Is legal the same as legitimate: AI reimplementation and the erosion of copyleft

There's a long history of people creating clean room implementations of other people's software based on specifications, reverse engineering, etc. A lot of that software is even distributed under GPL. Most drivers in the Linux kernel are good examples. There are things like Dosbox. Databases, video encoders, etc.

So, you could argue that people are using double standards here a bit. It's fine when people take proprietary software and create GPL versions of it. But it's not OK when people take GPL software and create permissively licensed or proprietary versions of it. That's of course not how copyright actually works. The reason all of this is OK is that copyright allows you to do this thing. This isn't some kind of loophole that needs closing but an essential feature of copyright.

The friction here, and common misunderstanding about how copyright works is that you don't copyright ideas but the form or expression of something. Making a painting of a photograph is not a copyright violation. Same idea, different expression. Patents are for protecting ideas. Trademarks are for protecting brands. Some companies have managed to trademark certain color codes even, which is controversial.

There's a lot of legal history for interpretation of what is and isn't "fair use" under copyright of course. It gets much more complicated if you also consider international law and how copyright works in different countries. But people being able to make reasonable use of copyrighted material always was essential to the notion of having it to begin with.

The reason we can have music that uses samples from other people's music without that being a copyright violation is exactly this fair use. In the same way, you can quote from books and create funny memes based on movie fragments. Or create new theater plays, movies, etc. reinterpreting works of others. All legal, up to a point. If you copy too much it stops being fair use and starts being plagiarism.

With software copyright violations, you have to prove that substantial parts of the software were lifted verbatim. Lawyers and judges look at this in terms of how they would apply it to a plagiarism case. Literally - software doesn't get special treatment under copyright. Copyright long predates the existence of software and computers and did not change in any material way after that was invented.

That code isn't going to be open source. And if you use someone else's closed source code you are violating laws that have nothing to do with copyright.

> The point of IP is to encourage the creation of new things.

Then why are corporations allowed to milk successful works for all eternity? Why do we have Disney monopolizing films made half a century ago? Why do we have Nintendo selling people the exact same Mario ROMs from the 80s every single console generation?

They should have like 10 years of copyright so they can turn a profit. Once it expires it's over and the work enters the public domain where it belongs. If they want to keep profiting they should have to keep creating new things. They shouldn't be able to turn shared culture into eternal intellectual property portfolios that they monopolize and then sit on like dragons.

Despite the tech layoffs and rise of AI, programmer hubris is alive and well, that is heartening.

Here we see three engineers writing — at length! — about a hugely complicated matter of law.

No one outside your bubble cares what you think. You are unqualified and your opinions irrelevant. You might as well be debating open heart surgery techniques.

> I feel as though, from an information-theoretic standpoint, it can't be possible that an LLM (which is almost certainly It doesn't need to for my argument to make sense. It's a problem if it reproduces a single copyrighted work (near)-verbatim. Which we have plenty of examples of.