User talk:Jimbo Wales

Mark is the PM of Canada now

I will only ask once, and you will not try to deflect this - do you think he will save us from those that want us to give in to their own interests? 2601AC47 (talk·contribs·my rights) Isn't a IP anon 00:33, 10 March 2025 (UTC)

I'll give a direct answer, if you promise not to tell me I'm not Jimbo (I already know that!) US-Canadian relations will be as bad as they ever have been over the next 3+ years, but Trump would never bomb Ottawa, nor will Canadians ever invite him in to be their dictator. Why do you think most of them live only a couple of hundred miles north of the USA, but don't just move to the USA? Because they would prefer to be Canadians rather than Americans. Smallbones_(smalltalk) 03:42, 10 March 2025 (UTC)

In my dream, it gets spicier and deadly: The US declares all-out no-holds-barred war against them, Ottawa is bombed out, millions in the crossfire and with little choice but to either fight or submit their surrender, and Canada is fallen and captured ceremonially with Trump calling it the "biggest thing we've ever done in the history of our country, not even my people didn't think we actually could". All in 1 week. Mark "the corny"… "Governor"… is put to trial over ordering power from his owning land to the US cut, found guilty without evidence or a legitimate chance of proving innocence, sentenced to lifetime imprisonment… And shot and killed by a Proud Boy.

So, Smallbones, what will it take to realize we're all but lost? Until Jimbo can answer that, I'll dream that he's forced to volunteer and face the very armed forces from which Jimbo once called home. 2601AC47 (talk·contribs·my rights) Isn't a IP anon 09:59, 10 March 2025 (UTC)

Is there a particular link between Mark Carney and Jimbo Wales that prompts this quite broad question? CMD (talk) 04:06, 10 March 2025 (UTC)

There is none, and I'm getting very close to asking 2601AC47 to go away from my talk page permanently for wasting people's time. I have no expertise and no opinion in this area, and see absolutely no reason to comment.--Jimbo Wales (talk) 09:36, 11 March 2025 (UTC)

Fine… But remember what happened to those that tried to help you steer things in the best way possible, only to be rebuffed or ignored. This can happen, and it's your great responsibility and trust on the line. 2601AC47 (talk·contribs·my rights) Isn't a IP anon 10:47, 11 March 2025 (UTC)

It's actually ironic, after looking through the history of your Talk. You speak of "wasting people's time", yet your own history is littered with instances that might be considered wrong. Perhaps you recall the unfortunate fate of a certain admin, blocked under circumstances that some might deem... Questionable? I've seen everything the last dozen years, Jimmy. Every statement, every action, every dismissal. It's all quite illuminating. And just to play that up, I'm sure that admin that was blocked would have loved to have been able to just declare "no opinion" when called to task, or tricked to it. 2601AC47 (talk·contribs·my rights) Isn't a IP anon 13:56, 11 March 2025 (UTC)

Reading this talk page now, if I were still an admin, I would probably block you. Please don't use me as an excuse for your childish antics. "I will only ask once, and you will not try to deflect this " is just not an acceptable way to address anyone, and it's just one example of everything you are doing wrong here. Fram (talk) 14:00, 11 March 2025 (UTC)

Would it help making it up to tell you that I just nominated Jlwoodwa for Adminship? 2601AC47 (talk·contribs·my rights) Isn't a IP anon 18:46, 11 March 2025 (UTC)

Not really, I have nominated it for deletion. I have no idea what you are trying to achieve by all this, but I guess it would be best if you just stopped it before you piss off even more people. Fram (talk) 19:37, 11 March 2025 (UTC)

I don't get why you're still prosecuting this, when the ex-admin in question made clear he wants nothing to do with it. MiasmaEternal☎ 21:04, 11 March 2025 (UTC)

Did you see Larry on Fox?

