12:02 AM
@F1Krazy I wasn't detailed enough, and perhaps one explanation can't cover all SEs. Here is a real Writing.SE question. "How do I write in a sophisticated manner (though not convoluted)?" Suppose the questioner has a plan, "Paraphrase sophisticated stories", and wants a critique of their plan. Would it be appropriate for the questioner to ask the question and then provide an answer, "Paraphrase sophisticated stories"? — Hall Livingston 30 secs ago
On the compensation: does anyone else interpret the press release as "OAI get SO's data in exchange for SO agreeing to beta test OAI features"? Doesn't really seem like a balanced trade. — AShelly 48 secs ago
12:51 AM
@Catija I read this as "Indeed sucks" -- here's how a platform that does not suck should work. So if Stack is moving to Indeed, it should still allow Stack-like manipulation of the data. Then it will suck less than Indeed. The last paragraph is the key. — mdfst13 36 secs ago
1:07 AM
The instructions regarding what to do to make the post on-topic that appear in the close-banner seem clear-enough to myself. If you need further clarification, then I suggest you post a query on the per-site meta: maths meta. Please be aware that there is a sandbox you can post in "for long and complex posts": maths sandbox. They will help you refine the post as needed. — W.O. 37 secs ago
1:30 AM
Please also remember that according to section 6 of the CC-BY-SA license, any breach of the license not cured within 30 days of you being informed of it will terminate the license. This would render displaying affected contributions on any of your sites and using your AI illegal, with or without attribution. I promise to that I will do everything in my power to see my copyright upheld and I urge all users to do the same. This kind of willful ignorance of copyright on part of SE should not be tolerated. — Kryomaani 33 secs ago
@AykhanHagverdili if they do not attribute each an every user whose contributions are used, they are in direct breach of the CC-BY-SA under which all contributed content has been licensed. It seems quite clear from how unrealistic fulfilling this is that SE is acting in bad faith and not intent on upholding the license in the first place. — Kryomaani 11 secs ago
1:48 AM
LLMs are pretty much the same as a lossy compression algorithm combined with a compiler, so the output naturally is a mechanically combined derivative of the input. This means all licences on all of the input must be honoured, which means they’d have to open up all the “training data” if they include copylefted works in them. — mirabilos 58 secs ago
Remember what was done before: forced relicensing - comprising (alleged) breach of contract with the userbase - nothing to stop them doing it again irrespective of our wishes or concerns. @Kryomaani — W.O. 50 secs ago
@mirabilos I'm more concerned with the situation where the trained model itself is ruled to be a derivative work. — bta 45 secs ago
2:03 AM
@W.O. well, that is sadly true, but it doesn't mean we should go down without a fight. The only logical course of action is to 1) Stop contributing and 2) Issue DMCA or other suitable process of copyright notice any time SE breaches the license of your contributions. I've already done 1 and will be doing 2 if this move goes through as advertised. — Kryomaani 20 secs ago
2:15 AM
@Piper what settings? what do they do? who are they surfaced to for changing? — super-starball-ultra 53 secs ago
2:26 AM
it is, of course; just because you can lossily JPEG-compress a picture doesn’t make the JPEG file not a derived work of the original picture, and you can decompress it and get a substantially similar result back, which people have proven for LLMs as well (by now, really substantial amounts of “training data”) — mirabilos 56 secs ago
2:54 AM
Well that's the other side of things - SE's acting as an upstream for other AI properties, and the threat there is what they do if that source of revenue dries up. — Journeyman Geek just now
Why would someone who's good with Pets be considered qualified to edit code, or edit vote on complex physics or mathematics problems? — W.O. 28 secs ago
1 hour later…
4:24 AM
4:37 AM
Hi Irfan, welcome to the Stack Exchange Network Meta site! I'm not sure which search brought you here but the problem you describe will not be answered on this specific site. To get an expert's answer for the topic of your question you'll have to find and then re-post on the proper site. Check How do I ask a good question and What is on topic on the target site to make sure your post is in good shape. Your question is definitely off-topic on Meta and is better deleted here. — Journeyman Geek 28 secs ago
