User talk:Jimbo Wales

WMF technical team

Today, after 17 years and 240,000 edits, I stopped contributing to Wikipedia. The last straw was the WMF's dismissal of its tech team, but that is just one symptom of a more widespread problem.

The WMF has completely lost the plot. It was formed to facilitate Wikipedia and initially did so well. The disks spun reliably; the software mostly worked. Since then, the WMF has bloated exponentially. Funded by aggressive and deceptive begging, it has turned to ever more tangential ways to dispose of its embarrassingly large cash surplus, yet refuses to fund basic technical fixes which the community has been requesting for years. The WMF is now so out of touch that it cannot remember its true purpose or who funds it: its mission statement and annual goals don't even mention an encyclopedia.

With hindsight, handing over our brand, domain and trademarks to the WMF was unfortunate. We are its cash cow, and it has us over a barrel. An editing strike now seems likely, but I don't think even this will stop the WMF, as it can afford to replace departing editors with less experienced substitutes.

We are losing the Wikipedia you founded. Please have a word with those who can still save it. Thank you for reading. Certes (talk) 14:49, 21 May 2026 (UTC)

"This: "refuses to fund basic technical fixes which the community has been requesting for years" is false. What is happening is something that you should attempt for a minute to understand. Let's acknowledge: there are technical fixes which the community has been requesting for years and which have not been resolved. There was a structure (community tech time, 5 engineers, 1 manager) which - per the very point you are making - did not solve that. After a lot of internal analysis a decision has been made to recognize that the existing structure has been a bottleneck. "In shifting Community Tech from a single team into a program that multiple teams are officially responsible for means the wishlist will continue to work on wishes from across languages and wikis. It will still have dedicated staff managing the wishlist intake and triage process and will continue the same financial support for this work, just under a different structure." - This is directly from Suman. Jimbo Wales (talk) 05:28, 23 May 2026 (UTC)

So the question there becomes is, does a team championing community requests perform better on achieving those requests than part-time responsibilities foisted on those who haven't had those responsibilities in the past? Econ Geek 876 (talk) 07:10, 23 May 2026 (UTC)— Econ Geek 876 (talk · contribs) is a confirmed sockpuppet of Nrcprm2026 (talk · contribs). SuperPianoMan9167 (talk) 19:39, 24 May 2026 (UTC)

The problem to the bottleneck argument is that having more employees to solve the same long list of issues isn’t a necessarily a bottleneck, it's only a bottleneck if you make it one. There was never anything that stopped other teams from resolving wishes on their own, in fact, I don’t think it required co-ordination with CommTech in most cases, so how is it a bottleneck to also have CommTech maintaining a set of tools no one but them maintained and solve issues no one but them wants to solve? stjn 09:41, 23 May 2026 (UTC)

> how is it a bottleneck to also have CommTech maintaining a set of tools no one but them maintained

I'm not sure exactly what you are asking here. The laid-off engineers supported many community tools, which noone else in CommTech understands how to support. This was a blunder, and needs to be acknoleged as such. Econ Geek 876 (talk) 10:01, 23 May 2026 (UTC)— Econ Geek 876 (talk · contribs) is a confirmed sockpuppet of Nrcprm2026 (talk · contribs). SuperPianoMan9167 (talk) 19:39, 24 May 2026 (UTC)

I know that, that was a rhetorical question. What Jimbo doesn’t seem to get that laying off a bunch of engineers doesn’t help anyone at getting more community wishes fulfilled. stjn 14:20, 23 May 2026 (UTC)

Thank you for taking the time to follow this issue. I still cannot see how a reduction in resources will ease the tech support bottleneck. More generally, I fear that the WMF has little understanding of, or interest in, ordinary editors' concerns (and the only redundancies which could improve that situation would come much higher up the management structure). I respect you and trust your judgement, but some tangible sign of the WMF descending from its ivory tower to give the support we need would really help to restore our confidence in it. Certes (talk) 20:09, 23 May 2026 (UTC)

If you wish to follow my advice, I think doing all this would be effective crisis management. It may be a weekend but if there's anyone you can convince at the C-level to say something meaningful like that, I can almost guarantee that people would be more satisfied than they are right now. Clovermoss🍀 (talk) 20:17, 23 May 2026 (UTC)