Larry on Fox A 3 minute video, the first half is not too important. But the 2nd half raises some interesting questions and suggests a very important conversation. I'd love to see that conversation happen. I'm not sure Wikipedia has the institutional framework, or even the technical bandwidth to host that conversation. Can you suggest anything that would get that conversation going? Smallbones_(smalltalk) 03:53, 10 March 2025 (UTC)

Forget him. 2601AC47 (talk·contribs·my rights) Isn't a IP anon 10:07, 10 March 2025 (UTC)

Let me guess, he's banging on about Wikipedia being left-wing, woke, or whatever epithet these people use now. He has been irrelevant for years. Black Kite (talk) 11:08, 10 March 2025 (UTC)

There's no need to guess - just watch the video. But a quick summary. He wants to know if the US and other governments employee people to edit Wikipedia. He wants to have a conversation about whether the US and others governments do, should they (I think he's agin it), and what to do about it. Smallbones_(smalltalk) 11:50, 10 March 2025 (UTC)

I don't know if it counts, but I read that Elon Musk donated 3 million dollars to WP. That was before he was in the government, though. Gråbergs Gråa Sång (talk) 12:37, 10 March 2025 (UTC)

Sanger really doesn't like it when Wikipedia articles don't agree with his version of reality, but I must admit it's a new idea that the Deep State is influencing the content of our articles. Black Kite (talk) 12:44, 10 March 2025 (UTC)

Jimbo, you and Larry Sanger could use a sweat lodge together or something, please heal any personal animosity if it exists. You and he are destined to be on a statue together somewhere, or a well-designed postage stamp, so put in a half-hour shouting at each other privately and let's get you both on stage at the same time at the 25th anniversary-year Paris conference (if not before). Thanks. Randy Kryn (talk) 12:20, 10 March 2025 (UTC)

Good idea, but IMHO it should be a broader conversation - at least 4 people. Heck, invite Elon if he can get a visa. And overall, I think Larry was hoping for a much bigger, very public conversation. Is there a website that could host a big in-person meeting with "audience participation" gizmos and maybe 1,000 virtual invitees? That could be a broad, but orderly meeting that could come up with some actual recommendations. Maybe TED could pull off something like that, or maybe that Irish tech organization that Katherine Maher chaired for awhile. It really couldn't just be on Wikipedia, we'd need broader participation. Smallbones_(smalltalk) 14:59, 10 March 2025 (UTC)

Since you asked for recommendations, the only recommendation I thought of awhile back was that since Sanger is very much an advocate of early childhood reading and writing (I get it, as my sainted mom had me reading and writing by the time I got to kindergarten), and since that fits right in with the concept of the world's knowledge being available free to anyone (they have to know how to read to actually accept that gift), maybe Sanger could be asked to chair some kind of WMF monetary distribution to increase worldwide very young child literacy. Alongside a WMF program having the same goal? They both might be able to team up on something like that. Randy Kryn (talk) 15:52, 10 March 2025 (UTC)

I may be too pessimistic but I think that Larry Sanger has gone so far off the rails that any useful collaboration with him is highly unlikely. Cullen328 (talk) 09:45, 11 March 2025 (UTC)

If a project would greatly benefit early childhood reading and writing I have faith that our fellow Wikipedian would at least listen. It rises above political views. Randy Kryn (talk) 15:55, 11 March 2025 (UTC)

This is not a matter of political differences. The fact of the matter is that Sanger has been demonstrably and repeatedly and spectacularly wrong about online encyclopedias for decades, and is now engaging in hallucinatory, evidence free attacks on Wikipedia. He is encouraging an increasingly authoritarian regime to "investigate" the project, which is a euphemism for industrial strength harassment. Civility is not a suicide pact. Cullen328 (talk) 22:36, 11 March 2025 (UTC)

I think most here are happy to see an Wikipedia:Academic bias because most of the sources used in non-popculture articles are academic in nature.

*Jimbo said years ago "The type of people who were drawn to writing an encyclopedia for fun tend to be pretty smart people."

• Geiger, Abigail (April 26, 2016). "A Wider Ideological Gap Between More and Less Educated Adults". Pew Research Center.
• Scott, Dr Ralph (August 13, 2024). "Why are graduates more socially liberal?". UK in a changing Europe.