4:48 AM
1 hour later…
5:56 AM
Hi Chickecode, welcome to the Stack Exchange Network Meta site! I'm not sure which search brought you here but the problem you describe will not be answered on this specific site. To get an expert's answer for the topic of your question you'll have to find and then re-post on the proper site. Check How do I ask a good question and What is on topic on the target site to make sure your post is in good shape. Your question is definitely off-topic on Meta and is better deleted here. — Journeyman Geek 43 secs ago
Why wouldn't it be an acceptable username? How is it profane or otherwise violating the Code of Conduct? Even if you drop "fsck" into Google, you get fsck (file system consistency check) on Wikipedia as the first result, and nothing that looks inappropriate in at least the first dozens of results. What actually prompted you to think that it might be a problem? — Makyen 14 secs ago
6:10 AM
Why wouldn't it be an acceptable username? How is it profane or otherwise violating the Code of Conduct? Even if you drop "fsck" into Google, you get fsck (file system consistency check) on Wikipedia as the first result, and nothing that looks inappropriate in at least the first couple/few dozen results. What actually prompted you to think that it might be a problem? — Makyen 59 secs ago
@Makyen I imagine they might think it would be a problem because it is one letter off from a problematic word. I don't agree with that idea though. — Daedalus 49 secs ago
It is, of course, possible that someone tries to use any text as a stand-in for a problematic four letter word. If it is inappropriate would depend on the surrounding circumstances. If we are going to say that it's a problem just because there's one character difference, then we need to ban "buck", "duck", "luck", "muck", etc. Now, that doesn't mean that all sequences of four characters which are a single letter off from that word are OK, as there are some that are very clearly intended as profane/inappropriate, and others that are clearly intended that way based on usage. — Makyen 38 secs ago
Everyone can flag, unless you've been suspended from doing so for being "naughty". @roundabout — W.O. 8 secs ago
Somewhat related: Why do I need 50 reputation to comment? What can I do instead?. Read the second paragraph in the answer. — HolyBlackCat 39 secs ago
"Will this abide by the CC by-SA license that posts are under?" Just as a comment. It may not need this. The posts may also be under a more permissive license for the company as written in the TOS. — NoDataDumpNoContribution 10 secs ago
Good question. From past experience I would not expect many answers anytime soon. But nice that you asked. — NoDataDumpNoContribution 12 secs ago
A flippant answer would be "engage more on the site or get 200 reputation somewhere. A more serious answer is I'm unsure what's the right way for a "low rep" user who has almost no reputation on a site to do more advanced meta moderation tasks. Quality control very much is a meta-moderation task that expects a certain level of platform and subject matter expertise. — Journeyman Geek 39 secs ago
The answer is not relevant to the actual question, sadly. (OP with 1 rep only, can't even flag.) — Shadow Wizard Love Zelda 23 secs ago
"There aren’t specific details yet because the work is just starting..." It would be very nice if more specific details could be shared when the work has progressed more, just to be more assured. — NoDataDumpNoContribution 49 secs ago
@mirabilos I agree with the technical description although there may be an argument that such LLMs compress so strongly that maybe only a tiny amount is taken from each work and that may make a difference. Anyway the courts or the legislation may well decide one way or the other. We should wait for a few of such court cases to see how it's seen legally. — NoDataDumpNoContribution 22 secs ago
OP never said that tho- and I'd argue this is a perfect opportunity to do what I said into action :D. What should an OP with 1 rep only do? — Journeyman Geek 53 secs ago
7:11 AM
Well OP didn't explicitly said something wrong either, it's case of not enough details leading to confusion and misleading question. But your previous comment is actually a fitting answer to the question. — Shadow Wizard Love Zelda 18 secs ago
7:28 AM
7:46 AM
As also mentioned in a comment in the announcement post, it is not unlikely that this will be a form of Retrieval-augmented Generation, in which "In response to a query, a document retriever selects the most relevant documents (= some SE questions)", and then "Following document retrieval, the LLM generates an output that incorporates information from both the query and the retrieved documents". This means user attribution and deletion would work exactly the same as on SE currently. — Marijn 6 secs ago