The most recent statement is nowhere near satisfying people. I'm trying to warn you that you're on the path to having hundreds of vital editors not editing at all. See Wikipedia:Wiki Workers United solidarity. Clovermoss🍀 (talk) 02:35, 25 May 2026 (UTC)

We're up to 270 signatures now, many of which are 20+ year editors, and countless social media posts about how people are cancelling their recurring donations. This story is #8 on Hacker News. This situation is only going to get worse the longer the foundation does nothing. Clovermoss🍀 (talk) 01:11, 27 May 2026 (UTC)

Now we're at 400+ signatures and fastly growing, making this the 8th most supported thing ever on Wikipedia. I'm feeling a lot like Cassandra right now. Clovermoss🍀 (talk) 14:04, 27 May 2026 (UTC)

If you won't listen to me, maybe you'll read Risker's long comment about why this was a horrible decision on every front. I'd like to emphasize the It seems the "volunteer support goal" mentioned in the annual plan is just some words on the screen part. Clovermoss🍀 (talk) 15:01, 6 June 2026 (UTC)

I just want to reiterate how disappointed I am in the foundation. This will be my go-to example for how out of touch it is with community needs and values going forward. I'm never going to forget this. Clovermoss🍀 (talk) 09:40, 7 June 2026 (UTC)

for those unaware of the context, see m:Talk:Community_Wishlist#May_20_update and Wikipedia:Village_pump_(WMF)#Petition:_Editors_willing_to_join_in_collective_labor_action

Jimbo, this looks like something the board at least needs to look at (and maybe you are, already) before there's another FRAM situation. -- Aunva6^{talk - contribs} 17:32, 21 May 2026 (UTC)

The board is aware of it and supportive. It's time to get serious about meeting community needs, and restructuring away from a system that clearly wasn't working is a positive move. Jimbo Wales (talk) 05:30, 23 May 2026 (UTC)

@Jimbo Wales: I would strongly urge you to read my comment to gain some context on why Suman's proposed restructuring is (in my opinion) not the correct approach. TLDR, the restructuring proposed by the WMF does not actually solve the underlying problem, rather it will exacerbate it, requiring that the community justify every wish in terms of "why X team should drop their OKR priority and work on my wish". This is not a positive thing. We need to first solve the community intake and prioritization challenges before we can remove the only pillar that was holding the Wishlist together (which I'm not terribly convinced needed removing in the first place). Sohom (talk) 06:04, 23 May 2026 (UTC)

Sorry in advance for the bluntness, but you and the board are extremely out of touch with the community/ies (and this is a widely-held view). Re your role, my 2c (informed by hearing others' thoughts on this): enwiki's outgrown the need for a god-like figure to provide checks and act as the movement's 'spiritual guide' as it were. Partly why I and others recently suggested editing, being a pleb isn’t so bad. My thoughts on this saga are here Kowal2701 (talk, contribs) 10:12, 23 May 2026 (UTC)

Restructuring away from a system that wasn't working well enough is only a positive move if there is solid evidence that the new structure will do better. I haven't seen any. Econ Geek 876 (talk) 23:50, 23 May 2026 (UTC)— Econ Geek 876 (talk · contribs) is a confirmed sockpuppet of Nrcprm2026 (talk · contribs). SuperPianoMan9167 (talk) 19:39, 24 May 2026 (UTC)

@Certes, I can tell you from my own experience in conversations with WMF people that it can afford to replace departing editors with less experienced substitutes is completely untrue. The WMF is extremely concerned with declining readership and what that means for the future of the editorial and administratorial corps. Of course, that doesn't mean that particular WMF employees or leadership as a whole will always make the decisions you or I think are best for these groups, but they do very much have this in mind. -- asilvering (talk) 19:25, 21 May 2026 (UTC)

The WMF is extremely concerned with meeting their self-set KPIs. An edit that reverts nonsense and an edit that makes a featured article are the same to them: it's just contributor engagement. Trying to dress up the current WMF situation as anything other than a disaster from an out-of-touch bureaucracy is extremely unhelpful. Johnuniq (talk) 23:21, 21 May 2026 (UTC)

