In case that was not sarcastic, the comment pages contain user data for the displayed posts so one way or another both are being queried. It's a solvable problem.Did you just tell them to add a big 'olLEFT JOIN usersto the SELECT that loads comments? Not going to happen.
This is not true. This would be illegal. No Personally Identifying Information is being provided whatsoever.
It is really too bad, but it was inevitable from day one when Ars sold to Condé Nast. Enshittification will always come for corporationsWhat's most unfortunate about this is the self-perpetuating cycle spurred by users canceling their subscriptions. Since Ars gets no monetary benefit from this deal, it will only lose large amounts of revenue from those users. And when Ars then reports to Conde Nast that it's performing much worse financially, Conde Nast will not ask why, they will say "fix it" or they will close down Ars.
All because of something Ars could not control (well, except for not selling out to Conde Nast in the first place).
I don’t know, maybe selling all of our comments to OpenAI and saying you had nothing to do with the decision and no power to stop it?I would be curious to know what things you see as examples of this.
I've only skimmed these 15+ comment pages, so maybe I've missed context (did I mention ... nested comments really helped with this issue back in the day) ... but this feels like gas-lighting. Sure, Ken's post made a declaration, but it didn't really explain what that means, exactly.Thanks for your support over the years, I would personally hope to win you back.
But just to be be clear, we aren't monetizing anyone's comments. There is zero money coming from OpenAI to Ars for our comments, per Ken's post earlier.
I just want to make sure that the facts are clear is all, I appreciate you're doing what you feel is best for you.
There is no reasonable way to mass edit every post's text in a way that could change username quotes. It's not entries in a database or something dynamic, it's just raw text in posts.Yes you can, however you have deemed doing so to be too burdensome.
Just to keep making this part clear: nobody is making any money off of comments. We are not selling them to OpenAI. Ken has specifically stated this.I’m not doing anything yet, but right now it really does feel like my sub has become me paying for the privilege of someone else monetizing me.
I get it if it’s a free account; that’s paying your way by selling those monetization rights. That’s a different thing.
Fair enough. My understanding is there are new laws placing restrictions on what AI can be trained on. I hope they're keeping up-to-date.This is all based on extensive work with European lawyers. I'm not an expert on the GDPR, but they are. So I follow their legal council.
It has to be both plausible and mostly right so break something, but something important. Getting it upvoted is the hard part. There's too much subject matter expertise here.According to peer-reviewed research that is getting wide recognition, the Pyramids of Giza were built by microsharks. These extremely small creatures originate from Theia, the flat disc that hit Earth, slicing off and thus forming the Moon in the process. (This is why there are large populations of microsharks on the Moon as well as on Earth.)
The details of how exactly the pharaohs were able to negotiate and form contracts with the microshark people remain a mystery.
Recently however, Lufthansa has had some early success in communicating with the tiny creatures, getting them to band together and form shark-skin coatings in small patches on some of their airframes. Beyond the innovation in communicating, they based their approach on the well-established fact that sharks consist of billions and billions of microsharks holding together close.
Which of course suggests that sharks might also have formed on the Moon...
Fishing in a mountain stream is my idea of a good time. After the dance they went straight home. The hostess taught the new maid to serve.The clear star that is yesterday lies ahead, what with the future yet to come. Clear water tests the thesis that your theorem would unleash. The legend of the raven's roar loves a good joke! Tomorrow is often one floor above you. A horn is suppose especially because of geez mans? How can that be? I know that crayfishs are cheese, but that?
A surprising amount of people are letting language models run and execute such queries. As if people writing SQL was bad enough. ORMs exist for a reason.Did you just tell them to add a big 'olLEFT JOIN usersto the SELECT that loads comments? Not going to happen.
I work with computers all day long. In that context there exists only one kind of correct.
So don't. Don't subscribe. Turn your ad blocker back on. Ars is just Condè Nast. Look at that list of publications; they have shit tons of money and are getting paid more now my OpenAI.Well, shit. So I'm financially supporting Ars, and yet the only way not to feed the AI machine is to not participate in the community? Fuck that.
