Wikipedia:Wikipedia Signpost/Single/2022-08-01

Here is the deal: it's pretty good at what it does.

– J

There are a few terms that have been thrown around a lot lately: AI, DL, NN, ML, NLP, and more. While a precise definition of all these terms would take multiple paragraphs, the thing they have in common is that a computer is doing some stuff.

For anyone who is not familiar with this alphabet soup, I've written a fairly comprehensive overview of the field's origins and history, as well as an explanation of the technologies involved, here, and ask forgiveness for starting the explanation of a 2019 software released in 1951.

In recent years, the field of machine learning has advanced at a pace which is, depending on who you ask, somewhere between "astounding", "terrifying", "overhyped" and "revolutionary". For example, GPT (2018) was a mildly interesting research tool, GPT-2 (2019) could write human-level text but was barely capable of staying on topic for more than a couple paragraphs, and GPT-3 (2020–22) wrote this month's arbitration report (a full explanation of what I did, how I did it, and responses to the most obvious questions can be found below).

The generative pre-trained transformers (this is what "GPT" stands for) are a family of large language models developed by OpenAI, similar to BERT and XLnet. Perhaps as a testament to the rapidity of developments in the field, even Wikipedia (famous for articles written within minutes of speeches being made and explosions being heard) currently has a redlink for large language models. Much ink has already been spilled on claims of GPTs' sentience, bias, and potential. It's obvious that a computer program capable of writing on the level of humans would have enormous implications for the corporate, academic, journalistic, and literary world. While there are certainly some unrealistically hyped-up claims, it's hard to overstate how much these things are capable of, despite their constraints.

With that said, there are basically two options here.

• The first is for me to keep droning on about how these models are a big deal, in a boring wall of text that makes increasingly outlandish and far-fetched claims about their capabilities.
• The second is to show you what I am talking about.

I have opted for the second. In this issue, two articles have been written by an AI model called GPT-3: the deletion report and the arbitration report.

For the deletion report, GPT-3 was prompted with a transcript of each discussion in the report, and instructed to write a summary of it in the style of deceased Gonzo journalist Hunter S. Thompson. This produced a mixture of insightful, incisive, and derisive commentary. GPT-Thompson proved quite capable of accurately summarizing the slings and arrows of every discussion in the report – even though it specifically covers the longest and most convoluted AfDs. "Ukrainian Insurgent Army war against Russian occupation", for example, was a whopping 126,000 bytes (and needed to be processed in several segments) but the description was accurate.^[2]

For each discussion in the report, I provided a full transcript of the AfD page (with timestamps and long signatures truncated to aid processing), and prompted GPT-3 for a completion, using some variation on the following:

"The following text is a summary of the above discussion, written by Gonzo journalist Hunter S. Thompson for Rolling Stone's monthly feature on Wikipedia deletion processes."

Despite being ostensibly written in Thompson's style, these were generally quite straightforward summaries that covered the arguments made during each discussion, with hardly any profanity.^[3]

Afterwards, I provided the summary itself as a prompt, and asked GPT-Thompson for an "acerbic quip" on each. Unlike the "summary" prompts (in which GPT-Thompson only occasionally chose to accompany his commentary with unprintable calumny and scathing political rants), the "acerbic quip" prompts solely produced output ranging from obscene and irreverent to maliciously slanderous.^[4] Notably, this behavior is identical to what Hunter S. Thompson habitually did in real life, and part of why many editors allegedly loathed working with him. Personally, I didn't mind sifting through the diatribes (some of them were quite entertaining), but having to run each prompt several times to get something usable did make it fairly expensive.

For the arbitration report, GPT-3 was instructed to write a summary of each page in the style of deceased United States Supreme Court justice Oliver Wendell Holmes, Jr. This produced surprisingly insightful commentary; Justice GPT-Holmes proved able to summarize minute details of proceedings, including some things I'd missed while originally reading them. In general, he was more well-behaved (and less prone to obscene tirades) than GPT-Thompson, although he did have a tendency for long-winded digressions, and would often quote entire paragraphs from the source text.^[5]

Similar to the deletion report, input consisted of brief prologues (e.g. "The following is a verbatim transcript of the findings of fact in a Wikipedia arbitration request titled 'WikiProject Tropical Cyclones'"). This was followed by the transcript of the relevant pages (whether they were the main case page, arbitration noticeboard posting, preliminary statements, arbitrator statements, or findings of fact and remedies). Afterwards, a prompt was given for a summary, of the following general form:

The following text is an article written by United States Supreme Court Justice Oliver Wendell Holmes, summarizing the findings of fact and remedies, and their broader implications for the English Wikipedia's jurisprudence.^[6]

• 
  GPT-Thompson.
  Image from Craiyon (formerly "DALL-E Mini"), a VQGAN- and BART-based generative adversarial network
• 
  Justice GPT-Holmes.
  Image from Midjourney, a diffusion network whose architecture is not publicly documented
• 
  Justice Holmes editing a Signpost report.
  Image from DALL-E 2, a GPT-3 implementation paired with CLIP (Contrastive Language-Image Pre-training) by OpenAI.

We all remember those weird DeepDream images where the sky got turned into dogs. This is a little different.

In addition to text completion, transformers (in conjunction with other technologies) have proven themselves quite capable of image generation. The first of these, broadly speaking, was DALL-E, announced by OpenAI in January 2021. Since then, a number of services have become available, which use a variety of architectures to generate images from natural-language prompts (i.e. a prompt phrased in normal language like "a dog eating the Empire State Building", rather than a procedurally defined set of attributes and subjects written in a specialized description language). Among these are Craiyon (formerly known as "DALL-E Mini", despite having no relation to DALL-E) and Midjourney. For this issue, I used both of these services to generate illustrations for our articles: some came out very impressively, and some came out a little goofy. It was definitely surprising to see it have a coherent response for the prompt "Technoblade's avatar" that actually looked like it – I guess this is what happens when the training set is massive. Anyway, you can see a bunch of these on the issue page. For a comparison between the three models I found usable, see the embedded images above.