Moxy🍁 01:19, 12 March 2025 (UTC)

I made a thing that some may find interesting

I wrote a script which I can feed a Wikipedia url and then it does the following: fetch the article, fetch all the external sources, get the text of those sources, and then feed it to a large language model which is asked two questions, first about whether there's information in the sources which should be in the Wikipedia entry, and second about whether there's information in the article which is not supported by the sources.

It is not well tested! I'm just sharing information here. I've put a first example on the talk page for Esther Meynell. Jimbo Wales (talk) 20:35, 13 March 2025 (UTC)

It's an interesting idea. Do you know if the llm was able to access the cited source for the Yorkshire Quaker sentence [1], or did it only hit the login/subscribe page? CMD (talk) 03:38, 14 March 2025 (UTC)

I'm virtually certain that it did not. What it did was to find the page you linked to, which is just a paywall page. My guess is that one of two things happened here - either the human who wrote our page did have access, or that the combination of someone being from Yorkshire, and being a Quaker, meant that they were described as a Yorkshire Quaker. I don't know enough in this current instance to know if that's potentially a major error, for example if Yorkshire Quaker is a particular group with different beliefs or practices from other Quakers.

If we look at this first stab at using an llm for an initial review, I would say it did a pretty good job on this point - it would ideally say a bit more, but it is true to say that one can't just click on the link and confirm the information.

I thought it also interesting that it quibbled over the idea that the house has been extended since - the only source is an airbnb listing which doesn't really say when the extension happened. But a human Wikipedian would likely say (as I'm saying now!) that an airbnb listing isn't really a great source in the first place.--Jimbo Wales (talk) 11:38, 14 March 2025 (UTC)

For the purposes of this test it isn't important what happened to create the text. It's that ideally if an llm hit a page like that (and I don't know what they might have access to, hence my inquiry), it was marked among the sources it couldn't access rather than the sources it could. There's probably a lot of edge cases where it might trip up, but a paywall page feels like something an llm should be able to pattern recognise. CMD (talk) 14:14, 14 March 2025 (UTC)

Yes, I agree.

Basically my script just loads the page and then loads all the (non-internal) links. In some cases I can tell there's a problem right there - 404 not found error for example. There's also in theory a meta tag which tells if a page is paywalled but I don't think it's widely used. [2] My script originally tried to detect paywalls by looking for some basic keywords like "subscribe" but that wasn't useful at all. However, you have a good point - a decent llm ought to be able to (somewhat imperfectly) notice that a page is not the actual source but rather a description of a likely paywalled source. That'd be good to report back, I'll play with that. Very good idea thank you!--Jimbo Wales (talk) 14:31, 14 March 2025 (UTC)

Update: I think I know why that meta tag is not much used - it's not even about paywalls I think. I suspect it's useless to even look for it in my context. Jimbo Wales (talk) 14:32, 14 March 2025 (UTC)

I believe past attempts at this have found even when given the text of the references, LLMs are too unreliable for this. Happy to be proven wrong though. — Qwerfjkltalk 11:45, 14 March 2025 (UTC)

Can you point me to past attempts? Basically my view is that LLMs are improving fairly quickly but still have very serious hallucination problems. However, when given specific texts to compare, they don't tend to simply make things up. Also, the development of a tool (this is just a quick script I threw together to start exploring ideas, not a tool, and please keep in mind this is my own personal project, not a WMF initiative or anything - I'm just a geek who likes to learn!)... the development of a tool to make suggestions doesn't have to be perfect, it only has to be good enough to be useful. If a tool suggested edits that we rejected 90% of the time, it'd be a waste of time to try to use. If a tool suggests edits that we accept 90% of the time, that's much more interesting. Jimbo Wales (talk) 11:51, 14 March 2025 (UTC)

New example: Takayuki Takayasu talk page with suggestions. It's interesting, because it makes a suggestion about a date that seems valid to me, at least initially, but the link (our link, which it looked at) doesn't seem to be working properly for me, and it doesn't say that. But even though I can't see the content of the press release, our own date for the press release, plus the text of the article, does suggest that the table has an error (likely a typo).Jimbo Wales (talk) 12:07, 14 March 2025 (UTC)

I'm fairly sure I saw it on one of the village pumps, and Wikipedia:Village pump (idea lab) makes most sense, but searching through the archives there didn't yield anything. So I'm afraid I can't. — Qwerfjkltalk 12:18, 14 March 2025 (UTC)

That's cool, I appreciate you trying. I'm sure when other people see this, they'll chime in with what they know. — Preceding unsigned comment added by Jimbo Wales (talk • contribs) 12:29, 14 March 2025 (UTC)

Very interesting! By the way (you were probably going to do this anyway, but just in case): It would be great to document the specific model and prompt(s) used in the script, as these choices can make a big difference for how well such efforts work.

