5 ms·

About 5 years ago, StackOverflow messed up and declared that they were making all content submitted by users available under CC-BY-SA 4.0 [1]. The error here is

About 5 years ago, StackOverflow messed up and declared that they were making all content submitted by users available under CC-BY-SA 4.0 [1]. The error here is that the users-content agreement was that all users' contributions are made available under CC-BY-SA 3.0 (and not anything about later). In the middle there were also some licensing problems concerning code vs noncode that were confusing. I remember thinking that if any of the super answerers really wanted, they could have tried to sue for illegally making their answers available under a different license. But I thought that without any damages, this probably wasn't likely to succeed. But now I wonder whether making all content available to AI scrapers and OpenAI in particular might be enough to actually base a case. As far as I can tell, StackOverflow continued being duplicitous with what license applies to what content for half of the year 2018 and the first few months of the year 2019. Their current licensing suggests CC-BY-SA 3.0 for things before May 5 2018, and CC-BY-SA 4.0 for things after. Sometime in early 2019 (if memory serves, it was after the meta post I link to), they made users login again and accept a new license agreement for relicensing content. But those middle months are murky. I should emphasize that I know nothing. [1]: https://meta.stackexchange.com/q/333089/205676 https://meta.stackexchange.com/q/333089/205676

kragen 2y ago
> if any of the super answerers really wanted, they could have tried to sue for illegally making their answers available under a different license. they can plausibly sue people other than stackoverflow if they attempt to reuse the answers under a different license. but i think it's very difficult to find a use that 4.0 permits that 3.0 doesn't
- StevenXC 2y ago
  3.0 has a "bug" that makes it risky to use materials without very careful attribution: https://doctorow.medium.com/a-bug-in-early-creative-commons-licenses-has-enabled-a-new-breed-of-superpredator-5f6360713299 https://doctorow.medium.com/a-bug-in-early-creative-commons-...
  - miohtama 2y ago
    I don't think this is a practical issue, really. I assume linking to the original answer is sufficient attribution. In the link you can find name, license and figure out if the answer was modified. Also linking the answer in a source comment is the smallest professional courtesy everyone should be doing. If you have some issue of not linking an answer then you likely do not deserve the answer in the first place.
    - eviks 2y ago
      The blog illustrates that such assumptions about what's a sufficient attribution are fraught with danger, so "the smallest professional courtesy" can expose you to a $150k risk
drivingmenuts 2y ago
People put their content on the site for the public to use, and now the public is using it, it's just that "the public" includes AIs. Admittedly, a non-human public, nonetheless ...
- aaron695 2y ago
  [dead]
- _xivi 2y ago
  The problem is LLMs don't provide attribution/credit which directly violates the license[0] Otherwise search engines were already "non-human public" that scraped the site but directly linked to the answers, which was great. They didn't claim its their work like these models. The problem isn't human vs non-human. LLMs aren't magic, they don't create stuff out of thin air, what they're doing is simply content laundering. [0] https://creativecommons.org/licenses/by-sa/4.0/#ref-appropriate-credit https://creativecommons.org/licenses/by-sa/4.0/#ref-appropri...
- postepowanieadm 2y ago
  You have to agree on how your work may be used, no one has expected it will be sold for ai training.
frognumber 2y ago
My understanding of licensing law is that something like 3.0 -> 4.0 is very unlikely to be a winnable case in the US. Programmers think like machines. Lawyers don't. A lot of confusion comes from this. To be clear, there are places where law is machine-like, but I believe licensing is not one of them. If two licenses are substantively equivalent, a court is likely to rule that it's a-okay. One would most likely need to show a substantive difference to have a case. IANAL, but this is based on one conversation with a law professor specializing in this stuff, so it's also not completely uninformed. But it matches up with what you wrote. If your history is right, the 2019 changes is where there would be a case. The joyful part here is that there are 200 countries in the world, and in many, the 3.0->4.0 would be a valid complaint. I suspect this would not fly in most common law jurisdictions (British Empire), but it would be fine in many statutory law ones (e.g. France). In the internet age, you can be sued anywhere!
- sidewndr46 2y ago
  It is worth remembering that law professors have a vested interest in making sure the system work as you described. If contract law was straightforward, they'd be out of job.
  - frognumber 2y ago
    I agreed in the abstract, but not in the specific (the specific professor was one of integrity, and sufficiently famous this was not an issue). However, it's worth noting the universe is a cesspool of corruption. If you pretend it works the way it ought to and not the way it does, you won't have a very good time or be very successful. The entire legal system is f-ed, and if you pretend it's anything else, you'll end up in prison or worse.
  - AnarchismIsCool 2y ago
    That's an admirable goal but if there are any "bugs" in the contract you probably don't want it executed mindlessly. Human language isn't code and even code isn't always perfect so I'd rather not be legally required to throw someone out a window because someone couldn't spell "defederate".
trueismywork 2y ago
If it is indeed CC-BY-SA then, openAI needs to publish their weights under the same license.