@Kryomaani you need to attribute every word of the english language you have learned, where did you see it, which english books did you read, every single time you type, just attribute every single thing you have ever seen. — Rainb 7 secs ago
@Rainb Because money. "Partnerships are another revenue stream for us," — S.L. Barth is on codidact.com 36 secs ago
Hi Elvis Kibet, welcome to the Stack Exchange Network Meta site! I'm not sure which search brought you here but the problem you describe will not be answered on this specific site. To get an expert's answer for the topic of your question you'll have to find and then re-post on the proper site. Check How do I ask a good question and What is on topic on the target site to make sure your post is in good shape. Your question is definitely off-topic on Meta and is better deleted here. — Journeyman Geek 50 secs ago
8:24 AM
Hi caron polwar, welcome to the Stack Exchange Network Meta site! I'm not sure which search brought you here but the problem you describe will not be answered on this specific site. To get an expert's answer for the topic of your question you'll have to find and then re-post on the proper site. Check How do I ask a good question and What is on topic on the target site to make sure your post is in good shape. Your question is definitely off-topic on Meta and is better deleted here. — Journeyman Geek 12 secs ago
8:39 AM
A deeper Google search into the etymology of the initialism claims the oldest known use was in 1983 (and relates to IBM, no less!) In short, a 30 y.o. from 1983 lines up pretty much with being in the central age of a "Boomer"... Me? It was only last week I found out what FGITW means... Social circles... — Fe2O3 5 secs ago
"There aren’t specific details yet because the work is just starting...". It is pretty scary (to say the least) that this kind of core questions were not answered prior to any agreements and any announcements. When such a dividing topic is announced, I think we could expect to have some FAQ with explicit answers about such obvious issues for SO/SE users — Double Sept 1 min ago
9:29 AM
"Why is it not known who closed your question?" The same reason it's not known who downvoted it - to prevent those users from being subjected to harassment or abuse as a result. — F1Krazy 43 secs ago
9:39 AM
What could possibly go wrong? Dear Stack Overflow denizens, thanks for helping train OpenAI's billion-dollar LLMs. Seems that many have been drinking the AI koolaid or mixing psychedelics into their happy tea. So much for being part about being a "community", seems that was just happy talk for "being exploited to generate LLM training data..." The corrupting influence of the profit-motive is never far away. — David C. Rankin 17 secs ago
9:59 AM
"those updates will be coming in weeks, not months". Well, it's been over 8 weeks now. Did I miss an update? — Shadow Wizard Love Zelda 5 secs ago
10:46 AM
Problem is we heard the same assurances lots of times already for many of the past projects that crashed and burned. Another "we don't have anything to share and work is just starting but don't worry, it will be totally fine when the feature is rolled out" doesn't cut it at this point. I'd bet a full years' salary that this will not be delivered in a way that satisfies the community, and probably not even in a way that satisfies the letter of the license. — l4mpi 32 secs ago
@dan1st nope, from the way of writing sounds like Rosie actually meant it literally, making us hope it will really be under 8 weeks. — Shadow Wizard Love Zelda 53 secs ago
And also, the "socially responsible AI" thing is a bad joke. What's supposed to be responsible about wasting tons of energy and resources on an inaccurate text generator that (assuming this will be based on GPT-4 or later) has a foundation stemming from egregious data slurping and copyright abuse, and includes all manner of problematic content in its training data (racism, conspiracy nutters, etc). I'd take you more seriously if you were talking about socially responsible cannibalistic meth addicts. — l4mpi 36 secs ago
Re "which completely misinterpreted it ... also very short and lacks explanation": That is consistent with blind plagiarism (though not much is known about it) — This_is_NOT_a_forum 10 secs ago
What site? You currently have 123 reputation points on Stack Overflow. — This_is_NOT_a_forum 38 secs ago
11:11 AM