It may be of interest that there are two academic research projects ongoing right now which try to do somewhat related things by developing custom AI systems. Both are currently recruiting Wikipedia editors to test and provide feedback on the tools they are developing:

• m:Research:Wikipedia Inconsistency Detection from Stanford's OVAL group (which previously gave us STORM - which btw ran into limitations with paywalls too -, WikiChat, and SPINACH). They have made a browser extension which highlights statements in a Wikipedia article that are inconsistent with information elsewhere on Wikipedia.
• m:Research:AI-Assisted Wikipedia Updating (The Blender Lab at University of Illinois Urbana-Champaign (UIUC) is developing a research tool to help Wikipedia editors identify and add information across articles, by monitoring news coverage)

Researchers from both groups attended the San Francisco meetup in person yesterday and have presented there about these projects (see notes, with link to a slide deck).

Also a general note that some of us are trying to keep m:Artificial intelligence updated with links to relevant policies, discussions, tools etc. on Wikimedia projects. The "Wikimedia AI" Telegram group currently seems to be the most active movement-wide venue for discussing such topics (I mentioned your experiment there already).

Regards, HaeB (talk) 06:09, 15 March 2025 (UTC)

Technical debt

Hi Jimbo! People say they don't even bother reporting problems to the WMF anymore because they believe the bugreports will be ignored anyway. What percentage of the money of the WMF is used to squash bugs? Can we please significantly increase that number, whatever it is?

Can you tell the WMF to spend a bunch of money to hire a bunch of nerds to work on Phabricator tickets? Wikipedians appear to be unable to contact the WMF.

Can we pause shiny new projects, and prioritize working on existing problems and technical debt instead? Thanks, Polygnotus (talk) 21:46, 14 March 2025 (UTC)

Can you point me to specific examples? Basically it's like super duper easy to contact the WMF, so I'm not sure what you mean. But if you're experiencing that, and especially if lots of people are, then yeah, let's roll up our sleeves and solve that.--Jimbo Wales (talk) 23:47, 14 March 2025 (UTC)

Communication: It is not easy for someone like myself to figure out which WMF-er to contact (but maybe there is some trick I have not discovered yet?). What do they use to communicate internally? Maybe I can contact them there? Or maybe that can alert them of pings on Wikipedia? For example if they use Slack to communicate internally we could build a Slack bot that notifies them if they get pinged.

Technical debt: I haven't made a list, and my memory is far from perfect, but I have heard that sentiment expressed several times, which is rather discouraging. Focusing on specific examples may detract from the truism that any website that exists for literal decades(!) builds up technical debt over time, and that it is good to focus on that once in a while. For example the Action API is missing a bunch of features (e.g. those in XTools).

I have some ideas that I believe may improve Wikipedia, but I am not sure how to reach the right person in the WMF who is willing and able to build something like that.

One idea was making it easy to add extra buttons to DiscussionTools.

Another idea was an alternative to the conventional talkpage notification Polygnotus (talk) 00:43, 15 March 2025 (UTC)

A dozen years in and countless dollars spent, the mobile site is still not fully functional and is an impediment to collaborative editing. I am a highly active editor and administrator who does 99% of my editing on smartphones, and I use the misnamed "desktop" site on my phone. It works just fine, which is ironic since the mobile site doesn't. It would be funny if it wasn't so tragic that we are losing out on potential editors because of the "I can't hear you" problem. Cullen328 (talk) 17:44, 15 March 2025 (UTC)

@Cullen328 I am trying to get Jimbo to agree with me that it would be a good idea to hire some more nerds. If there is too much negativity in this section Jimbo is less likely to agree with me. So there is a fine balance, enough negativity to show there is a problem, but not so much the message gets lost.