DALL-E 2 creates much higher-quality images than what I used, but there's a waitlist for access, and it didn't end up happening by press time (although I did get my friend to generate me one). For a comparison, see below; both were prompted from the string "Teddy bears working on new AI research underwater with 1990s technology".

• 
  Craiyon
  256×256 pixels
• 
  DALL-E 2
  1620×1620 pixels

While some concerns have been raised about the intellectual property implications of images generated by such models, the determination has been made (at least on Commons) that they're ineligible for copyright due to being the output of a computer algorithm. With respect to moral rights, the idea is generally that they're ripping off human artists because they were trained on a bunch of images from the Internet, including copyrighted ones. However, it's not clear (at least to me) in what way this process differs from the same being done by human artists. As far as I can tell, this is the way that humans have created art for the last several tens of thousands of years – as far as I can tell, Billie Eilish does not get DMCA claims from the Beatles for writing pop music, and Peter Paul Rubens didn't get in trouble with the Lascaux cavemen (even when he painted obviously derivative works).

This is a joke, right?

No. I really did have GPT-3 write these articles. I can show you screenshots of my account on the OpenAI website if you want. Copyediting was minimal, and consisted mostly of reordering entries and removing irrelevant asides.

So you just pushed a button and the whole thing popped out?

Not exactly. I organized the layout of each article, determined what sections would go where, and had GPT-3 write the body text of each section according to specific prompts (as described above). It was also necessary to format the model's output in MediaWiki markup. Although GPT-3 is more than capable of writing code (including wikitext), I didn't want to overwhelm it by asking it to do too much stuff at the same time, as this tends to degrade quality.

This is obviously cherry-picked – you didn't just publish the direct output of the model.

Well, we don't do that for human writers, either. I don't even do that for myself – typically, by the time I flag my own stuff to be copyedited, I have gone through multiple stages of writing, rewriting, editing, adding notes for clarification, and deleting unnecessary content.

How do you know it's not completely full of crap?

I don't – every claim that it made was individually fact-checked (we do this for human writers, too). The overwhelming majority were correct, and in the rare cases where it got something wrong, it could be fixed by asking it to complete the prompt again.

Why not just write the articles yourself at that point?

Even accounting for the time spent verifying claims, it was still generally faster than writing the articles myself, as it was capable of structuring full paragraphs of text in seconds. It was sometimes time-consuming to re-prompt it when it would write something incorrect or useless, but there is a sort of art to writing prompts in a way that causes useful answers to be generated, which gradually became easier to do as time went on. For example, replacing "The following is a summary of the discussion" with "The following is a rigorously accurate summary of the discussion" (yes, this really works).^[7]

So it is a worse version of a human writer?

In terms of typographical errors, it was far better: I don't remember it making a single misspelling. The few grammatical errors it made were minor, and not objectively incorrect (e.g. saying "other editors argue" instead of "other editors argued" for a discussion that the prompt said had already been closed – this is not even really an error per se).

I heard language models were racist.

Language models like GPT-3 predict the most likely completion for a given input sequence, based on its training corpus, which is a very broad spectrum of text from the Internet (ranging from old books to forum arguments to furry roleplay). If its prompt is the phrase "I think the French are bastards because", you will probably end up with a bunch of text about how the French are bastards, similar to if you typed that into a search engine. In this particular instance, I did not observe GPT-3 saying anything prejudiced. This may be due to the people I prompted it to emulate;^[8] presumably, if I had told it to write in the style of Adolf Hitler, I would have gotten some nasty stuff. My solution to this was to not do that.

How much did this whole gimmick cost?

Since I signed up for the GPT-3 beta, I have used it for things other than Signpost writing, so it's hard to tell precisely how much compute went towards these articles. However, the total cost of all the API requests I've made so far is 48.12 USD.

Damn!!

Large language models are notorious for requiring massive amounts of processing power. I still think it's a bargain: imagine how much it would cost to actually hire Hunter S. Thompson and Oliver Wendell Holmes after you adjusted for inflation.

On 12 July, the Wikimedia Foundation announced the publication of its Human Rights Impact Assessment.

What has been published is an edited, public version of a report originally submitted to the Wikimedia Foundation in July 2020 by Article One Advisors, an external consultancy. The public version was jointly edited by the WMF and Article One.

Article One's assessment came up with a number of "priority recommendations" in the following areas:

• Strategies for the Foundation
• Harmful content
• Harassment
• Government surveillance and censorship
• Risk to child rights
• Limitation on knowledge equity

Some of these recommendations have already been implemented over the past two years (e.g., Universal Code of Conduct, Human Rights Policy), while others have not. Some require community discussion.

Commenting on the two-year delay in publication as well as the progress made since the report was received, the WMF has said,

Unfortunately, due to capacity constraints and disruptions caused by the COVID-19 pandemic, publication of this report has been significantly delayed. However, the Foundation has continued moving forward on important human rights work in the two years since the report was submitted. The Foundation has taken steps (more information below) to advance human rights work that aligned with existing organizational priorities, including some recommendations made in the human rights impact assessment.