So attribution for the training data should be given to the person who posted an AI-generated answer, which was obtained from training the AI on data for which attribution should be given-... Why doesn't OpenAI see that this is a loop that only serves to make their AI increasingly dumber with every iteration? The company was first insisting that AI generated posts are good for the site, only stepping back after massive criticism. And now you are trying to sell access to the data which the company itself has actively encouraged to get polluted with AI content. What's the longterm plan here? — Lundin 9 secs ago
11:27 AM
@Lundin To let the Community filter out all the AI
*** (insert whatever you want there, e.g. posts) and get money for it? — dan1st 34 secs ago11:43 AM
@dan1st That cannot have been the company's longterm plan since they first opposed such moderation and tried to make it difficult. Now of course, it seems quite likely that the company has no longterm plans at all. These kind of announcements rather suggest that this is indeed the case. — Lundin 5 secs ago
11:58 AM
@Lundin The CEO increasingly reminds me of the Joker. "Do I really look like a guy with a plan? You know what I am? I'm a CEO chasing hype." — S.L. Barth is on codidact.com 15 secs ago
12:15 PM
This question should probably have at least one of these tags: [generative-ai], [ai-tools], [overflowai] (I’m not sure which.) — Elements In Space 35 secs ago
I’ve added the tag [generative-ai], but maybe [ai-tools] or [overflowai] would be more appropriate. — Elements In Space 41 secs ago
1 hour later…
1:20 PM
The argument from OpenAI might be that a model should be treated the same as a person that re-expresses the ideas in an SE post, but doesn't reproduce or remix the content. In that sense, the data and the model may need to be relicensed, bif they were ever distributed. However, the only thing OpenAI publishes is the model's answers, and they only need to be CC-BY-SA if they overlap enough with the training data to count as derivative. — Peter 45 secs ago
All of this is pretty orthogonal to the deal above, though, since the license applies just the same whether OpenAI scrapes the data from the website or gets it directly from SE. — Peter 35 secs ago
Note that posting AI-generated content is still banned regardless of this policy. — Nzall 58 secs ago
"UserIds are also null for deleted posts, for obvious reasons" - I wonder, what are the obvious reasons?
OwnerUserId would be nice to have. As Kelly Bundy mentioned just above, deleted posts remain _fully accessible for non-moderator 10k+ users. — avm23 23 secs ago1:45 PM
@rene Ah, so maybe not that hypothetical. Still, I guess the questions stand. If we all move to codidact, how much of a tantrum could StackExchange throw? — Peter 11 secs ago
Well, we have at least one founder there that is well-seasoned when it comes to tantrums thrown by SE. I assume that fork has their stuff sorted, even when it comes to legal stuff. — rene 7 secs ago
@PresidentJamesK.Polk not by default but members that moved their seeded the site with their SE posts — rene 42 secs ago
There isn't an "s" in <censored>. How can you come to that conclusion? I have never seen a substitution of <censored> or <censored>ing with an "s" in it. — This_is_NOT_a_forum 53 secs ago
I hate this. I'm just going to delete/deface my answers one by one. I don't care if this is against your silly policies, because as this announcement shows, your policies can change at a whim without prior consultation of your stakeholders. You don't care about your users, I don't care about you. — henning 38 secs ago
There isn't an "s" in <censored>. And (real) typos are unlikely as it requires at least two keys over. How can you come to that conclusion? I have only seen one substitution of <censored> or <censored>ing with an "s" in it. I don't remember the context, but it could very well have been in a Unix fsck context (thus very contrived and rare). — This_is_NOT_a_forum 25 secs ago
considering SE essentially copied the format from EE and started from scratch, I'd argue the trick wouldn't be moving content, it would be effectively moving over and supporting communities, both ones that are currently still on the platform, and diaspora ones. — Journeyman Geek 1 min ago
@VLAZ As I said in my question, I would be much, much more sure that the "for 5 minutes" is wrong (and hence I marked this as a bug). I have heard about "4 minutes long edit window" for ages, ever since I started using SE. It was never ever even a mention about "5 minutes". — trejder 9 secs ago
No need to deface anything. Just admit that all your contributions were made with the help of a GenAI. Then all your posts will get deleted. — Lundin 55 secs ago