If Jimbo agrees that hiring more nerds is a good idea then we can ask them to work on the mobile interface.

@Jimbo Wales The mobile interface is much much better than it was, and a lot of improvements have been made, but there are still some things that could be improved, as Cullen328 points out. It would be awesome if there would be some more nerds to work on stuff like that. Polygnotus (talk) 18:18, 15 March 2025 (UTC)

Here is a good example. Ten years ago it was announced (with no prior community discussion) on mediawiki.org (which is the Wikipedia equivalent of In a cellar, behind a locked door marked "beware of the tiger.") that "More than 2,000 Notifications, will start to be removed". The reason given was "This change is mainly intended to reduce a performance bottleneck." There is no explanation what this bottleneck is, or what discovery has been done to see what the improvement will be, or what alternative have been considered. The Gerrit task to implement it merely says "Add job to keep user notifications in reasonable volume." A priori it's likely that a SQL or database fix could have resolved the bottleneck.

As far as I know this is the first time that WMF has deliberately deleted data people want to keep. I and a few others, however, came across this page and objected. I noticed, after a few years, that I was loosing notifications. Trizek kindly raised T227853 in 2019. MMiller reponded with "We'll need to wait to see if this becomes a widespread issue for many users before spending any time with it" to which I responded (as I do now) with "Meanwhile data is being irretrievably lost."

A few years later, an editor undoes thousands of my edits, which means all my notifications are gone - note failing to preserve Wikimedian's data is probably contrary to GDPR, as well as best practice and movement ideals.

I raised 2 tickets

1. T367755 Restore lost notifications. 8 months later closes with "There is no way we keep copies of deleted data for that long. I'm afraid that your old notifications are gone forever, and there's nothing anyone can do to restore them." Which is quite likely wrong if a WAL is maintained (and if it isn't it probably should be), but thanks for at least responding.
2. T367754 Attack deletes user's notifications. This is a security issue, as well as deleting data it makes it possible to hide relatively new notifications. A commentator suggested that this was not an easily achievable attack, and the ticket was closed. I have reopened it, with a more comprehensive (but I hope not too WP:BEANS]y) explanation of how this is not hard to achieve.

Summary, WMF fails to respond to community objections. Fails to explain the underlying issue. Fails to restore lost data. Fails to stop deleting data. Fails to close security hole (although they succeed in closing the ticket). This may never have happened if they had discussed the original issue. 10 years later this is still having an impact.

I have other examples which are less complex, but much older than ten years.

I'd emphasise that this [things not getting done] is a subtle issue, and certainly not a personal criticism of those involved. All the best: Rich Farmbrough 23:41, 16 March 2025 (UTC).

\volunteer developer hat. This just sounds like "decisions I don't like". There is no amount of developers that will result in "everyone gets everything they want whenever and however they want it". There were performance issues, people took action. It's cool people want to think along, but that doesn't mean most people have a proper understanding of the scale of performance issues that Wikimedia experiences. Those who do can already find the people in question. It took 7 years before wmf was even able to do any sort of substantial database migration again, they had bigger issues to take care of.

I'm all for hiring more developers to fix bugs (which I agree is heavily underfunded), but notifications by nature are ephemeral (they didn't even exist originally and are still optional). Everything can be build / fixed, but that doesn't mean it makes economic sense (time/money/effort, whichever economy you pick) to do so. This is a nice to have at most. And you just repeating and demanding that Wikimedia developers take action, doesn't change that, and will indeed cause people to ignore the repetitive requests. —TheDJ (talk • contribs) 12:54, 18 March 2025 (UTC)

Bit of an ad hominem there, I think. I'm not demanding that anyone do anything. I raised some tickets.