Johnuniq, let me just stop you right there: you are talking absolute rubbish. Literally *no one* at the foundation is extremely concerned with meeting pointless KPIs. Seriously that's not even remotely close to accurate. Why make stuff like that up out of thin air just to cause people who don't know better to have incorrect ideas?--Jimbo Wales (talk) 05:22, 23 May 2026 (UTC)

Thanks, that's encouraging. I look forward to a response explaining why it was appropriate to sack extremely dedicated and talented developers, at least one of whom has spent two decades ensuring that Wikipedia runs efficiently and securely. Johnuniq (talk) 06:01, 23 May 2026 (UTC)

I have read enough about the Foundation from their frequent and very appreciated outreach, to believe that they do very often if not almost always measure employee performance by KPIs, which I think is a good thing. The question of how meaningful those KPIs are is separate, with reasonable people extremely likely to reasonably disagree on that second question. (If you don't believe me, look at the diversity of responses to the GRDC requests for feedback.[1]) Having said that, and pulling back a bit to avoid comparing hypothetical KPI constructs, let me just say that removing a longstanding community support department, instead of distributing them amongst the teams which support the goals that the community surfaces through the annual wish list survey, seems like a managment blunder, but not a particularly suprising sort of such blunder. What has impressed me is how the community has come together in support of corrective action for this blunder, which makes sense because it cuts off their ability (and the primary traditional means) to communicate their needs to Foundation engineering staff. I commend the community, and, Jimbo, I think you need to take a closer look. Econ Geek 876 (talk) 06:40, 23 May 2026 (UTC)— Econ Geek 876 (talk · contribs) is a confirmed sockpuppet of Nrcprm2026 (talk · contribs). SuperPianoMan9167 (talk) 19:39, 24 May 2026 (UTC)

"It will still have dedicated staff managing the wishlist intake and triage process" - that doesn't make it sound harder to communicate needs to the engineering staff at all. Jimbo Wales (talk) 09:28, 23 May 2026 (UTC)

The quote is, "We still have dedicated staff managing the wishlist intake and triage process and will continue the same financial support for this work, just under a different structure."[2] My question, whether "a team championing community requests perform better on achieving those requests than part-time responsibilities foisted on those who haven't had those responsibilities in the past" stands. Econ Geek 876 (talk) 09:54, 23 May 2026 (UTC)— Econ Geek 876 (talk · contribs) is a confirmed sockpuppet of Nrcprm2026 (talk · contribs). SuperPianoMan9167 (talk) 19:39, 24 May 2026 (UTC)

@Jimbo Wales: I'm unsure how deeply you've been informed about this, but when Suman says "dedicated staff", what he means is "a single product manager" (who, while being a great guy has had less than 11 months at the company, compared to the CommTech team's years of experience working with the community). Also I find that doesn't make it sound harder to communicate needs to the engineering staff at all. kinda baffling. Having a system where a product manager talks about a wish behind closed doors to another product manager who can then arbitrarily decline said wish does make it infinitely harder to talk to the actual engineers (as was possible with CommTech). Sohom (talk) 18:10, 23 May 2026 (UTC)

@Jimbo Wales and @Econ Geek 876: WMF has indeed measured a significant majority of it's team performance through metrics, that can be considered KPIs that are set on a yearly basis through the formulation of OKRs during the Annual planning process that they try and hit throughout the year. Whether or not they are pointless is a different discussion to have. (also if folks think that a OKR is pointless/unneeded for the next year, you can voice concerns until the end of May at m:Talk:Wikimedia_Foundation_Annual_Plan/2026-2027) Sohom (talk) 07:29, 23 May 2026 (UTC)

I feel the discussion around KPIs etc distracts from the two big discussions: 1. How to give the community more say in the direction of technological development. Firing of highly-skilled engineers with a great understanding of the community cannot be a solution here, even if we might disagree with how CommTech needs to be reformed. My preferred route is a combination of an annual wishlist organised by CommTech, with an obligation on different teams to select prioritised wishes. 2. How we can create a culture within the Foundation where people are not afraid to voice their opinions, and employment is not put at risk for voicing these opinions. I'm dismayed to hear from multiple directions that employees are afraid to speak out, and that those who have spoken out against senior management have faced dismissal not long after, whether related to the union formation or not. Jimmy, I hope the board takes an active role in fighting for increased community input, putting more value on expertise of staff with community expertise, and supporting a cultural shift where dissent is encouraged. In solidarity, —Femke (talk) 🐦 09:38, 23 May 2026 (UTC)