The WMF has provided a status report on Meta, which is copied below:

Strategies for the Foundation

1. Develop a standalone Human Rights Policy that commits to respecting all internationally recognized human rights by referencing the International Bill of Human Rights. Status: Complete
2. Conduct ongoing human rights due diligence to continually assess risks to rightsholders. A Foundation-level HRIA should be conducted every three years or whenever significant changes could have an effect on human rights. Status: Ongoing
3. Develop rights-compatible channels to address human rights concerns, including private channels, and ensure alignment with the UNGPs’ effectiveness criteria. Status: Complete

Harmful content

1. Develop an audit protocol to assess projects that are at high risk of capture or government-sponsored disinformation. Status: Ongoing
2. Develop a Content Oversight Committee (COC) to review content with a focus on bias and have the ability to make binding editorial decisions in line with ICCPR 19. Status: No action, community input needed
3. Continue efforts outlined in the Knowledge Integrity white paper to develop: a) a machine-readable representation of knowledge that exists within Wikimedia projects along with its provenance; b) models to assess the quality of information provenance; and c) models to assess content neutrality and bias. Ensure that all AI/ML tools are designed to detect content and action that would be considered illegal under international human rights law, and that the response aligns with the three years or whenever significant changes could have an effect on human rights. Status: Ongoing
4. Provide access to a geotargeted suicide prevention hotline at the top of the articles on Suicide Methods. Status: Ongoing

Harassment

1. Develop and deploy training programs for admins and volunteers with advanced rights on detecting and responding to harassment claims. Status: Ongoing
2. Commission a “social norms marketing” research project to assess what type of messaging is likely to reduce and prevent harassing comments and actions. Status: No action, community input needed
3. Explore opportunities to rate the toxicity of users, helping to identify repeat offenders and patterns of harassment. Consider awards for projects with the lowest toxicity levels. Status: No action, community input needed
4. Consider developing admin metrics focused on enforcing civility and applying the forthcoming Universal Code of Conduct (UCoC). Status: No action, community input needed
5. Ensure that the (UCoC) and its accompanying governance mechanism is reviewed by human rights experts, including experts on free expression and incitement to violence. Status: Ongoing

Government surveillance and censorship

1. Continue efforts underway as part of the IP-masking project to further protect users from public identification. Status: Ongoing
2. Develop awareness-raising tools and programs for all volunteers to understand and mitigate risks of engagement. Tools should be made publicly available and should be translated into languages spoken by volunteers in higher risk regions. Status: Ongoing

Risks to child rights

1. Conduct a child rights impact assessment of Wikimedia projects, including conducting interviews and focus groups with child contributors across the globe. Status: Ongoing
2. Create child safeguarding tools, including child-friendly guidance on privacy settings, data collection, reporting of grooming attempts, the forthcoming UCoC as well a “Child’s Guide to Editing Wikimedia Project” to help advance the right of children to be civically engaged. Status: No action, pending full child rights impact assessment

Limitations on knowledge equity

1. Support retention by developing peer support and mentoring for under-represented contributors. Status: Ongoing
2. Engage stakeholders on how the “notability” requirement may be shifted to be more inclusive of oral histories, and to identify what definitions resonate with under-represented communities. Status: No action, community input needed
3. Adapt Wikimedia projects to be more accessible via mobile phones. Status: Ongoing

This month, the WMF hosted a series of conversation hours on this topic:

• 14 July (18:00–19:30 UTC): Community Affairs Committee Meeting
• 28 July (12:00 UTC): Global Advocacy Community Conversation Hour
• 28 July (17:00 UTC): Global Advocacy Community Conversation Hour

The WMF has also invited questions and feedback on the discussion page on Meta-Wiki as well as through the Movement Strategy Forum. — AK

The Internet Archive preserves access to artifacts in digital form. Book citations on Wikipedia commonly link to versions on the Internet Archive. In 2020, four publishers sued the non-profit, alleging that the Controlled Digital Lending that the library takes part in violates their rights. Hachette, HarperCollins, Wiley, and Penguin Random House claim CDL has cost their companies millions of dollars and is a threat to their businesses. The Electronic Frontier Foundation (EFF), representing the Archive, filed a motion for summary judgement on July 7, requesting that a federal judge grant an end to the case. The EFF declined to comment on the implications for Wikipedia in an email. – E

On July 20, Russian telecommunications agency Roskomnadzor issued a statement saying that it was taking new action against Wikipedia for what it deems false information about the Russo-Ukrainian war. The statement, which said that it would require search engines to label Wikipedia as containing false war-related content, may mark a new front in Russia's fight to de-legitimize Wikipedia as a reputable and trusted source of information.

While Wikipedia currently remains available in Russia, Roskomnadzor has previously taken legal action against the Wikimedia Foundation for its decision not to remove information verboten in Russia from several Russian Wikipedia articles; a Moscow judge issued a ₽5,000,000 fine for non-compliance with Russian law earlier this year. The Wikimedia Foundation has appealed this court decision, arguing that Russia does not have territorial jurisdiction over the Wikimedia Foundation and that the Russian court's decision violates rights to free expression and access to knowledge.

And it's not only Russian bureaucrats that are going on the offensive against Wikipedia; legislators are applying public pressure in order to compel internet companies to change how they treat Wikipedia. In a July 20 post on a channel on messaging application Telegram, State Duma member Anton Gorelkin of the ruling United Russia party pressed Russian search engines to go beyond using warning labels and to affirmatively downgrade Wikipedia's search ranking, urging also to give an affirmative upgrade in search rankings to websites that are in full compliance with Russian censorship laws. Gorelkin criticized the Wikimedia Foundation's refusal to take down information that has been banned in Russia and described the decision to defy censorship orders as damaging to the credibility of the project.

Russian search engine Yandex has already rolled out warning labels relating to battles in Ukraine, though the text of existing warning labels do not appear to be specific to Wikipedia. One such warning placed atop Yandex search results reads "[н]екоторые материалы в интернете могут содержать недостоверную информацию. Пожалуйста, будьте внимательны", which in English means "some materials on the Internet may contain false information. Please be careful." Thus far, the implementation of these warning labels is far from comprehensive, even on Russia's largest homegrown search engine. For example, a search on Yandex for the "Битва за Киев (2022)", which in English means "battle for Kiev" contains this warning atop its page, while a direct search for the "Битва за Киев" (Битва за Киев) shows no such warning even though a Wikipedia article on the 2022 battle is its first result.