• I would appreciate knowing what the "performance issue" was. It is very unlikely that a significant number of accounts had over 2,000 notifications at the time, so a that raises questions, if the issue was the total number of notifications on the system this wouldn't help much. If something was falling over on a per-account basis then there would likely be a way of limiting a query, without actually deleting notifications. If it was syncing, then there would probably be a chunking solution. It's also possible that what was a performance issue 10 years ago would no longer be a performance issue today. Of course it could be that this was a wicked problem that could only be solved by limiting the number of notifications a user can have, but without sharing the issue, I don't think it's a very convincing case.
• I don't know what you mean by "find the people in question" - unless it's that getting things fixed is better done by talking directly to devs than raising tickets. In which case it's not what you know but who you know.
• You say "notifications by nature are ephemeral," I don't agree. Nothing on-wiki is supposed to be ephemeral, every byte is supposed to be preserved for all time. It's true that we hide a lot of stuff by faux-deleting it, also true that much is probably more hidden than it should be, but we very rarely actually delete anything, pretty much only for legal reasons.

All the best: Rich Farmbrough 11:27, 19 March 2025 (UTC).

I'll just add a coda, without clarity on the "performance issue" it makes no sense to "demand" a reversion of the initial change. I have however, offered a solution which would allow the maximum number of items to be set on a per-wiki basis. This might be useful for non WMF users of the software regardless.

All the best: Rich Farmbrough 11:52, 19 March 2025 (UTC).

This looks very much like you holding a grudge to me and not a real problem with too much technical debt (which definitely exists). And the WMF can't maintain a four-year-long record of all database changes, both because that would no doubt be terabytes (or maybe even petabytes) of content, and because it would contain private data which they have promised to not keep for more than 90 days. * Pppery * _{it has begun...} 18:19, 19 March 2025 (UTC)

Perhaps more concerning than MediaWiki design flaws is the Foundation's inability to handle reports to legal-reports@. I've been waiting on one report for over two months now and the other report I sent only got actioned when I emailed the volunteer Commons Oversight queue. JayCubby 21:19, 16 March 2025 (UTC)

Ok, but I am asking Jimbo to hire more nerds. So if you want the WMF to hire more people who respond to emails to that email address it may be a good idea to start a new section so that we don't go offtopic here. Polygnotus (talk) 21:26, 16 March 2025 (UTC)

I figured it was in the same vein of the Foundation prioritizing more the shiny things and less the more urgent things, like CSAM or MediaWiki maintenance. Feel free to move my comment to a new section if you find it irrelevant. JayCubby 00:51, 17 March 2025 (UTC)

@JayCubby Thing is, I find it very relevant. And it saddens me to hear that you had to wait so long. But the more we ask the smaller the chance is that Jimbo agrees. Polygnotus (talk) 00:53, 17 March 2025 (UTC)

Polygnotus is correct—please start another section to discuss anything that does not involve reducing technical debt. Johnuniq (talk) 03:02, 17 March 2025 (UTC)

I have seen lots of issues but finding them now is tricky. A simple case is phab:T281921 where (nearly four years ago), it was reported that visiting bn:Template:Arguments generates an error. It's a minor matter but there should be a WMF technical person who notices stuff like that and spends a few hours finding the cause (is it indicative of a fundamental problem?) and solution. Johnuniq (talk) 03:02, 17 March 2025 (UTC)

Yes it is unfortunate that many such operational tickets remain open. Additional and dedicated staffing to take care of operational bugs and problems that the community encounters would be highly appreciated. This goes even more so for the sister sites, which are in FAR deeper holes than wikipedia is. —TheDJ (talk • contribs) 13:11, 18 March 2025 (UTC)

This has been investigated. It's some old corrupt DB in the database that doesn't indicate any sort of fundamental problem. And that was already pointed out years ago. * Pppery * _{it has begun...} 18:19, 19 March 2025 (UTC)

Here's another, that just bit me today - I see from Phabricator that it has bitten me in the past. T4700 (soon to celebrate it's 20th birthday) the pipe trick doesn't work inside refs (or other extension tags). All the best: Rich Farmbrough 11:06, 19 March 2025 (UTC).

Could this be... good news... from India?