Suppose team A is given performance metrics based on their previous performance, which are easy for them to reach. Suppose team B is given performance metrics based on difficult achievements which are known to be strongly aligned with community goals. If team A's metrics are usually greater than team B's, does that mean team A is better than team B? Econ Geek 876 (talk) 10:09, 23 May 2026 (UTC)— Econ Geek 876 (talk · contribs) is a confirmed sockpuppet of Nrcprm2026 (talk · contribs). SuperPianoMan9167 (talk) 19:39, 24 May 2026 (UTC)

I do not believe this is pertinent to the two key discussions here, where we could use Wales' help. In solidarity, —Femke (talk) 🐦 10:16, 23 May 2026 (UTC)

I'm not sure if you're asking me or Femke or just the room in general, but my answer would be no, that wouldn't be a very sensible way to do things. There's a very famous essay in management science, "On the Folly of Rewarding A while hoping for B" which, if you don't know it, you may want to read as an interesting exploration of these points. Jimbo Wales (talk) 10:35, 23 May 2026 (UTC)

That is an excellent paper which is exactly on point here. I would urge everyone to read it and propose how its findings might be applied to the current situation here. Econ Geek 876 (talk) 23:40, 23 May 2026 (UTC)— Econ Geek 876 (talk · contribs) is a confirmed sockpuppet of Nrcprm2026 (talk · contribs). SuperPianoMan9167 (talk) 19:39, 24 May 2026 (UTC)

Jimbo, beyond the restructuring itself, many community members are becoming increasingly concerned about reports of how dissenting staff are treated internally. Multiple former staff members have described a culture where criticizing leadership decisions, raising concerns about product direction, or advocating for long-requested community needs could negatively affect their careers. Some also describe being discouraged or even prohibited from discussing certain technical ideas publicly or across teams. Whether every individual account is fully accurate or not, the fact that so many experienced people independently describe similar patterns should concern the Board.

Do you believe the Foundation currently has a healthy internal culture where staff can openly disagree with leadership without fear of retaliation? And will the Board investigate these concerns seriously rather than treating them as isolated complaints? Nemoralis (talk) 11:06, 23 May 2026 (UTC)

Since Selena isn't answering questions

Hi Jimbo. Since SDeckelmann-WMF has decided that, apparently, 100+ editors showing willingness to go on strike is the sort of situation where it's fine to walk away for the next few days and not answer any questions, I'm hoping that you can address what I said to her on Meta. I'll condense it to two questions:

Given that the Foundation has said this reorganization is meant to improve outcomes with the Wishlist; and given that the Foundation has not explained how these employees' 10,000+ annual worker-hours of effort on the Wishlist will be replaced by existing workers on other teams (presumably not 10,000 hours of overtime), and given that the Foundation is assuring us it is quite willing to re-hire everybody (or, at least, keeps coming very close to saying that without actually saying it):

1. What reason did the Foundation have to lay these people off rather than transfer them to other teams, as has been done in the past for some teams' dissolutions?
2. Regardless of the answer to (1), is the Foundation willing to remedy the harms they've caused and rebuild community trust by offering those transfers now?

-- Tamzin^{[cetacean needed]} (they|xe|🤷) 11:45, 23 May 2026 (UTC)

Selena has posted an update on VPW and on Meta. SuperPianoMan9167 (talk) 22:40, 24 May 2026 (UTC)

700+ editors now, at least. Wikipedian12512 (Talking is fine | contribs) 20:27, 29 May 2026 (UTC)

926. Herostratus (talk) 20:53, 1 June 2026 (UTC)

Reason for a union

I have observed three extraordinarily talented developers over many years. One was fired a while ago. It was obvious that some manager with zero understanding of the situation thought he could increase efficiency by firing an incredibly dedicated super-developer. Now we have this fiasco where two others I am familiar with have been chopped because [something goes here]. Senior management should get them back and tell the managers to leave them alone. They don't need to be managed—just get out the way. You cannot buy 20 years of experience. We assume it's just a coincidence that engineers were talking about a union. A thoughtful person would ask what is the underlying reason senior developers would waste their time arranging a union? Developers like to solve problems. The only reason they would spend time on a union is because management is broken. Johnuniq (talk) 03:08, 24 May 2026 (UTC)