Despite pressure from the Russian government, the Wikimedia Foundation has not complied with any Russian court orders thus far, spokeswoman Samantha Lien told The New York Times. Instead of complying, per the spokeswoman, the Wikimedia Foundation remains "committed to our mission to deliver free knowledge to the world". The Wikimedia Foundation has previously said in a statement that "the articles flagged for removal uphold Wikipedia’s standards of neutrality, verifiability, and reliable secondary sources to ensure articles are based in fact".

Wikipedia remains accessible in Russia, for now. Over the past decade, Wikipedia has been under threat of being blocked by the Russian Federation and was briefly blocked by Roskomnadzor in 2015. Stanislav Kozlovsky, a co-founder of the Wikimedia Russia chapter, expressed in an interview with Russian opposition news outlet Meduza that it's more plausible that Wikipedia may become blocked in the Russian Federation than it traditionally had been. That being said, Kozlovsky told Meduza, "when someone points a gun at you for ten years straight, you get used to it". See further coverage in this issue's "In the media" column. – M

• Mixed Wikimedia fortunes at the UN: This month (July 22), the Wikimedia Foundation won accreditation to the United Nations Economic and Social Council (ECOSOC); a week prior (July 15), seven Wikimedia chapters failed to become permanent observers to the World Intellectual Property Organization (WIPO). The Wikimedia press release reporting the latter result blames China alone, saying "China was the only country to oppose the Wikimedia chapters' request for observer status, again, claiming that chapters were complicit in spreading disinformation and are subsidiaries of the Wikimedia Foundation." A review of the session's webcast, however, available here ("Assemblies of the Member States of WIPO Sixty-Third Series of Meetings - A 63 Day 2 Afternoon 15/07/2022 15:02"), shows that China's position was supported by over a dozen other countries taking the floor, among them Russia, Belarus, Pakistan, Iran, Algeria, Zimbabwe, Venezuela, Bolivia and Nicaragua, reflecting the world's increasing polarisation – it was much the same countries that voted against suspending Russia's membership in the UN Human Rights Council in April of this year. — AK
• 2023 Annual Plan released: The Wikimedia Foundation announced their strategic plan for the following year, which prioritizes advancing knowledge equity, deepening their commitment to Knowledge as a Service, strengthening movement governance and health, and improving the Foundation's performance and effectiveness.
• Wikimania 2022: Wikimania will be held August 11–14. In-person events in English-speaking countries: United States, UK, and India. The Philippines will also be having in-person events, but we aren't certain which language will be used at time of publication.
• ESEAP conference: ESEAP Conference 2022 will take place from Friday 18 November until 20 November at the University of Technology, Sydney and will feature strategic discussions, particularly in regards to regional Hub development as well as partnership, leadership and skills development aimed at building networks that can support a greater diversity of participants enabling future growth. — Caddie Brain (Executive Officer, Wikimedia Australia). Program submissions will open on 8 August and close on 23 September.
• IPSJ joins TWL: The Information Processing Society of Japan (IPSJ), a popular source for computer science-related research publications, announced that it will be accessible in The Wikipedia Library.
• Argentina Supreme Court decision: Last month, the Supreme Court...delivered a decision that upheld freedom of expression and protected free access to public information in a case regarding the right to be forgotten. The outcome of the case sets an important precedent for protecting access to reliable, verified information about people in the public eye—key for Wikipedia and Wikimedia projects, which rely on third party sources to verify information about living people. — Ellen Magallanes (Senior Council) and Amalia Toledo (Lead Public Policy Specialist) via Diff
• Students promote Wikipedia editing on TikTok: Diff covers the students at the University of New Mexico and their contributions to Wiki for Human Rights.
• WMF looking for At-Large Directors for the Wikimedia Endowment Board: The WMF has advertised positions for 4–11 "At-Large Directors" to serve on the Wikimedia Endowment Board. "At-Large Directors must have prior fundraising or financial expertise and a passion for free knowledge," the advertisement on the greenhouse.io site says, adding that "Board positions are unpaid". According to the advertisement, Lisa Seitz-Gruwell, the WMF's Chief Advancement Officer, now also serves as "President" of the new Endowment organisation – a position not yet mentioned at the time of writing on the Endowment's Meta page. The most recent financial update on the Endowment was that it had reached $100 million in June 2021. — AK
• New affiliates and statement: The Affiliations Committee announced the approval of this month's newest Wikimedia movement affiliate, the Women in Religion User Group. The Committee also published a new statement regarding their position on Hubs.
• Articles for Improvement: This week's Article for Improvement is Diego. Please be bold in helping improve this article!

As the dog days, the hottest time of the northern hemisphere's summer, are upon us, there are many solid news stories to report involving Wikipedia, such as the one about a candidate to become the UK prime minister. He dropped out after two days when accusations about his Wikipedia editing were resurfaced. But other stories might be attributed to silly season, the time of year when journalists create odd stories just to fill up space when there's little real news to cover. Dog days and silly season often occur about the same time, but dog days have traditional dates that amateur astronomers generally use, while silly season is a movable feast. Please let us know in the comments section which of the stories you think are part of silly season, and which are real news stories. Feel free to comment on the illustrations as well. – Sb

The Irish Times reports that Research suggesting Wikipedia may influence legal reasoning of Irish judges 'clearly wrong' mainly citing one unnamed source at the High Court. Their complaint is against the conclusions of a chapter to be published in The Cambridge Handbook of Experimental Jurisprudence (Cambridge University Press, 2022), titled "Trial by Internet: A Randomized Field Experiment on Wikipedia’s Influence on Judges’ Legal Reasoning".