I'd like to see your work reposted on Codidact. They don't have a site for Academia yet, but you could suggest one and post your work as samples. Also - if you delete your posts here, do it slowly. Otherwise you'll get a suspension to stop you. — S.L. Barth is on codidact.com 42 secs ago
If -- as you have finally corrected me -- it is "5 minutes edit window" then why we cannot have a 2-3 lines of additional client-side code that would trigger comment's timestamp field to be updated right after displaying "Comments may only be edited for 5 minutes" message? That wouldn't cost much (in both developing it and adding some load to the server) and would shave off all the doubts and misunderstandings, as per my opinion. — trejder 30 secs ago
2:45 PM
"they used a lot of profane words" Not sure if there might be other revisions before the migration, but looking at the history of the post, there only seems to have been one problematic word. Not sure if that counts as "a lot of" ... — samcarter_is_at_topanswers.xyz 18 secs ago
You'd be putting most of the strain on the community mods rather than the staff or the gen AI providers. There's probably better ways to remove all your content from the network than that if it came to it - but its probably better to have a considered reaction, backing up your posts and such, than threatening to delete all your posts and such. I'd note defacement/deletion dosn't actually remove SE or potentially others access to your post either. — Journeyman Geek 41 secs ago
Yeah, there is many low-hanging fruit to implement by the dev team. Unfortunately they are not given the time by top level management to spend time on these type of UX improvements. — rene 34 secs ago
Vandalism is usually reversed. If you no longer want to be associated with the content, there are better paths to go. — Mast 37 secs ago
Porting the content to Codidact would actually be viable, although mass imports from SE were attempted and didn't really end up well. A proposal for a new community would have to be made in this case, but once that is done, there are ways to import SE content. You could pick your best posts to serve as example questions for the new community. Although since this is getting off-topic, a separate question about it on the appropriate Codidact meta would be the best place to continue that discussion. — Lundin 53 secs ago
currently, there's a top N% of the site/year link on profiles linking back to the leagues. For 'lower' ranked users, where would it go? — Journeyman Geek 17 secs ago
"the moderator teams were not given any prior notice of any of this" - this seems a violation of the agreement that was reached less than a year ago. — S.L. Barth is on codidact.com 17 secs ago
I'm going with this one tomorrow, and will rename the imgur tag to imgur-image-hosting and update it's guidance a bit to mention it is obsolete now — Tinkeringbell ♦ 13 secs ago
@S.L.Barthisoncodidact.com Lots and lots of empty words, who would have guessed. Unless it hurts the company financially (and they understand that it this is the case), then nothing will change. There's no reasoning with them, everyone should have realized that at the 99th attempt. How the board can sit passive and watch their flagship product getting dragged in the dirt over and over again, I just don't understand. All other companies in the world tend to cherish their flagship product. — Lundin 18 secs ago
@Lundin To be fair, I can understand there's a dilemma. Negotiating a partnership is likely a business secret until the contracts are signed. Informing the moderators ahead of time would make this difficult. Not sure what is the right solution here.... — S.L. Barth is on codidact.com 7 secs ago
I don't understand point 3. Does it mean that when a post contains an URL to imgur, only a simple
REPLACE(body, 'i.stack.imgur.com', 'i.sstatic.net') is done? — rene 6 secs agoWell, its out now. I'm seeing a fair amount of confusion, so I'd say as much as a press release plan, having one to provide us with basic information, and bidirectional handling of social media might be handy. Essentially, giving us the tools to help, and not needing us to. — Journeyman Geek 39 secs ago
@rene We are checking to make sure that URL has an image ID associated with it. So it won't replace
https://i.stack.imgur.com/, but it will replace https://i.stack.imgur.com/<image ID>.<extension>. The ID has to look like a "real" image ID, too. So it's a hair more nuanced than a simple find-and-replace on i.stack.imgur.com. What we're not checking, however, is whether the associated image ID actually points to a real image before running the migration. If the image ID looks "real" but doesn't actually point to an image, it'll be converted anyway, as though an image exists there. — Slate ♦ 32 secs ago@JourneymanGeek That link will only be displayed in the user profile if that user is ranked on the leagues, e.g. in the top 400 users. (We'll also update this link so the text displayed stays accurate.) — Slate ♦ 47 secs ago
