britta

Editors have worked together to rewrite most of it: https://sco.wikipedia.org/wiki/Wikipedia:WikiProject_The_Big...

WikiEdu is an independent nonprofit that coordinates a lot of programs where college professors assign Wikipedia editing to students, and they recognized that a lot of students were attempting to add AI-generated text to articles. The organization has done a lot of work over the past couple years to train students on using AI constructively (such as to help with finding sources, not to generate article content), detect potential AI-generated contributions from students, and coach students toward doing the right thing.

https://wikiedu.org/blog/2026/01/29/generative-ai-and-wikipe...

https://wikiedu.org/blog/2026/07/14/generative-ai-and-wikipe...

Wikinewsie is also fun, showing regional current events articles from Wikipedia:

https://wikinewsie.toolforge.org/en/portal/california

https://wikinewsie.toolforge.org/en/portal/europe

Also, in case anyone is inspired, Wikimedia Foundation happens to have a SRE position open: https://job-boards.greenhouse.io/wikimedia/jobs/7972965

Brief technical writeup after a similar event in 2009 when Michael Jackson died: https://wikitech-static.wikimedia.org/wiki/Michael_Jackson_e...

I’ve compiled the entire sordid history of TreasuryDirect authentication in their Wikipedia article: https://en.wikipedia.org/wiki/TreasuryDirect

You might find it interesting to read some of the 2026-2027 infrastructure OKRs: https://meta.wikimedia.org/wiki/Wikimedia_Foundation_Annual_...

Come help! Here's the August 2026 unreferenced articles backlog drive, with instructions and a leaderboard: https://en.wikipedia.org/wiki/Wikipedia:WikiProject_Unrefere...

If it were that easy to recruit new dedicated volunteers who would contribute a non-trivial amount of constructive work and stick with it over time, I'd be delighted and relieved, but it's not. Contributing to Wikipedia, at the level of many of the contributors who signed the petition, requires a lot of patience, enthusiasm, and time, and it requires building quite a lot of specialized skill. If you're doing it right, you get really quite good at a certain kind of research, writing, and reasoning. I've been an editor for 25 years, with 12k edits, and have not yet written an article that qualifies as a "featured article" (https://en.wikipedia.org/wiki/Wikipedia:Featured_articles) - it's super tough!

The foundation's 2026-2027 draft annual plan explains a bit of their current strategy for recruiting more editors, including by deepening engagement among readers in meaningful ways: https://meta.wikimedia.org/wiki/Wikimedia_Foundation_Annual_...

The solidarity petition is public (https://en.wikipedia.org/wiki/Wikipedia:Wiki_Workers_United_...), and it includes a section with statistics summarizing the contributions of editors who have signed the petition, which suggests the impact of signatories deciding to withhold their volunteer labor: https://en.wikipedia.org/wiki/Wikipedia:Wiki_Workers_United_...

A couple highlights:

"5 oversighters [https://en.wikipedia.org/wiki/Wikipedia:Oversight] ... who collectively performed 587 of the 1,463 (~40%) suppression actions in April 2026"

"5 checkusers [https://en.wikipedia.org/wiki/Wikipedia:CheckUser] ... who collectively performed 876 of the 5,419 (~19%) checkuser actions in April 2026"

There are a bunch of research papers that compare Wikipedia and Grokipedia in some depth, and they've found and interpreted a lot of interesting differences. A couple of recaps from the Wikipedia Signpost, which is the internal newsletter for Wikipedians:

2026-03-10: https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2...

2026-01-15: https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2...

I appreciate that this article actually cites some of the relevant discussions. Wikipedia community processes are unusual in that they are largely public, including that foundation staff members respond to community questions in public, so you can read a lot of the backstory yourself. Here are a few links.

Main discussion about Community Tech team on the Village Pump (discussion board): https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(WMF)#W...

* Response from the WMF (21 May): https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(WMF)#R...

* Note from the Wikimedia Foundation on unionization: https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(WMF)#N...

* Response from the WMF (22 May): https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(WMF)#R...

* WWU statement (May 23): https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(WMF)#W...

* Response from WMF 24 May: https://en.wikipedia.org/wiki/Wikipedia:Village_pump_(WMF)#R...

The 24 May statement includes "I know many of you asked why we cannot just guarantee people new roles...we have 4 countries represented, with a wide variance in required actions. I want to note one specific requirement that came from these laws: we could not pre-select certain staff for new roles, as that would appear to be circumventing legally required processes in some countries."