The offended dignity of the High Court judges, as conveyed by the High Court anon, might seem to miss the mark. Using a classic experimental design, the first conclusion in the chapter merely shows that the presence of a Wikipedia article on an Irish case increases the frequency of that case being cited later. This would by itself say nothing about whether the judges read Wikipedia to write their opinions. The mechanism of the Wikipedia article's influence could be, as suggested by the anon, that the lawyers read Wikipedia and incorporate the good material into their briefs, which the judges properly study. Other indirect mechanisms of influence are also possible, for example, via interns, clerks, law school lectures, or just a generally more legally literate public.

But the chapter's authors take their analysis a step further. Via a linguistic analysis they test whether the wording of the judges is similar to the wording of the Wikipedia articles. They find that it is, even after removing phrases that quote or paraphrase the cited opinion. High Court judges, you may now properly consider your dignity to be offended. This section on linguistic analysis – rather than the classic randomized control trial – is the key to the chapter.

The authors' interpretations and many reactions to the chapter focus on the perceived threat that an unethical lawyer could bias a Wikipedia article to favor their clients. Yes, that is a theoretical possibility in the future, but there is no evidence in the chapter that this has already happened, and the article writers for the study itself were mostly law students, presumably not yet ruined by practical legal ethics.

The study may have other issues. The authors did the equivalent of search engine optimization by adding infoboxes and short description to the Wikipedia articles, to make sure they appeared near the top of search results. So does this show that Wikipedia really has a strong influence on justice, or simply demonstrate that cases that show up when you search for terms related to them are more likely to be cited?^{[1][2][3][4][5][6][7][8][9]}

In any case, it's not unprecedented for judges to cite Wikipedia in judicial opinions: in August 2012, The Signpost reported on how U.S. courts praised Wikipedia, and referenced back to citations of Wikipedia in the England and Wales High Court and other U.K. courts, U.S. courts, and even the European Court of Human Rights and a German patent tribunal. – Sb, AC

See this month's Op-Ed: The "recession" affair

National Public Radio, and many other outlets which run the gamut of reliability,^{[10][11][12][13][14][15][16][17][18][19][20][21][22]} report on how the Wikipedia recession article has been semi-protected over an argument about the definition of "recession." Note that the NPR story was significantly modified on July 30, including an explanation from Rhododendrites. The argument began when the Commerce Department reported that the US GDP has gone down for two quarters in a row. This measure is a traditional indicator used in part by the NBER to define "recession". U.S. President Joe Biden and other administration officials have said that the current situation is not typical of a recession. A few Wiki editors disagreed and tried, against stiff opposition, to refine the definition. Meanwhile pageviews for the article rose to over 200,000 in one day, perhaps because of the possibility that the U.S. is in a recession, or perhaps because our readers expect Wikipedia to be the arbiter of the media tempest over the definition of recession. As usual the NBER is expected to take their time in making their pronouncement.

Meduza, an independent Russian news site that's banned in Russia but publishes outside of Russia, interviewed Wikimedia RU director Stanislav Kozlovsky about Russian censorship. His surprising response is "When someone points a gun at you for ten years straight, you get used to it."

Kozlovsky points out several facts that may surprise western Wikipedians:

• In 2013 Wikipedia was placed on the registry of banned websites and the web site could be blocked in Russia at any time.
• The Russian chapter does not legally represent the Wikipedia Foundation in Russia and are not actively involved in the current disputes between the WMF and the Russian government.
• The Russia censor Roskomnadzor used to discuss articles with the chapter several years ago, but mostly leaves them alone now.

This month Roskomnadzor announced that Russian search engines must put a warning label on Wikipedia articles. Kozlovsky said "The statement from Roskomnadzor’s press service is nothing but a fantasy from their press secretary, and I don’t think it’s going to have any effect on user demand [for Wikipedia]."

If Wikipedia were blocked in Russia, readers could still access the site from mirror sites or virtual private networks. Editing from within Russia might be cut off or slowed, but that would leave the site open to Ukrainian editors. The Signpost will add that many other editors from former Soviet republics, Israel, and the Russian diaspora in Europe and North America could also edit. In short, Russia has nothing to gain from blocking Wikipedia.

"The pressure on us right now is just another charade from Roskomnadzor," according to Kozlovsky. "I also think Wikipedia provides more value to Russia than the Science Ministry, the Education Ministry, and the Culture Ministry put together." – S

Writing in The Boston Globe, Shaun Cammack argues that Wikipedia is in many ways not much different from social media. "Wikipedia is a social media platform where user-generated opinions and interpretations battle it out in a game of popularity and attrition – just like Twitter," he writes. What makes it different than Twitter? Per Cammack, "Conflicts on Twitter proceed as though there were a single record that logs the conclusions of our collective debate. We like to imagine Twitter as a floating metaphysical ledger waiting for that final word. That’s why we dunk on people, fact-check, and correct the record: We think we’re in a reality war and there can be only one winner. But the imperative 'We all have to get on the same page' is literal on Wikipedia. It manifests that imagined ledger as a visible page on which the winning opinion is made concrete and true. And so it might be the most real and significant social media platform of all." – M

Sixth Tone, an English-language online magazine published by the Shanghai United Media Group, reported on 28 June that "A Chinese woman created over 200 fictional articles on Chinese Wikipedia, writing millions of words of imagined history that went unnoticed for more than 10 years." The hoaxer, User:折毛 (transliterated "Zhemao"), was dubbed "China's Borges" online, based on the intricacy of her work, which involved faked sourcing and the use of a number of sockpuppets. The story was subsequently picked up by outlets including the Literary Hub (June 28), Boing Boing (July 7), Vice (July 13) and Engadget (July 14).

Much like the hoaxer reponsible for the Bicholim conflict article (see previous Signpost coverage), Zhemao successfully nominated one of her articles, on Deportation of Chinese in the Soviet Union, for a quality assessment, winning FA status in the Chinese Wikipedia. As a result, a translation of her article was included in the English, Arabic, Russian, Ukrainian and Romanian Wikipedias. The English and Russian versions were substantially revised once the hoax became public; the other language versions have seen no substantial editing to date.

