6 ms·

What irks me about this is that 100% of their data is provided for free, by the community that they have fostered, the people like myself who have answered > 25

What irks me about this is that 100% of their data is provided for free, by the community that they have fostered, the people like myself who have answered > 2500 questions[0], and now SO feels hard-done-by by LLMs using all their hard work to create tools like CodeGPT, GitHub copilot, etc. Were it really a site for helping developers to improve their skills and increase their productivity through the give-and-take model that SO was, at least once upon a time, SO should perhaps take a deep breath and realise that this might not change a thing apart from causing their contributors to feel like they were never part of it in the first place. I'm not sure if I've correctly articulated that, but I do find SO's stance to be quite revealing. It feels to me like they're crying foul that ChatGPT and the how many other systems out there are stealing their revenue. None of the contributors (apart from the employee ones, I suppose) ever got paid any currency other than high-fives in the form of rep, medals, the gamified stuff, moderation rights, and at certain rep levels some swag in the form of t-shirts and the usual. I never wanted any money from SO, but the revelation of this attitude has left me feeling, well, a little sad to say the least. [0]https://stackoverflow.com/users/70393/karim79 https://stackoverflow.com/users/70393/karim79

hyperbovine 3y ago
Fair enough, but you were fully aware of this arrangement going in, and chose to participate. SO didn't opt into being training data for ChatGPT, and I doubt they would have given the chance. You may object that SO implicitly did so by making their site available to the public, but the ethics of GAI training data a new moral gray area that we're still navigating. They at least have something of a case to be made.
- karim79 3y ago
  > Fair enough, but you were fully aware of this arrangement going in, and chose to participate. Yep, but that's not the point. > SO didn't opt into being training data for ChatGPT, and I doubt they would have given the chance. Neither did Wikipedia (at least to my knowledge). I thought the point of opening up information was to benefit the public, first and foremost, and without hidden terms which state something along the lines of "it's free and open information built by the community, but when something disrupts our ads-driven business model and we make it unfree". It would have been nice if they had at least allowed their contributors to vote on this, or have some sort of a say.
  - baliex 3y ago
    > "it's free and open information built by the community, but when something disrupts our ads-driven business model and we make it unfree" I feel the same, that it would be much better today if that was the agreement we all entered into. (Though, I doubt anyone would have read it anyway so I’m not sure it would have made any difference.) But it seems that no-one saw this disruption coming, so it wasn’t possible to plan ahead for this outcome. Call it complacency, call it ignorance, it’s too late to plan now, so we get this kind of reaction instead
abetusk 3y ago
I think I see. The economies of the internet are changing. Now with LLMs being accessible at an exponentially cheaper rate, we're seeing old models crumble and new models rising. The era of moderated user content is changing drastically and the stalwarts of social networking, or adjacent, services are closing ranks to try and anticipate the change. Thanks for the insight. I had a vague notion that these new policies were because of some recession or some other basic economic issue. I think a better theory is that the lowering economic cost of LLMs that are becoming available are the reason for all these changes.
- karim79 3y ago
  Absolutely, and very well put. The costs are pretty close to zero at this point for something that would have been mostly considered, publicly at least, as science fiction just a year ago. It's what the printing press was to the scribe, only way, way more disruptive. I don't know if I'm correct here, but that is what it looks like right now and we're basically still at year zero. I mean, it even got the Google all riled up. The company which basically equates to "The Internet" is having to react, I'm pretty sure that's the first time I've seen such a thing in my life. It goes way beyond their reaction to the iPhone, way beyond the threat of MS exercising their monopoly(ish) power back in the day to try to mitigate the Google threat. History repeats itself albeit in a fascinating way which I'm still trying to grasp. I don't blame SO. I think they are acting rationally and as anyone would facing such a threat. Secretly, I want the internet ad economy of nothing to go away. I won't mention any names (cough taboola cough cough) but that might be the only upside to this tech. Let's see what happens six months or so from now. Side note, I'm running llama on a really crappy old server and that was enough to convince me that I'll be able to run an LLM on my watch in the near future.
ryan29 3y ago
> None of the contributors (apart from the employee ones, I suppose) ever got paid any currency other than high-fives in the form of rep, medals, the gamified stuff, moderation rights, and at certain rep levels some swag in the form of t-shirts and the usual. I would love to see some kind of identity and reputation system where the "high-fives in the form of rep" could follow people across communities. It may not feel like much compensation if you've contributed over 2500 answers, but having reputation gained in your area of expertise grant you a high level of trust to interact in other communities could be valuable, at least in my opinion. Assuming they're making this move to protect against AI / LLMs, I think SO is in an impossible situation here. When all the ChatGPT hype started, one of my first questions was "what happens to the incentive for contributors and creators?" Why would I want to contribute on a platform if I know an AI model is going to come in, take my contribution, and regurgitate it back to the masses in a way that I can't control? Even if I get some attribution from the AI/LLM, do I even want it? If the LLM is blending content from multiple sources, which changes the context and presentation I put effort into, is the quality going to be high enough to match what I strive to achieve for myself when I'm trying to build a reputation as a high quality contributor? What if the AI is hallucinating objectively poor quality content and giving me partial attribution? So, for me, part of the social contract with SO is that I provide answers, but I get to control the entire interaction; the context, the presentation (mark up), defending criticism in the comments, etc.. In addition to that, since the entire conversation happens inline, I can be corrected by someone even more knowledgeable than me and use that feedback for self improvement. I think AI is going to be disruptive and the whole idea, for me anyway, behind disruption is that you break an existing system and then everyone is free to take a shot at claiming part of the new gold rush that occurs while trying to build the replacement. The problem with AI is that it's going to break a lot of services that do a good job of serving the community and shouldn't be broken. SO is a great example of a healthy community that doesn't need disruption, but the massive amount of high quality, curated content is going to make them a prime target for LLM training. Personally I think the only solution is for "noai" variants of popular open source licenses so contributors have the ability to make it clear they don't want to contribute to AI/LLM companies. If SO had an option to flag contributions as CC-BY-SA-NOAI, I'd enable it on my stuff going forward.
- 3y ago