1,000 English Wikipedians 100,000 edits

A group of 100,000 edit Wikipedians toasting you in Paris2026 (child editor alert, the kid mostly used bots)

Here's something which deserves a glass held high toast to you and the community: Ecangola just became the 1,000 English Wikipedian to reach 100,000 edits. For Wikipedians who notice round numbers. Randy Kryn (talk) 13:41, 2 June 2026 (UTC)

Bah, humbug. Carlstak (talk) 16:58, 2 June 2026 (UTC)

That's awesome!Jimbo Wales (talk) 20:07, 6 June 2026 (UTC)

A really good essay by Jake Orlowitz

A link to this essay by WMF Good Egg Jake Orlowitz showed up on Wikipediocracy yesterday and I pass it along here. A really good piece of work on the state of Wikipedia, well-crafted and smart. Highly recommended reading. Carrite (talk) 17:50, 6 June 2026 (UTC)

• @Clovermoss: have you had a chance to read this? Thanks to Carrite for sharing. ↠Pine (✉) 20:46, 6 June 2026 (UTC)

I'm not an AI hater, and agree that this essay is smoothly written and brings several ideas and concepts into the debate that not everyone might be familiar with and that are worth knowing about (by Raymond, Hayek, Hirschman, Scott, Chesterton). And of course it reiterates important concerns that many other community members have voiced over the past few weeks.

However, people should be aware of the fact that Pangram (the only AI detector which enjoys a good reputation for a generally very low false positives rate) scores this essay as 100% AI-generated with high confidence, and that it contains various other strong indicators that this is LLM-generated text, such as metaphors that are contorted to the point of nonsense (e.g. The wire to the bazaar is not merely unplugged; the current has reached the center before, and the center decided it could just take the shock). And there are several parts that read superficially plausible and convincing, but fall apart upon closer scrutiny, due to what are quite likely LLM hallucinations/confabulations.

For example, Orlowitz (or the LLM he tasked with writing up this essay) claims that The dissolved [Community Tech] team [...] are the people behind the projects’ dark mode. But that Dark Mode feature was built by an entirely different team, after Community Tech had explicitly rejected the corresponding wishlist request (The wish is out of scope for the team and conflicts with the Desktop Improvements initiative).

What's more, Orlowitz (or the LLM he tasked with writing up this essay) completely misrepresents the Community Wishlist as a grassroots community ritual[] that was later taken over by the Foundation in a disastrous attempt at making the process legible to the cathedral [WMF] and meaningless to the bazaar [the Wikipedia community] (in reference to Raymond's The Cathedral and the Bazaar):

There is a second failure, one the anthropologist James Scott describes. In his study of why grand schemes of improvement so often end in ruin, he names the recurring villain: legibility, the urge of any central authority to remake a messy, living reality into something simplified and readable from their own central vantage point. [...]
The Wikimedia case is a textbook instance. For most of a decade the Community Wishlist Survey was one of the community’s defining rituals. Each year editors proposed technical wishes and registered support in the open. A ranked list emerged, and the team built the ones at the top.
The declarations of support and opposition are what Wikipedians half-jokingly write with an exclamation mark, the !vote, because consensus is meant to be evaluated, not counted. It was the one place in the apparatus where the volunteers, not the managers, set the engineering agenda.
At some point it was rationalized into something more legible to a product manager: a perpetual backlog with no season, no vote, no commitment, sorted into focus areas that read cleanly on a roadmap. Participation collapsed by as much as nine in ten.
Here two failures fuse into a single, almost poignant error. The collapse, the direct result of making the process legible to the cathedral and meaningless to the bazaar, was offered as proof that the new system was working.

Now, I personally agree that WMF's changes to the Wishlist in recent years appear to have been counterproductive. However, the essay makes self-contradictory claims here (how could WMF/the cathedral/the team have built the ones at the top before, if they were not legible?). And most importantly, Orlowitz' overall thesis here (cathedral takes over bazaar process and thereby destroys it) is contradicted by the simple fact the Wishlist had been implemented top-down by WMF from the start, with pretty much the explicit goal to make things more legible to a product manager. See for example this 2016 Signpost article by a WMF product manager, which describes the first iteration of the wishlist as follows: we invited the community to participate in a cross-project survey to set our agenda for the year. Again, the essay's claims here looks superficially plausible on first read, and eruditely connect to broader ideas (that can be assumed to be present in any modern LLM's parametric knowledge), but entirely fall apart after some concrete fact-checking.

