos, and my claim was specifically scoped to on-nadir satellite photos (and US copyright law). I've cited three judicial precedents to support that claim; you have a weak argument about one of those precedents that suggests you don't understand it, and even the 30-year-old law review article you chose to support your position specifically disclaims the claim you're making, in the case it discusses, which is indeed a public satellite.

Google Maps and Google Earth output are not on-nadir satellite photos. They are composite images produced by an elaborate production pipeline drawing on many sets of data and possibly enough human creativity to produce a copyrightable work. That's a question for a court to decide, not Google or Wikipedia. It seems clear that if their results are found to be copyrightable, it will not be due to the satellite photos that some of them draw on.

The single example that you were able to find referencing photography was specifically scoped to artistic works. None of those cases back up the claim that "on-nadir satellite images are not copyrightable".

If it were "crystal clear" you'd be able to find a more specific example, which you can't.

None of the sources you provided attempt to claim that raw, unmodified satellite photos are copyrightable. They all have to do with "enhanced" or "watermarked" photos. There is no question that Google owns the copyright and trademark to their logo, and adding it to an image will allow them to enforce their copyright on its distribution. Additionally, Google has done processing on the photos to remove clouds, add roads, city names, and a lot of other info.

But a claim that a raw, uncreative, unprocessed, mechanically produced photo is subject to copyright would be a major exemption from the requirement that anything copyrightatlbe be significantly creative. Unless you have an actual case involving an unprocessed satellite image, it's pretty safe to assume they are not copyrighted.

> In this specific case, it was a computer that created the music, not a human.

You don’t really get any music made by humans these days, the sound of hitting them against each other just isn’t that popular. So we get music made by guitars, pianos, trumpets ect. But of cause all that music is still “made by humans using those instruments” just like this was made by a human using a computer.

If he just remembered to take a photograph of the photograph and only share that version, he’d have been all set :)

Some people have been claiming copyright on algorithmic outputs. Wolfram is one such prominent example. I don't know if it's ever been challenged, but he does claim all output of Alpha is copyrighted to him.

What if you record a robot playing the notes on a piano?

Better question: what if a robot records itself playing the notes on a piano?

Reply on news.ycombinator.com