@rene It's a hair more nuanced than a simple find-and-replace. A post to qualifies for the URL update if it has a
https://i.stack.imgur.com/<image ID>.<extension> URL anywhere within it. The image ID has to match the format of a "real" image ID, too. If the post contains any matching Imgur image URL in it, then yes, it's essentially a find-and-replace on //i.stack.imgur.com/ (slashes included). If the image ID looks "real" but doesn't actually point to an image, the post will still qualify for updating. Therefore, in some cases, a bare reference to i.stack.imgur.com might change. — Slate ♦ 14 secs agoIt seems like this would be good as a synonym: image-hosting --> stack-exchange-image-hosting. — Makyen 34 secs ago
@henning Please don't do that; you'd only be making more work for the folks around you that also care about the site(s) you contribute to. The company did a spectacularly horrendous job of announcing this, whatever it's supposed to be, but we truly have zero information at all about what this will practically mean. I think we should get a little bit more of an idea of concrete impact before anyone considers nuclear options. — zcoop98 1 min ago
4:02 PM
@Magisch Interestingly, for some users this will underestimate reputation, not overestimate. E.g. your query with
site=math, UserId=460999, UntilDate=2021-01-15 returns -586, whereas the activity tab graph shows 31 on that date. What might be the reasons for that? The user might've earned reputation from deleted posts. Are there any other possibilities? Refunded downvotes for deleted posts wouldn't enter the equation in the first place. — avm23 10 secs ago@ShadowWizardLoveZelda I think this was after the omission of the 6-8 weeks joke was noticed in a comment: meta.stackexchange.com/questions/398127/… — ColleenV 40 secs ago
@samcarter_is_at_topanswers.xyz - That's just what I thought... just one example, and even that example was self-censored (with an asterisk). — Greenonline 9 secs ago
Okay, that sounds less bad then I imagined it. I expect we can deal with the few mishaps that occur due to how these replaces are done. Thanks for clarifying. — rene 26 secs ago
Very much appreciate this feedback in terms of what is most valuable to you on a job site. I can't make any promises about what functionality will be included in future iterations of SO Jobs, but we will consider the ideas presented here. — Sasha ♦ 6 secs ago
@user1937198 That sounds entirely unenforceable. The data is freely copyable by anyone, and regardless of whether you sell it or let someone else copy it of their own volition, once that's done, you no longer control it. That's not how reality works. — TylerH 47 secs ago
4:44 PM
"if Stack Exchange ends up liable for license violations". Which license? — Franck Dernoncourt 27 secs ago
4:55 PM
If that were the intention of the ToS, it wouldn't say “as reasonably necessary to”. The only thing they've listed that comes close to this is “Aggregate data to provide product optimization”, and "training a transformer model" clearly doesn't fall under that. Besides, they don't have the right to sublicense except under CC BY-SA 4.0. — wizzwizz4 34 secs ago
This remains an open question. Some people are convinced there is a dual license. I believe that it is an enumeration of the minimum rights that SE needs to function which are currently covered by CC BY-SA 4.0. Unless SE legal clarifies or someone tests it in court, we may never know for certainty if there's one license or two. — Thomas Owens 35 secs ago
Once you have 20 network-wide rep, which you currently do, you could try going to one of the chat rooms and mentioning the answer there. — CPlus 1 min ago
5:23 PM
So instead of improving the queries and making them faster, you choose the quick way of just cutting this in 200000% (yes it's accurate based on the 800000 users you said here yourself). OK. — Shadow Wizard Love Zelda 9 secs ago
5:41 PM
@TylerH Its enforceable because the law makes a distinction between where a data controller has made data publicly available and someone has used it, in which case the burden is 'reasonable steps taking into account cost of implementation', and in case where they have privately transferred data to an entity they have a direct relationship with, where the standard is 'obligation to erase personal data without undue delay'. — user1937198 42 secs ago