Another example of an apparently hallucinated/confabulated claim is that Brooke led development of MediaWiki, the software the projects run on, for more than two decades (repeated later: ran the core software for two decades). Now, Brooke is of course a very well respected senior MediaWiki engineer who indeed led MediaWiki development for a time in the 2000s (and has continued to do very valuable work on neglected areas like video since then), so people can and should be concerned about the fact that she was fired. But the quoted claims paint an extremely misleading picture of who led MediaWiki's course after the 2000s. E.g. while B. Vibber was still on TechCom/ArchCom until 2020, it was as one of ~10 members, and she appears to have not been involved in its successor at all. (Personally I find this error especially disappointing because it had already been present in a somewhat milder form in an earlier Medium post by Orlowitz that described Brooke as the longtime lead developer of MediaWiki, and pointed out to Orlowitz in a Wikipedia Weekly discussion last week, where even his defenders merely argued that he could have meant that term historically or as an informal honorary title, and agreed it would be misleading to assert that "the lead developer" was still her formal position as of 2026. But this also makes it even more likely to me that the error in the new post was due to AI use, because Orlowitz had been aware of the Facebook discussion - so either he deliberately disregarded it to double down on an assertion he knew was likely to get further pushback, or the LLM he tasked with writing up this essay wasn't aware of that discussion.)

I'm not quite sure what WMF Good Egg refers to above, but it might represent a misunderstanding that Orlowitz still works at WMF (IIRC he left over half a decade ago). The more relevant professional background is that, as mentioned in his bio on the linked Medium post, he runs a Wikipedia consulting company. Churning out thinkpieces of this sort is a well-known strategy for promoting such consulting businesses, and they have appeared on that Medium account with high frequency recently.

One reason I'm pointing all this out in some detail is to encourage people to instead read some much more thoughtful, better-researched observations by community members who evidently take more time to conduct their analysis and arrive at their conclusions. Like for example this recent summary by Risker (who BTW wisely separates longstanding community-WMF tensions from possibly misguided labor rights activism, see also this new user essay by Thebiguglyalien).

Regards, HaeB (talk) 09:13, 7 June 2026 (UTC)

HaeB makes some good points but I'm not so sure about their use of Pangram, which is itself an AI tool. I'm interested in this because Pangram was used to make a similar claim of AI authorship for parts of the recent Papal encyclical, Magnifica humanitas. See AI-generated text in the encyclical for discussion of that and how Wikipedia should cover it.

In the case of the encyclical, there was a specific tell which was the word "genuinely" which is characteristic of Anthropic's Claude. I'm not seeing such a good clue in this case. I tried Pangram out myself on the Orlowitz essay. It gives the same 100% AI figure but doesn't justify it. There's an option to see evidence and its response was

Supporting Evidence

0 instances found

Human-written text of this length normally contains ~8 AI-associated patterns. This document is within that range.

These patterns aren't used to determine our AI score; they help you see why AI text often reads differently.

My understanding of this is that most human text of this length will contain some false positives and this text is within the normal range. So, the 100% score seems mysterious. My impression is that it's based on Pangram's AI intuition rather than any clear tells and this seems quite weak. See America Has a Pangram Problem for a sceptical view.

As a control, I tried Pangram on Haeb's post here. It would only check the first 1000 words with my remaining credit but gave a score of 100% human for that. But again there was no supporting evidence; just the same bland statement saying that all human text will contain false positives and that this was in the normal range. When the methodology seems to be the AI equivalent of gut feel, I'm not trusting these scores of 100% as they seem too confident in their black and white assessments.