Discussion about proposed direction for Community Wishlist: https://meta.wikimedia.org/wiki/Talk:Community_Wishlist#Prop...

https://www.eff.org/pages/internet-still-works-wikipedia-def...:

"A decade ago, Wikimedia Foundation, the nonprofit that operates Wikipedia, received 304 requests to alter or remove content over a two-year period, not including copyright complaints. In 2024 alone, it received 664 such takedown requests. Only four were granted. As complaints over user speech have grown, Wikimedia has expanded its legal team to defend the volunteer editors who write and maintain the encyclopedia."

https://meta.wikimedia.org/wiki/Wikimedia_Foundation_Annual_...:

"Protect our projects: Sustain and defend our model. Wikipedia's volunteers and values are increasingly under threat. We will provide volunteers with the safety and legal support they need, and defend neutral point of view in a world where facts are increasingly contested and politicized. We will strengthen protection for Wikimedia infrastructure from bot abuse, and lean into responsible reuse that drives value back to Wikipedia, not just traffic away."

FYI, I took Plucker out of the lead in November, after a PG volunteer recommended that update on the article talk page. Plucker is currently only mentioned in a sentence about formats offered in 2009.

Happy to make other updates! Writing specific notes on the talk page is helpful.

My late family member was a technical editor at Bell Labs, and I inherited a few of his papers, and his internal resume is truly boring in many places:

“From October 1970 through December 1970, served as a member of the Western Electric Committee on the Technical Data Design Classification Plan.”

“Worked on studies and publication specifications for an automated Maintenance Data System (MDS).”

He did work on some slightly more interesting things, like documentation for a cable-laying ship, and he was proud of helping with the monumental book project “A History of Engineering and Science in the Bell System”. But overall it was just a job for him, for several decades; he retired as soon as he could.

Aw yeah! I’d love to see somebody from the DevTeam talk about this, or literally anything else they might want to talk about, at the Roguelike Celebration in October (https://www.roguelike.club/), if anyone has a connection and could encourage them to consider it. It’s a super lovely community-run online event, and everyone would be thrilled. (I was a volunteer for the first few annual events, as a person who played about a zillion games of Nethack as a kid.)

Come help! When you come across a math article on Wikipedia that you find difficult to understand, consider writing a talk page comment with specific, polite, constructive feedback. That can help other editors figure out how to improve the article. We have a goal of making articles understandable: https://en.wikipedia.org/wiki/Wikipedia:Make_technical_artic...

Wikipedia has a range of protective mechanisms that admins can apply to high-traffic or frequently vandalized articles: https://en.wikipedia.org/wiki/Wikipedia:Protection_policy

"Protection restricts the modification of pages to specific groups of users. Pages are protected when there is disruption that cannot be prevented through other means, such as blocks. Wikipedia is built on the principle that anyone can edit, and therefore aims to have as many pages open for public editing as possible so that anyone can add material and correct issues. This policy states in detail the protection types and procedures for page protection and unprotection, and when each protection should and should not be applied."

These mechanisms do include a "Pending changes" mode: https://en.wikipedia.org/wiki/Wikipedia:Pending_changes

I want the equivalent of Mythos for Wikipedia - I want world-class tooling that helps human editors efficiently find, prioritize, and mitigate attempts to add deceptive and low-quality content - and I know it's possible to build this kind of thing. A whole bunch of long-time editors, including myself, are excited about building better tools, trying a range of experiments. This is one of the really fun parts about a community-built encyclopedia: you can help build tools too! A few interesting experiments - you can also use these as a Wikipedia reader (some require logging in):

* Cite Unseen (https://meta.wikimedia.org/wiki/Cite_Unseen): show icons in an article's References section that indicate what the Wikipedia community knows about that source, such as whether a website is a known unreliable source - such as whether a source is banned on Russian and/or Ukrainian Wikipedia. [https://gitlab.wikimedia.org/kevinpayravi/cite-unseen]

* AI Source Verification (https://en.wikipedia.org/wiki/User:Alaexis/AI_Source_Verific...): use LLMs to help check whether the citations in an article support the claims, providing a summarized report. [https://github.com/alex-o-748/citation-checker-script]

* Suggestion Mode (https://www.mediawiki.org/wiki/VisualEditor/Suggestion_Mode): provide automatic in-line edit suggestions, including using small language models to detect potential tone issues. Demo: https://www.tiktok.com/@wikipedia/video/7634591061553237266?...

* Microtask Generator (https://meta.wikimedia.org/wiki/Research:Micro-task_Generato...): provide a list of prioritized edit suggestions based on the editor's choice of category. [https://gitlab.wikimedia.org/toolforge-repos/microtask-gener...]