@TylerH Or in other words, if an SO publish the data on the Q&A site, and an LLM manufacturer comes along and slurps it, then SO need to stop serving the data when its deleted, but ultimately, they don't know who slurped it. If SO sells the data to the LLM provider directly, then they need to ensure that that data will be used in a way that is consistent with any deletion requests, including any received after the sale. — user1937198 30 secs ago
Hi caron polwar, welcome to Meta! I'm not sure which search brought you here but the problem you describe will not be answered on this specific site. To get an answer from users that have the expertise about the topic of your question you'll have to find and then re-post on the proper site. Check How do I ask a good question and What is on topic on the target site to make sure your post is in good shape. Your question is definitely off-topic on Meta and is better deleted here. — Sam Onela 26 secs ago
@TylerH The issue legally is not that LLMs are using data from the network. Its that SO is directly partnering with them, that creates the burden on SO to have the data deleted. — user1937198 15 secs ago
@ShadowWizardLoveZelda There aren't that many ways of making a query like this faster, and most of them would likely be a very large amount of work. I don't see how the reputation leagues would be worth that kind of effort — Mad Scientist 18 secs ago
What %’s are we expecting this to result in on SO? For example, for a lot of currently active users they may see today “You’re in the top 5%!” If thst % is still based on all users, not just the top 400, where can we expect the cut off to be on a site the size of SO? — Kevin B 13 secs ago
6:05 PM
@Shadow Mad Scientist hit the nail on the head honestly. We have to weigh the value of the feature against the work it takes to maintain. In this case, fixing the feature in-place would be a very significant amount of work, but doubly so account of the existing backend work taking place. If folks strongly feel that the rep leagues need to be maintained as-is, I will of course pass that along as usual, but I'm inclined to feel that they can be safely shrunk. — Slate ♦ 16 secs ago
So, if on Stack Overflow, 2 of us lowly "flag voters" select "Not about programming" and 1 selects "Blatantly off topic", has consensus been reached? If not, is there intentionality in making it hard to flag stuff that is clearly off-topic? — rand'Chris 45 secs ago
6:18 PM
Flag for moderator attention. But moderators are already automatically notified about rollback or edit wars with a lot of edits or rollbacks in a short time. The mod can handle it from there. — Thomas Owens 41 secs ago
Flag a mod. Also on enough rollbacks on a post the system contacts a mod. Also the poster is defacing multiple posts. But please research before considering posting a question. — philipxy 38 secs ago
Does this answer your question? What to do when the OP mutilates their own question? — bobble 28 secs ago
Try Writing. Be sure to take their tour, read-up in their help centre before posting. We also have a Worldbuilding site for the creation of fictional worlds/creatures/physics etc. for your interest. — W.O. 40 secs ago
@Mast I don't care about dissociation, I want my contributions removed so they can't be misused. — henning 20 secs ago
7:19 PM
I don't believe that we've agreed to a second licence which permits SE (or their partners) to republish the content we created without giving adequate CC attribution. OTOH, I agree that the situation is legally murky, and meta.stackexchange.com/questions/388571/… — PM 2Ring 14 secs ago
Since you are convinced that that answer is wrong, that implies that you know what the right answer is. Or at least you understand the question and have some ideas regarding a correct answer. So do some research and write an answer. — PM 2Ring 44 secs ago
7:39 PM
While I agree that it's probably not a great use of dev time to try to fix things to optimize things enough to maintain the current functionality, could the number potentially be larger (say, 10k) without significant work required to further optimize? My concern is that at a certain scale (glances at Stack Overflow) the amount of reputation required to break into the top 400 becomes somewhat impractical (currently ~205k for the all-time league on Stack Overflow; I don't even have enough to quite get into the all-time top 10k). Of course, the time-based leagues do mitigate this somewhat. — Ryan M 44 secs ago
@user1937198 Then it sounds like I already covered that in my comment. If the AI tool just looks at code freshly every time there's a dump then it'll be fixed during the next dump. If it doesn't, that's not SO's problem, it's the AI tool's problem, because it's not SO's code or content anymore. — TylerH 13 secs ago