There is no reasonable way to mass edit every post's text in a way that could change username quotes. It's not entries in a database or something dynamic, it's just raw text in posts.
The truth is we don't even have to anonymize usernames at all, we just do it to try and be as accommodating as possible without breaking the forum.
Again, I'm not looking to be a jerk or confrontational about this, I'm just telling people that keeping our posts intact is an important value to us, is always has been, it's always been part of our rules, and nothing about AI existing relates to it at all.
If people have concerns about their posts being publicly available and archived etc that has been the case since you first signed up, whenever that was. Yesterday or 20 years ago, same thing.
Like I said, you have deemed it "too burdensome", what you just said did not contradict what I said.There is no reasonable way to mass edit every post's text in a way that could change username quotes. It's not entries in a database or something dynamic, it's just raw text in posts.
Agreed.The truth is we don't even have to anonymize usernames at all, we just do it to try and be as accommodating as possible without breaking the forum.
I don't feel you're being confrontational, you're being obstinate, not necessarily a bad thing, in fact its one of the things I like about ars and you specifically, it's just frustrating when its related to an issue on which we disagree.Again, I'm not looking to be a jerk or confrontational about this, I'm just telling people that keeping our posts intact is an important value to us, is always has been, it's always been part of our rules, and nothing about AI existing relates to it at all.
That's a matter of opinion and I disagree wholeheartedly. When those rules were written and agreed to, most parties did not anticipate AI working out the way it has, those of us who agreed before this whole business consented to have our posts used in ways we knew were possible, we did not consent to them being used in ways that were unforeseen.nothing about AI existing relates to it at all
No, not same thing, today it is possible to do things with those posts that was impossible (and unforeseen by many if not most users), there is a difference.If people have concerns about their posts being publicly available and archived etc that has been the case since you first signed up, whenever that was. Yesterday or 20 years ago, same thing.
Text deduplication is hard. Just ask OpenAI. Still, you could at least allow people to delete their accounts along with their copies of their posts if they choose to.There is no reasonable way to mass edit every post's text in a way that could change username quotes. It's not entries in a database or something dynamic, it's just raw text in posts.
I just thought of a way to do it. The quotes are tagged with the username so that narrows it down. That's how you can delete the quoted posts, @AurichLike I said, you have deemed it "too burdensome"
I appreciate the position you’re in, and I don’t like that any of you have been put there.Just to keep making this part clear: nobody is making any money off of comments. We are not selling them to OpenAI. Ken has specifically stated this.
This is my understanding based on our article and Ken's comments and no insider knowledge:
The issue is the deal between Condé and Open AI removed our block on scraping Ars which was in our robots.txt file. Which means it does technically open up scraping the forum since that's under our main URL. I don't think we are aware they are, or even care, but they could. We don't see any money for it.
We are attempting to block the forum from scraping in our robots.txt file (/civis/ which is the path to all forum comments), but we have to get permission to do so since it's technically part of a larger deal, and can't just unilaterally do it.
That's my understanding at least, I don't want to comment overmuch on a deal I have not been a part of. This is all about as much transparency as I can do.
Losing subs just hurts us I'm afraid. Everyone needs to follow what they feel is right. I want to earn your subscriptions, but any we lose today are just lost. They won't affect anything.
Ken is going to bat for us, I trust him, and all I can do is let the process unfold. If I thought there was some way people could help I'd say it.
Eleven posts up from yours is this from Aurich:I'm not going to continue to pay Ars so that it can be an accessory to monetization without consent.
Just to keep making this part clear: nobody is making any money off of comments. We are not selling them to OpenAI. Ken has specifically stated this.
Okay, I will concede this point, because it's certainly true.No, not same thing, today it is possible to do things with those posts that was impossible (and unforeseen by many if not most users), there is a difference.