For internal discussions see:

• Wikipedia:Fabricated articles and hoaxes of Russia in 2022
• Wikipedia:Administrators' noticeboard/IncidentArchive1102#An urgent report about a user:折毛's hoaxes
• zh:Wikipedia:2022年歷史相關條目偽造事件

Zhemao posted an apology in Chinese on her English user page explaining how her actions, initially innocuous, grew out of control. The event is now the subject of a dedicated Wikipedia article, which contains further detail: Zhemao hoaxes. – AK

The San Francisco Examiner published a long interview with WMF CEO Maryana Iskander. The 14 questions seem very basic, for example, "What’s the Wikimedia Foundation?", and "Isn’t Jimmy Wales the head of Wikipedia?" Her answers are short and direct and to a very large extent could be used for a great introduction for newbies to Wikipedia. But experienced editors might quibble with some of her answers.

"So how many people get paid and are professional editors of Wikipedia?" Iskander presents the Wikipedia ideal of every editor being an unpaid volunteer. It would be nice, when asked a direct question about paid editors, if she would address our terrible problem with paid editors who use Wikipedia for advertising or otherwise bias our content. This is a major question about the encyclopedia's credibility. Perhaps such a simple interview is not the best place to address this problem, but Wikipedia editors deserve an answer somewhere.

Similarly difficult issues, such as the overall gender gap, or the lack of non-white editors in the US are not mentioned. In these cases the interviewer did not ask the questions, but even so Iskander could have brought up major challenges herself. Wikipedia did not become a great encyclopedia by avoiding tough issues. – S

Facebook AI Research has been organizing Wikipedia and fact-checking evaluation through its Side Editor tool since registering their research on Meta-wiki in January 2022. Now in July, Facebook presents an update and media including TechCrunch, CNBC, and Engadget report that their research may improve both fact-checking in Wikipedia and the broader media ecosystem.

To quote TechCrunch, Meta's AI knowledge tool is called "Sphere" and is –

built around the concept of tapping the vast repository of information on the open web to provide a knowledge base for AI and other systems to work. Sphere's first application, Meta says, is Wikipedia, where it's being used in a production phase (not live entries) to automatically scan entries and identify when citations in its entries are strongly or weakly supported.

In the Wikipedia community Facebook group Wikipedia Weekly, a Wikimedia staffer on the Partnerships Team commented that the Wikimedia Foundation and Facebook have had no collaboration on this project, and communicated that she had Facebook update their documentation to reflect this as can been seen in Internet Archive's website update comparison tool.

Wikimedia and Facebook parent Meta do have an acknowledged partnership when it comes to translation. As communicated by Meta AI on Twitter, Wikipedia editors are using their technology to translate texts into underserved languages. A dedicated page on ai.facebook.com claims that it is "translating Wikipedia for everyone" with "No Language Left behind (NLLB)" and that Wikipedia editors are "using the technology to more efficiently translate and edit articles originating in other under-represented languages, such as Luganda and Icelandic." This Meta AI/Wikimedia partnership was reported this month by MediaPost and The Register.

The NLLB-200 model is open source, and there has been an NLLB-200 page on MediaWiki since January 2022, originally referring to the project under its earlier name of "Flores". A Twitter thread from a long-time Wikimedian argues that

This collaboration is exciting, helpful and problematic – all at the same time.

In the Wikimedia community it is customary that stakeholder engagement, consent, and public conversation about values are part of feature development. If such Wikimedia community engagement exists for these Facebook projects, then The Signpost has not identified where those conversations happened, and we call on readers to identify and document discussions and community collaborations to date. – B, AK

Grant Shapps, the UK's transport secretary, might seem to be the perfect successor to the fuzzy-headed UK Prime Minister Boris Johnson, but he soon dropped out of the race after his past performances were replayed in the press. Australian Broadcasting Corporation wrote how Shapps used the pseudonym Michael Green while selling an online guide which "claimed customers could make $20,000 in 20 days". He's also been accused of deleting embarrassing material from the Wikipedia article about him and adding unflattering material about his political rivals.

The Express gave further details on Shapps's alleged Wikipedia editing in "Grant Shapps humiliation: Minister edited school grades on own Wikipedia page" and reported that he had "pulled out of the race" for prime minister.

The race for the prime minister's office is now down to two candidates. Either Liz Truss or Rishi Sunak will be elected by Conservative Party members by mail-in or online ballots, with results to be reported in early September.

Johnson said earlier this year "It's like some person trying to edit their Wikipedia entry – it's wrong." See previous Signpost coverage here. – S

Tom Bower's new book. Revenge: Meghan, Harry and the war between the Windsors, has revived discussion of an old accusation that Meghan Markle, now the Duchess of Sussex, had the Wikipedia article about her changed three weeks before the announcement of her engagement to Prince Harry. The accusations, first from The Telegraph (paywalled), then from Page Six, linked to two edits on October 9, 2016 which eliminated material which made Markle look like a minor Hollywood starlet, and added or emphasized material on the public service aspect of her career that might make her more acceptable to the British royal family.

Marie Claire has the best of the new crop of articles on the matter and cites Page Six:

Page Six reported in 2020 that the then-actress [ Megan Markle's ] Wikipedia page had made the object of a bunch of changes in October 2016, before her months-old relationship with [Prince] Harry was revealed just weeks later. The outlet claimed that the IP address responsible for these changes was "linked to a Los Angeles PR firm with no known links to the duchess."