@henning -- of note the Terms of service give Stack Overflow license to use your post content, even if they are deleted. — AMtwo 35 secs ago
@Machavity RE your edit - I've moved most of its content to the chat privilege FAQ, which I've linked in my edit. The scope of this answer is documenting the tools that mods can use - I've moved the info of when they can use it to another answer where it's in scope. — Sonic the Anonymous Hedgehog 16 secs ago
7:58 PM
If you dive into the site Terms of Service, you'll see that "Stack Overflow [has] the perpetual and irrevocable right and license to ... distribute, export, display and to commercially exploit [Q&A] Content" ......ie) Stack can do whatever they want with Q&A data. There's really no legal basis for a user to sue the company over what they do with Q&A data. — AMtwo 26 secs ago
let's come back here in a couple of months and we will see. Glad to see your will to defend an ecocidal money making scheme. Good use of your time. — Joel Falcou 20 secs ago
8:28 PM
This is good news... so that the community can open its eyes and realize who is the one who works and who is the one who receives benefits... until today's sun... from the sales that the company has had and the associations... that promised a better environment for end users... I haven't seen anything... and even less on sites in other languages such as Spanish. — Francisco Nuñez IA Lover 20 secs ago
8:45 PM
@TylerH Its depends entirely on if the dump is provided by SO to the LLM vendor, vs downloaded by the LLM vendor from public sources. If its the first, the next dump is not good enough, SO have to proactively get the vendor to remove personal data when they receive deletion request. — user1937198 5 secs ago
9:14 PM
On the last collaboration SE explicitly stated that they are not provided the content under a different, non-CC license when I asked about that. I assumed this case would be the same, but it certainly would be useful to ask it explicitly for this case as well. — Mad Scientist 59 secs ago
Please don't post the same question on multiple Stack Exchange sites, This is even more off-topic here than it was on Stack Overflow. This is not a free code-writing service: if you want this program created, and you're not good enough to make it yourself, then hire a professional developer. — F1Krazy 46 secs ago
What site? What ads? What exactly you expect to hear with such a question? — Shadow Wizard Love Zelda 42 secs ago
@NoDataDumpNoContribution there has recently been work showing that, no, the compression is nowhere even remotely as heavy to reach that. But having read where someone reimplemented a slightly older ChatGPT in a 498-line PostgreSQL query, I now understand even better why this is so, and how the output is naturally derived from the inputs. — mirabilos 39 secs ago
Can you provide a screenshot with a red rectangle around the ads you are concerned about? — Ramhound 10 secs ago
9:52 PM
10:41 PM
@Rosie what do you mean with "keeping tabs" and "answering when you are able to"? I know it's only been a day but there is exactly zero answers from staff to any of the questions the community has raised, and there are quite a few. There's no point pretending there's a discussion if one of the parties doesn't even participate. — Kryomaani 38 secs ago
@Kryomaani there's this answer post, which a lot people suggested as editing into the question post. — super-starball-ultra 49 secs ago
11:07 PM
@SonictheAnonymousHedgehog and OP, as far as I could remember, there have always been those banner ads for anonymous and users under a certain reputation threshold. Until you reach that threshold you can disable those ads. Perhaps the OP was more accustomed to a site where they had that privilege. — Cave Johnson 23 secs ago
You only have 111 reputation in Ask Different so you don't have that privilege. — Cave Johnson just now
SE never did go public. If it did, having a community representative on the board might be a good counter balance , and I suspect we could have made it happen collectively — Journeyman Geek 59 secs ago
11:30 PM
@CaveJohnson Another case I've seen is if a user with the privilege gets suspended on such a site: due to their reputation being 1, they'll suddenly start seeing ads for the period of their suspension. — Sonic the Anonymous Hedgehog 32 secs ago
11:43 PM
« first day (349 days earlier) ← previous day next day → last day (12 days later) »
Transcript for
May6
May7
May8
♦ Meta Stack Exchange Comment Archive
A comprehensive archive of all Meta Stack Exchange Comments. C...