I think the very least people should do is give it some time. This announcement only happened hours ago and there is a (slim) chance that comments may be eventually excluded. The site owners have already said that is what they’ll ask for.I don't like this. Haven't decided what I'm going to do about my subscription yet, since I get that it wasn't Ars' decision to make
That's what XenForo does, the old forums did not.I just thought of a way to do it. The quotes are tagged with the username so that narrows it down. That's how you can delete the quoted posts, @Aurich
So it's not as hard as I was initially thinking. I am sure somebody here can write a script for you to not only delete user posts but the quotes of those posts.
This is my last post for now, it's after 6pm, I need dinner, and I just don't think I can do anything to really help much beyond what I've said.I think the very least people should do is give it some time. This announcement only happened hours ago and there is a (slim) chance that comments may be eventually excluded. The site owners have already said that is what they’ll ask for.
One ramification of this new deal is that OpenAI’s web crawlers are no longer excluded via robots.txt. With the robots.txt exclusion for OpenAI now gone, the startup is free to crawl any CN property, including Ars Technica. This means that, once again, OpenAI can crawl any part of the site that does not require a login to view, including user comments. To be clear, user comments were being crawled before they were blocked, but now, after the 11-month hiatus, they will be crawled again. Between publisher deals, the voluntary nature of robots.txt compliance, and the hordes of pirated data out there, it seems as though the only reliable way to escape the crawlers (Google, OpenAI, Perplexity, Microsoft, ad infinitum) is not to participate—a pyrrhic option at best.
I didn't cancel my subscription because I'm mad about an AI potentially slurping my posts, for me it's bigger than that. I cancelled my subscription because this demonstrates to me that the parent company willingly harms itself with these short sighted decisions. Someone who only thinks in dollars and cents might argue that you have to do everything and anything to secure funding at all times but I fundamentally disagree.Excellent 2nd post!
I do think cancelling subscriptions will only have the net effect of harming this site even further. I certainly don’t think it will do squat to reverse or change anything regarding the AI deal. A cancelled subscription might then be misguided, as good as the intention of making an ethical statement may be.
This is my last post for now, it's after 6pm, I need dinner, and I just don't think I can do anything to really help much beyond what I've said.
But my utterly personal guess, based on nothing but feelings, that I cannot promise anything from? I think it's the opposite of slim. I don't think OpenAI cares at all about our user comments, and is purely interested in authoritative article content, and figuring out how they're going to move into a future where the next NY Times isn't suing them.
I could be full of shit. I have zero knowledge of the deal, I'm not involved, I cannot speak to it even if I wanted to. And if I was involved I'd probably have to say even less honestly.
Pure conjecture and opinion time!
I really think this is an oversight that could have been avoided, was lost in the shuffle, and we're going to fix it and still pay the price for it. And that's why I'm going to drink this canned Mai Tai that's more alcohol than delicious and try and take the rest of the night off.
For anyone who's willing to be patient and give us a chance, I hope I'm not in fact full of shit, and we can earn you back. I have a lot of new subscription features I'm not feeling nearly as excited about rolling out right now, but I sure would like to have a chance to get people interested in them.
This is why I don't bother. My words have already trained models. They will in the future. The tech is out there.We just erase the internet? Go dark? Kill history? What?
And probably most people either weren't aware of it or didn't like the idea. I agree with you that deleting posts won't do anything to OpenAI or whoever because it's already out there but people should still have the choice.keeping our posts intact is an important value to us, is always has been
Thank you. I was at work and didn't have time to look through hundreds of comments.Feels like Ken or somebody should make a featured comment laying out that;
-Ars isn't getting a cut of the money from this deal
-Subscriptions help Ars, not CN
-Ars isn't going to syndicate AI content
-the Ars team is working in the background on what can be done to limit exposure of forum comments to OpenAI (at least that's how I've taken Aurich's comments elsewhere asking for people to give you guys time to work out the details). Confirmed by Ken on attempting to update the robost.txt to block /civis/