Courts have to be tolerant: Supreme Court on Delhi HC's takedown order against Wikipedia in ANI case Gråbergs Gråa Sång (talk) 15:35, 17 March 2025 (UTC)

ADL report

A short note from the Anti-Defamation League: Editing for Hate: How Anti-Israel and Anti-Jewish Bias Undermines Wikipedia’s Neutrality Gråbergs Gråa Sång (talk) 10:47, 18 March 2025 (UTC)

If they would've contacted some Wikipedians to ask some questions they wouldn't get such basic facts wrong. There are millions of regular users on Wikipedia, but only 840 administrators and 15 arbitrators. Wikipedia has not developed large-scale or advanced technical solutions, such as automated detection, to enforce its policies against bias or harassment. Unfortunately they don't supply a list of diffs and accounts and ips they consider suspicious. I will email them and ask. Polygnotus (talk) 15:14, 18 March 2025 (UTC)

Well, if you count every person who's made an edit on en-WP, it may or may not add up to millions. Personally I reacted a little to the Table 3 source-list. Gråbergs Gråa Sång (talk) 16:02, 18 March 2025 (UTC)

I have emailed them to ask for the lists of suspicious accounts/IPs and edits. Table 3 is indeed interesting. Polygnotus (talk) 15:14, 18 March 2025 (UTC)

Some of the analysis is a bit confusing. "Google Books is the top most cited source, which represents links to a range of books, some of which are also biased." Well, yes. "Looking at these editors’ contributions since October 7, 2023, Al Jazeera moves to the top most-cited, though PalestineRemembered.com no longer appears.", PalestineRemembered.com wasn't in the top 10 for the previous table, it didn't appear in the top 10 for either. CMD (talk) 07:20, 19 March 2025 (UTC)

I have been known to use gbooks, Jstor and doi:s in my references myself on occasion. I guess that makes me suspicious from the ADL-pov. Gråbergs Gråa Sång (talk) 07:24, 19 March 2025 (UTC)

@Gråbergs Gråa Sång hamas-propaganda.com would probably get blacklisted quickly. And if you want to spread propaganda it would not be a great idea to use hamas-propaganda.com as a reference on Wikipedia. Polygnotus (talk) 07:42, 19 March 2025 (UTC)

You lost me, there is no hamas-propaganda.com in Table 3. Gråbergs Gråa Sång (talk) 07:46, 19 March 2025 (UTC)

@Gråbergs Gråa Sång I mean that it makes sense that a hypothetical group of anti-Israel/anti-Jewish propagandists would not use hamas-propaganda.com as a reference on Wikipedia. It would be weird if they did. So it is not very surprising that the table lists "normal" sources and not those you'd associated with anti-Israel/anti-Jewish propaganda. Polygnotus (talk) 07:54, 19 March 2025 (UTC)

My reading: the ADL-report made a "Top 20 sources bad-faith editors cited most frequently in pages on Israel/Palestine" table. This implies that editors who use these sources are to be suspected, the more use, the more suspicion. I use gbooks, Jstor, doi, BBC, WaPo, Haaretz, Aljazeera, Reuters etc when I ref stuff, ergo, from the ADL-pov I'm to be suspected for bad-fatih editing. Gråbergs Gråa Sång (talk) 08:03, 19 March 2025 (UTC)

I read it more as, "here is the list, and within that list, there are sources that indicate bias". It's just a bit farcical that they felt the need to say gbooks might be biased, it undercuts the rest of the message. CMD (talk) 08:22, 19 March 2025 (UTC)

That works too. Gråbergs Gråa Sång (talk) 08:41, 19 March 2025 (UTC)

I thought

"While Wikipedia’s guidelines provide editors with significant autonomy, certain activities are serious violations of Wikipedia’s code that result in a user’s account being deleted. Because contributors often spend so much time on Wikipedia, and are deeply invested in their efforts, threatening a user with account deletion carries real weight."

was a bit weird, but maybe the writer, human or not, confused delete with block/ban. We (Wikipedians) do have our own jargon. Gråbergs Gråa Sång (talk) 07:41, 19 March 2025 (UTC)

millions of regular users. That's the bit where they're full of it; if we actually had millions of regular users... God only knows what we could achieve. BarntToust 00:38, 19 March 2025 (UTC)

We'd have to change 3RR to 30RR; maybe more! Polygnotus (talk) 00:40, 19 March 2025 (UTC)