kragen 2y ago

> if any of the super answerers really wanted, they could have tried to sue for illegally making their answers available under a different license. they can plausibly sue people other than stackoverflow if they attempt to reuse the answers under a different license. but i think it's very difficult to find a use that 4.0 permits that 3.0 doesn't

StevenXC 2y ago
3.0 has a "bug" that makes it risky to use materials without very careful attribution: https://doctorow.medium.com/a-bug-in-early-creative-commons-licenses-has-enabled-a-new-breed-of-superpredator-5f6360713299 https://doctorow.medium.com/a-bug-in-early-creative-commons-...
- miohtama 2y ago
  I don't think this is a practical issue, really. I assume linking to the original answer is sufficient attribution. In the link you can find name, license and figure out if the answer was modified. Also linking the answer in a source comment is the smallest professional courtesy everyone should be doing. If you have some issue of not linking an answer then you likely do not deserve the answer in the first place.
  - eviks 2y ago
    The blog illustrates that such assumptions about what's a sufficient attribution are fraught with danger, so "the smallest professional courtesy" can expose you to a $150k risk

drivingmenuts 2y ago

People put their content on the site for the public to use, and now the public is using it, it's just that "the public" includes AIs. Admittedly, a non-human public, nonetheless ...

aaron695 2y ago
[dead]
_xivi 2y ago
The problem is LLMs don't provide attribution/credit which directly violates the license[0] Otherwise search engines were already "non-human public" that scraped the site but directly linked to the answers, which was great. They didn't claim its their work like these models. The problem isn't human vs non-human. LLMs aren't magic, they don't create stuff out of thin air, what they're doing is simply content laundering. [0] https://creativecommons.org/licenses/by-sa/4.0/#ref-appropriate-credit https://creativecommons.org/licenses/by-sa/4.0/#ref-appropri...
postepowanieadm 2y ago
You have to agree on how your work may be used, no one has expected it will be sold for ai training.

frognumber 2y ago

My understanding of licensing law is that something like 3.0 -> 4.0 is very unlikely to be a winnable case in the US. Programmers think like machines. Lawyers don't. A lot of confusion comes from this. To be clear, there are places where law is machine-like, but I believe licensing is not one of them. If two licenses are substantively equivalent, a court is likely to rule that it's a-okay. One would most likely need to show a substantive difference to have a case. IANAL, but this is based on one conversation with a law professor specializing in this stuff, so it's also not completely uninformed. But it matches up with what you wrote. If your history is right, the 2019 changes is where there would be a case. The joyful part here is that there are 200 countries in the world, and in many, the 3.0->4.0 would be a valid complaint. I suspect this would not fly in most common law jurisdictions (British Empire), but it would be fine in many statutory law ones (e.g. France). In the internet age, you can be sued anywhere!

sidewndr46 2y ago
It is worth remembering that law professors have a vested interest in making sure the system work as you described. If contract law was straightforward, they'd be out of job.
- frognumber 2y ago
  I agreed in the abstract, but not in the specific (the specific professor was one of integrity, and sufficiently famous this was not an issue). However, it's worth noting the universe is a cesspool of corruption. If you pretend it works the way it ought to and not the way it does, you won't have a very good time or be very successful. The entire legal system is f-ed, and if you pretend it's anything else, you'll end up in prison or worse.
- AnarchismIsCool 2y ago
  That's an admirable goal but if there are any "bugs" in the contract you probably don't want it executed mindlessly. Human language isn't code and even code isn't always perfect so I'd rather not be legally required to throw someone out a window because someone couldn't spell "defederate".

trueismywork 2y ago

If it is indeed CC-BY-SA then, openAI needs to publish their weights under the same license.