hyperbovine 3y ago

Fair enough, but you were fully aware of this arrangement going in, and chose to participate. SO didn't opt into being training data for ChatGPT, and I doubt they would have given the chance. You may object that SO implicitly did so by making their site available to the public, but the ethics of GAI training data a new moral gray area that we're still navigating. They at least have something of a case to be made.

karim79 3y ago
> Fair enough, but you were fully aware of this arrangement going in, and chose to participate. Yep, but that's not the point. > SO didn't opt into being training data for ChatGPT, and I doubt they would have given the chance. Neither did Wikipedia (at least to my knowledge). I thought the point of opening up information was to benefit the public, first and foremost, and without hidden terms which state something along the lines of "it's free and open information built by the community, but when something disrupts our ads-driven business model and we make it unfree". It would have been nice if they had at least allowed their contributors to vote on this, or have some sort of a say.
- baliex 3y ago
  > "it's free and open information built by the community, but when something disrupts our ads-driven business model and we make it unfree" I feel the same, that it would be much better today if that was the agreement we all entered into. (Though, I doubt anyone would have read it anyway so I’m not sure it would have made any difference.) But it seems that no-one saw this disruption coming, so it wasn’t possible to plan ahead for this outcome. Call it complacency, call it ignorance, it’s too late to plan now, so we get this kind of reaction instead

abetusk 3y ago

I think I see. The economies of the internet are changing. Now with LLMs being accessible at an exponentially cheaper rate, we're seeing old models crumble and new models rising. The era of moderated user content is changing drastically and the stalwarts of social networking, or adjacent, services are closing ranks to try and anticipate the change. Thanks for the insight. I had a vague notion that these new policies were because of some recession or some other basic economic issue. I think a better theory is that the lowering economic cost of LLMs that are becoming available are the reason for all these changes.

karim79 3y ago
Absolutely, and very well put. The costs are pretty close to zero at this point for something that would have been mostly considered, publicly at least, as science fiction just a year ago. It's what the printing press was to the scribe, only way, way more disruptive. I don't know if I'm correct here, but that is what it looks like right now and we're basically still at year zero. I mean, it even got the Google all riled up. The company which basically equates to "The Internet" is having to react, I'm pretty sure that's the first time I've seen such a thing in my life. It goes way beyond their reaction to the iPhone, way beyond the threat of MS exercising their monopoly(ish) power back in the day to try to mitigate the Google threat. History repeats itself albeit in a fascinating way which I'm still trying to grasp. I don't blame SO. I think they are acting rationally and as anyone would facing such a threat. Secretly, I want the internet ad economy of nothing to go away. I won't mention any names (cough taboola cough cough) but that might be the only upside to this tech. Let's see what happens six months or so from now. Side note, I'm running llama on a really crappy old server and that was enough to convince me that I'll be able to run an LLM on my watch in the near future.

ryan29 3y ago

> None of the contributors (apart from the employee ones, I suppose) ever got paid any currency other than high-fives in the form of rep, medals, the gamified stuff, moderation rights, and at certain rep levels some swag in the form of t-shirts and the usual. I would love to see some kind of identity and reputation system where the "high-fives in the form of rep" could follow people across communities. It may not feel like much compensation if you've contributed over 2500 answers, but having reputation gained in your area of expertise grant you a high level of trust to interact in other communities could be valuable, at least in my opinion. Assuming they're making this move to protect against AI / LLMs, I think SO is in an impossible situation here. When all the ChatGPT hype started, one of my first questions was "what happens to the incentive for contributors and creators?" Why would I want to contribute on a platform if I know an AI model is going to come in, take my contribution, and regurgitate it back to the masses in a way that I can't control? Even if I get some attribution from the AI/LLM, do I even want it? If the LLM is blending content from multiple sources, which changes the context and presentation I put effort into, is the quality going to be high enough to match what I strive to achieve for myself when I'm trying to build a reputation as a high quality contributor? What if the AI is hallucinating objectively poor quality content and giving me partial attribution? So, for me, part of the social contract with SO is that I provide answers, but I get to control the entire interaction; the context, the presentation (mark up), defending criticism in the comments, etc.. In addition to that, since the entire conversation happens inline, I can be corrected by someone even more knowledgeable than me and use that feedback for self improvement. I think AI is going to be disruptive and the whole idea, for me anyway, behind disruption is that you break an existing system and then everyone is free to take a shot at claiming part of the new gold rush that occurs while trying to build the replacement. The problem with AI is that it's going to break a lot of services that do a good job of serving the community and shouldn't be broken. SO is a great example of a healthy community that doesn't need disruption, but the massive amount of high quality, curated content is going to make them a prime target for LLM training. Personally I think the only solution is for "noai" variants of popular open source licenses so contributors have the ability to make it clear they don't want to contribute to AI/LLM companies. If SO had an option to flag contributions as CC-BY-SA-NOAI, I'd enable it on my stuff going forward.

3y ago