Wikipedia has not developed large-scale or advanced technical solutions, such as automated detection, to enforce its policies against bias or harassment. Hey Jimbo, once you're done working out that source scraper, do you think you could program something that could scrape discussions (ex. a given discussion within WP:PIA) to have an AI check editor bias? lol. BarntToust 00:43, 19 March 2025 (UTC)

@BarntToust According to that link above BFT already uses AI to detect bias. This may explain why I am not impressed. In terms of research, BFT leverages cutting-edge advances in generative AI and large language models (LLMs) to address pressing global challenges. Polygnotus (talk) 00:50, 19 March 2025 (UTC)

I read them as that their definition of regular users are those not admins and arbs. Gråbergs Gråa Sång (talk) 05:59, 19 March 2025 (UTC)

Using Google Books is like going to a bookstore or a library or a book table at a flea market. Or your own bookshelves. Some of the books that you find are reliable sources and others are worthless. Some may be entertaining but of little value in building an encyclopedia. One of the most important skills of a good Wikipedia editor is the ability to separate the wheat from the chaff. Cullen328 (talk) 07:33, 20 March 2025 (UTC)

A bit different - questions on methodology

I'll first note that I really don't want to discuss the Israeli-Palestinian question. It's just that it seems like that would be a sure way to get everybody on Wikipedia and elsewhere angry at me (given my beliefs). That's something I generally avoid doing. I'm not trying to blame anyone in the underlying conflict.

When I've read about previous efforts to do what ADL seems to be trying to do here (blame some group for poorly written Wiki articles), I'm generally pretty dismissive and do not even read the news, article, etc. in that it looks like they are starting out wrong and that nullifies the whole inquiry right away. It's always seemed that they identify groups based on subjective factors (i.e. do I agree with what they say). They define good guys and bad guys right from the start, and they will inevitably conclude that the good guys are good, and the bad guys are bad. Why even read it?

ADl does start writing at the top naming a group "bad-faith editors", but are they doing what I described above? There is a section Coordinated editors that starts

"We identified the group of 30 editors that appear to be in close coordination by calculating the minimum time between edits made on pages related to Israel, Jews, or the Palestinians, such as the 1948 Palestine War and Israeli War Crimes articles. We then analyzed all pages on which they commented or were mentioned, including backend discussion (“talk”) pages and user talk pages (discussion pages for individual editors)."

"Once we identified this set of 30 editors,..."

I don't fully understand where this group comes from. Can anybody fill me in? It would seem that there might be 2 groups of coordinating editors. Or maybe both "good guys" and "bad guys" might be part of one group of coordinating editors. And then why label them "bad-faith editors" right from the start? Why not just "Coordinating editors"? There's no hard and fast rule on Wikipedia about how fast you edit (or save edits) or against using talk pages. Do Israeli editors talk less that Arab editors? So I'm just asking whether this selection mechanism has any basis, or is maybe a just a method of convenience?

If that question can be answered, then we could start looking at the rest of the paper. Some of it looks interesting, a bit seems questionable, but it all depends on the "If" starting this sentence. My main interest here is on how you identify the groups. That's a question I've had to deal with in my work on The Signpost regarding paid editors, and I'd love to see that type of work move forward. So I'm not routing either for or against the ADL's paper, just wanting to know how they got their groups. Smallbones_(smalltalk) 18:37, 19 March 2025 (UTC)

I haven't examined the issue but surely this is another case of confirmation bias. They have certain beliefs and they keeping analyzing data until they find "evidence" that supports their beliefs. Johnuniq (talk) 04:01, 20 March 2025 (UTC)

The "bad-faith editors" label is likely just appropriating the en.wiki terminology. Some of the data, such as chart 4, suggest there was some form of network analysis, although it is unclear if this was used to identify a group or confirm a pre-existing concept of a group. It's not clear to me however from reading how the claimed group would be distinguished from "editors who are interested in the topic area", as one would expect such editors to naturally network more. Further, as all the data is presented as aggregate, the work of the group might just be of a few editors, and different charts may actually result from different subsets of that group. Hard to say without more details. CMD (talk) 05:46, 20 March 2025 (UTC)