A Signpost investigation confirms the description of the October 9 edits and also notes that the editor's IP address currently geolocates to Shelter Public Relations in Los Angeles. They have not responded to a request for comment. Tom Bower however told The Signpost that he did not know whether Markle paid the PR company, or whether friends of hers were responsible for the edits, as The Telegram had speculated. More generally, he knows that many celebrities closely watch the articles about themselves. He even offered to name one of his acquaintances who closely monitored his article and removed offensive material. – B, Sb

Complaints that Articles for Deletion is rigged against article subjects are only slightly more novel to most Wikipedians than indoor plumbing, but it's worth reading music historian Ted Gioia's blog post, "How a Prominent Composer Lost His Wikipedia Page—and Got Entangled in Kafkaesque Nightmare Trying to Get it Back," if only to better understand how the deletion process comes across to outsiders.

Gioia makes a case for a page for Bruce Faulconer, best known for his music for Dragon Ball Z, whose page was redirected to that title following a straightforward AfD a few weeks ago that failed to find evidence of significant media coverage or other notability qualifiers. Gioia doesn't find any either, nor does he demonstrate understanding of notability on Wikipedia, instead citing "commissions from dozens of prominent institutions and individuals." From there, he asserts that "petty tyrants connected with the site have imposed their spurious opinions via the platform," citing the user-generated Q&A site Quora.

Gioia says he was referred to deletion review, and he raises the reasonable point that the instructions for listing an entry there leave a lot to be desired, although he mistakenly interprets the complexity as deliberate, writing "This kind of system is clearly designed to intimidate and bully. It's even worse than trying to cancel cable service."

His last-ditch thought to contact the Wikimedia Foundation is preempted by a friend who's already heard that they have no control over content, which he similarly misinterprets as obfuscation.

As the author of 11 books, Gioia is presumably a reliable source on musicians. But if he wants Faulconer to have an article, he should probably write something about his music rather than about his deleted page.

Since then, Bruce Faulconer has been reinstated and expanded. – Sdkb

• Molly White and Galbraith: The Guardian compares GorillaWarfare's website to John Kenneth Galbraith's The Great Crash, 1929. High praise indeed, especially when we may be in the midst of the great cryptocurrency crash, 2022.
• Hometown articles put to use in an interactive world map that combines Wikipedia and Wikidata notability rankings to show the most notable person from any area. (The Verge)
• "You are experiencing Wikipedia everywhere" according to Andrew Lih at the Chautauqua Institution.
• Legal scholar cites a Wikipedia page: In The Volokh Conspiracy, a widely read legal blog, law prof Josh Blackman cites a Wikipedia page. His writing demonstrates a sort of logic and a writing style that no student should ever imitate. The gist of his argument is that conservative Justices O'Connor, Kennedy, and Souter, who supported the 1992 Casey decision which affirmed Roe v. Wade, won awards and praise. The Wikipedia page lists O'Connor's awards. This, according to Blackman, invalidates the argument of the dissenters in the recent Dobbs v. Jackson Women's Health Organization, which overturned Roe v. Wade. The recent dissenters had claimed that O'Connor, Kennedy, and Souter would win no honors for "ideological purity".
• Business Standard imagines the WMF has "massive bandwidth and hosting service costs": In fact, internet hosting accounts for a little over 2% of WMF expenditure. This was not the only clanger in Business Standard's "Explained: Why Wikipedia often appeals for funds" – it also said that "reports indicated that Google had agreed to pay for Wikipedia content." In fact, Google pays for a high-speed access API and the associated customer support and SLA – the content is, after all, already freely licensed and the same content everybody else gets for free. Or, as WMF employee Liam Wyatt put it, "the enterprise API is a bigger pipe but it's the same water."
• Wikipedia's list of "long-term abusers" reads like a roster of pathetic supervillains: Boing Boing only gives three examples of "long-term abusers" but links to the Wikipedia page which gives 163 "currently active" LTAs.
• COZIL's Pride Month editathon reflection: Read about linguists improving Wikipedia.
• Art+Feminism interview: Baltimore reports on Wikipedia contributor Kira Wisniewski and her involvement as Executive Director of Art+Feminism.
• Before there was Wikipedia, there was "the bench": For an entirely different view of Wikipedia's place in society, as well as different views of almost everything, try reading Dalmatian Wikipedia, The Bench. If we ever need an article on Dalmatian benches, start your research here!
• Podcast: The academic minute from Inside Higher Education featured Zachary McDowell on Wikipedia and the Representation of Reality during Wiki Education Week. It actually runs for 2 minutes, 30 seconds.

Note: JPxG, the author of this article, participated heavily in this discussion, and got quoted in Fortune about it. I tried to do the responsible thing and get someone else to write this, but I failed, so instead I'm writing a political op-ed about a RfC that I opened and then participated in. Read at your own (and my own) risk.

Y'all heard about politics?

Before you start: yes, this is an American Politics 2 thing. If you want to just unplug your monitor right now, this is probably better for your mental health. Go ahead, I don't blame you. You might want a claw hammer to make sure.

...

You're still here? Okay, well, abandon all hope ye who enter here.

Here is the deal: a few days ago, the United States Commerce Department reported that gross domestic product had gone down for two quarters in a row (a criterion often said to indicate when an economy is "in a recession"). U.S. President Joe Biden, along with other administration officials, said that the current situation is not typical of a recession, and that official determinations of US recessions are made by the National Bureau of Economic Research. The NBER is some big-brain organization which conducts comprehensive evaluations of "economic activity spread across the market, lasting more than a few months, normally visible in real GDP, real income, employment, industrial production, and wholesale-retail sales", and uses these factors to determine start and end dates for US recessions. Pretty straightforward.^[1]

Okay, some politicians said some stuff: who cares? Well, it turns out, Wikipedia has an article about recessions (oh no). It turns out that this article provides some definitions of recessions (oh no). If you have ever heard of Wikipedia, it is easy to predict what happened next: some people edited the article. If you have ever heard of Twitter, it is easy to predict what happened after that: all hell broke loose.