* WikiTask Pro (https://nethahussain.github.io/wikitask-pro/ + https://github.com/nethahussain/wikitask-pro) - another approach to integrating signals to recommend potential edits to editors.

There are also interesting conversations happening about developing and maintaining better data about questionable sources - check out this amazing compilation: https://en.wikipedia.org/wiki/User:Kuru/fakesources

Some places to stay in touch with these things if you're interested: https://www.wikicred.org/ + https://en.wikipedia.org/wiki/Wikipedia:WikiProject_AI_Tools (not all of these kinds of tools involve AI, but it's a component of various things people are working on). If you’re in the SF Bay Area, come to our meetups: https://en.wikipedia.org/wiki/Wikipedia:Bay_Area_Wikipedians...

For anyone at GitHub looking at this thread: please update your documentation page about how to report abuse (https://docs.github.com/en/communities/maintaining-your-safe...). I tried to follow the instructions, but I ran into a bunch of dead ends that slowed me down - I couldn't find the report abuse buttons for issues, comments, or repositories, only for the user profile page. I'm on Chrome on a Mac laptop, logged into GitHub.

Also, on the report abuse page that I got to from the user profile page, the green submit button is nearly hidden by the grey footer, even when I scroll the page around and complete the captcha.

The team is a mix of employees and contractors. They also offer customers (government agencies who use their service) the option to use Bing results or their in-house Elasticsearch results: https://search.gov/admin-center/content/content-overview.htm...

They do good work, and it’s an important service. I believe it saves a ton of money for the federal government by reducing reinvention of the wheel. As a former federal employee and current federal contractor, it’s been very helpful to be able to use their no-cost-to-customer search services on multiple projects. On my current project we eventually shifted to doing our own search (using Postgres full text search) so we could customize the indexing and ranking, but Search.gov was a useful interim solution.

It's not too late! Grab a ticket and come on in: https://www.eventbrite.com/e/roguelike-celebration-2021-tick...

The Forest Service worked with 18F on this Open Forest project to enable people to buy Christmas tree permits online, which uses United States Web Design System components: https://openforest.fs.usda.gov/christmas-trees/forests

You can learn more about that project here, including the open source code: https://18f.gsa.gov/what-we-deliver/forest-service/ - and a few blog posts about it: https://18f.gsa.gov/tags/forest-service/

Both that Forest Service project and Federalist run on cloud.gov, which is a Platform as a Service operated by 18F.

(I work for 18F but don't officially represent it here.)

You need to move to DC to work for USDS (https://www.usds.gov/join#relocation), but you can work for 18F from anywhere in the US (https://18f.gsa.gov/join/). The login.gov team includes both USDS and 18F team members (as the blog post says).

Also the salary cap is $161,900, as shown in this pay grade table for SF: https://www.opm.gov/policy-data-oversight/pay-leave/salaries... Not a top-of-the-line private sector salary, but not some kind of hardship.

I work at 18F but am speaking about my experience here and experiences heard from coworkers, not officially on behalf of 18F.

Applying to 18F didn't require a drug test for me. The application process asks this question in the background check part (details at "security clearance" at the bottom of https://pages.18f.gov/joining-18f/how-to-apply/ for anyone curious), but answering yes to the "have you ever done federally-illegal drugs" question is not an automatic disqualification. It can be a disqualification depending on the details of the drug use.

Yes, if a company receiving this letter isn't quite sure what level of detail the NHSTA needs, they could make a polite call to the explicitly-listed "call us if you have questions" person and ask about the expected scope, to make sure they're on the right track before they submit the final documents. Federal regulatory agencies are made of people.

Aw, thanks for posting this and being excited about it! This is a labor of love organized by my friend Noah with help from a few other friends and me, just for fun since we love playing roguelikes.

We did sell out of tickets; our very graciously donated venue (thanks to Eventbrite) has an attendee limit since it's basically an office rather than a large venue. I hope you all will watch the streams! The talks will also be recorded so you can watch them later.

That's just the schedule page - sorry this is confusing! The intro page (https://roguelike.club/) has the date - tomorrow (Saturday), September 17.

You might find this chart interesting for thinking about the uses of discussion outside of courts: https://twitter.com/scriptjunkie1/status/739545814150746113

I'd encourage you all to read the draft source code policy (https://sourcecode.cio.gov/) and consider commenting on it or responding to existing comments yourself (due by Monday night at midnight Eastern U.S. time): https://github.com/WhiteHouse/source-code-policy/issues

This is an unusually direct opportunity to share your expertise to help the entire U.S. federal government do a better job of building software.