Andrew🐉(talk) 10:42, 7 June 2026 (UTC)

My understanding of this is that most human text of this length will contain some false positives and this text is within the normal range. I'm under the impression that this applies to far shorter texts (50-100 words). Orlowitz's article is 2,400 words. For example, the false positive cited in your article as an example was only 50 words. ARandomName123 (talk)^{Ping me!} 17:09, 7 June 2026 (UTC)

Pangram says that there are 7 sections of AI generated text in Orlowitz's essay. These all seem to be about 400 words in length so my impression is that Pangram processes a long text in chunks of that size. I'm new to use of the tool but, so far as I can see, it doesn't provide any specifics to explain why it thinks these are AI generated. I suppose that it has been trained on large numbers of texts and so its thinking is a huge set of pattern-recognition weights which are inscrutable. If we don't trust such AI reasoning in general then I don't see why we should make an exception for Pangram.

What seems much more important than the rhetoric are the facts of the matter such as the point about Dark Mode. I am sceptical that the achievements of the various WMF teams are as black and white as some suggest. What seems to have been missing most at the WMF is strong leadership as my impression is that the organisation is rudderless. In such circumstances, the efforts of the individuals and teams will be lost in the maelstrom.

Andrew🐉(talk) 17:54, 7 June 2026 (UTC)

Thanks for the close read, appreciation, and critique. I crafted this piece with input from AI, as a tool for research and drafting. In this case I didn't verify specific claims the way I should have. Dark mode was built by the Web team, not Community Tech (though it was highly requested there), and that line is updated. On Vibber, she was employee #1, the Foundation's first CTO, and for a time Lead MediaWiki Developer, until shifting out of that position. The current essay clarifies that, and adds new details throughout.

On the Wishlist, though, you've misread me @HaeB, and your own source shows it. I never said the community created the survey. I said the community drove it. Here is the full passage you quoted from, from the 2016 Signpost: "Rather than those of us on the team coming up with our own ideas and proposing them to the community, the team decided to let the community tell us what to work on. To do this, we invited the community to participate in a cross-project survey to set our agenda for the year." It goes on to say the team "committed to investigating and addressing the top 10 wishes." You quoted the clause about setting the agenda and left out the sentence before it, where the team hands the agenda to the community, and the line after, where it commits to building what the community votes up. The community directly drove the agenda until the 2024 redesign took that ritual away.

On the notion that I write these pieces to grow my consulting business: that's exactly backwards. I'm spending capital and taking significant risks by doing so. A Wikipedia consultant who picks a public fight with the Foundation is not prospecting for clients. What I'm spending it on is the union's future, the community's say in what gets built, and some accountability for the nonprofit at the center of this movement.

I stand by the piece: the argument, the intellectual history under it, and the ultimate version and voice. The details I got wrong are fixed. The thesis isn't one of them. Ocaasi ^{t | c} 08:02, 8 June 2026 (UTC)

Thanks for the response which seems to vindicate and validate Pangram's intuition. I used today's Pangram credit ration to evaluate Ocaasi's response. It reports that it is 100% human with medium confidence but, again, there are no specifics. My further reading indicates that Pangram analyses the style of the text in a statistical way and that AI tends to be more consistent in its stylistic choices than a human author. Andrew🐉(talk) 08:45, 8 June 2026 (UTC)

Thank you for coming to Aotearoa New Zealand

Heartfelt personal thanks for making the long trip here and spending time with the small but dedicated group of us Wikipedians. We are re-energized by the visit and the perspectives you shared with us. If you get back this way, please drop a note. Your fellow Southerner, UnitedStatesian (talk) 03:55, 7 June 2026 (UTC)

It was great fun, a great meetup! Thank you for your thank you! :-) Jimbo Wales (talk) 11:24, 9 June 2026 (UTC)

Signal statement on mandatory government phone scanning in UK

I just want to make sure that the community is keeping an eye on this: Signal statement. Please post it anywhere relevant for awareness. Jimbo Wales (talk) 11:25, 9 June 2026 (UTC)

After all the recent WMF-related events affecting the community, I'm unsure whether the community is fully interested in (controversies of) online age verification in the United Kingdom or privacy in English law at this time. As I see, the government has wanted to reduce the effects of pornography on young people all along... or so I thought.

Signal claims that citing the effects was an excuse for supposed disregard for child privacy, but that's just assuming. (One of the links directs a user to a documentary Groomed: A National Scandal.) I just now am reading Signal (software)#Controversial use and Signal (software)#Use by Trump administration. George Ho (talk) 20:56, 9 June 2026 (UTC)