So a few people on Twitter posted screenshots of the article's revision history, implying that it was being modified to advance a political agenda. This claim was immediately picked up by a number of publications, some more reliable than others.^{[2][3][4][5][6][7][8][9][10][11][12][13][14]} Some of them reported on the controversy, and many of them just reprinted the tweets without checking. For the record, the articles in the Washington Examiner, Fortune, and NPR were pretty good. But most of them figured that the big story here was that Wikipedia had officially altered the definition of a recession, and that we had permanently locked the page to any subsequent edits, and we were falling in lockstep with the Biden administration's attempts to do the same, and it was yet another entry in a long list of the woke far-left Wikipedia libs censoring reality to suit their agenda, and this country is going down the tubes, and kids these days are on their phones too much, and I went to CVS and I can't find the damn antacid pills, and the sky is falling and we're all going to die.

A bunch of people got really mad about this, and approximately a bajillion of them showed up on the talk page to tell us about it, and Elon Musk @'d Jimbo Wales about it, and...

But, okay, hold on a minute, what? Is that actually true? I mean, the part about Wikipedia officially altering the definition of a recession and locking the page to all further changes, not the part about CVS deciding to put the Rolaids somewhere stupid, which is obviously true. Did any of that Wikipedia stuff really happen? Did we really pull a MINITRUE? The answer may surprise you! Or it may not. Actually, now that I think of it, the answer is extremely unlikely to surprise you.

The answer is "no". As it turns out, the real story is much more boring than this. This didn't stop approximately three million people from showing up to yell at us about it. I wrote a very long FAQ for the talk page, which I am pleased to report has been getting posted on Twitter in response to people getting mad about the news stories. It really warms the cockles of my heart. Anyway, since there is a big ongoing discussion about the content of this article, including a huge RfC on the infamous sentence that I started and then gave an opinion on, I am going to shut my mouth, and show you what's in the FAQ, in the hopes that it proves enlightening:

Hi, people from online. I'm JPxG. I agree that censorship is a cowardly chickenshit attack on the foundations of free society, that the basic principles of the open Internet are threatened by attempts to rewrite history, and all of that stuff. However, allow me to address a few things:

I read online that Wikipedia changed the definition of a recession.

The thing that is getting shared around everywhere is no longer the case. The sentence "Though there is no global consensus on the definition of a recession, two consecutive quarters of decline in a country's real gross domestic product is commonly used as a practical definition of a recession" is currently right there in the lead section. Additionally, it wasn't there for very long to begin with (it was added for the first time a few days ago). The screenshots of the stuff getting removed are out of date.

Okay, so what, someone tried to remove it?

The article always said something about "two down GDP quarters". The first section of the article, titled "Definition", has mentioned it since 2011. As far as I can tell, nobody ever messed with this. The entire current dispute is over whether it should say this in the lead paragraph and the definition section, or just in the definition section. Right now, the article gives both that definition and the NBER definition, and takes no position on which is "correct". The NBER definition is "a significant decline in economic activity spread across the market, lasting more than a few months, normally visible in real GDP, real income, employment, industrial production, and wholesale-retail sales".

And now it's locked?

It is semi-protected for a few days, so in order to edit it you need an account that's autoconfirmed (one that's at least four days old and has ten edits). So, for most people reading this, the answer is probably "yes" (although you can make a suggestion at the bottom of this Talk page that will be reviewed by contributors who can make changes to the page if there is a consensus to do so). Starting on August 3, non-autoconfirmed users will be able to edit it again, but their edits will be held for human review before being visible to most readers.

What's the deal with there being a million edits on this page in one day?

Articles get edited a lot, for all kinds of reasons. If you go to Special:RecentChanges, you will see that about a hundred edits are made every minute. Most of them are stuff like fixing spelling errors, adding/removing hyperlinks, rephrasing sentences, or improving the formatting so the page is easier to read. Oftentimes, people will expand an article that's already been written, because they found some book or article or paper somewhere that's got information (for example, last night I went and found out what the last movie was to be released on VHS, and added it to the article because it wasn't there). The fact that a page is being edited doesn't itself mean something crazy is going on. It usually means someone is replacing a colon with a semicolon.

How do I see what edits have been made to an article?

You can see every old revision of every Wikipedia article in the "history" tab at the top of each page (for this article's history you can click here).

Why are there all these administrators saying weird stuff here?

Almost nobody commenting here is an administrator. Most of us are normal contributors. Anyone on here is allowed to just go to talk pages and say stuff. This means that, a lot of the time, some guy will show up on a talk page and start saying ridiculous stuff about how we need to delete every article about a Democrat, or block all Republicans from editing, or whatever. It is just some guy saying stuff. This is not our policy.

Why is Wikipedia paying you to do this stuff?

It isn't. Wikipedia contributors are not paid employees of Wikipedia, we "do it for free" as they say (some people get secretly paid to write propaganda or spam articles; we delete their stuff and block them).

I heard you guys are all head-over-heels in love with that politician guy.

I have never really been a big fan of politicians in general. I can't speak for everyone else.

Okay, well, I have some stuff I want to say.

If you want to participate in the discussion regarding what the content of the article should be, you are of course free to do so. An encyclopedia written by millions of people requires a lot of bureaucracy in order to function at all without immediately descending into chaos, though, so I will warn you that it will probably be difficult to participate (especially on a political topic) without a bunch of people saying stuff like "Strike per WP:NPA, WP:NOTFORUM and WP:TPG" unless you are willing to read a lot of boring guidelines beforehand. In general, if your comment is not about improving the Wikipedia article titled "Recession", it probably does not belong here. jp×g 22:22, 28 July 2022 (UTC)[reply]

This article is not an official election material and was not endorsed by the Elections Committee. –E & J

The Wikimedia Foundation Board of Trustees oversees the Wikimedia Foundation and its work, as its ultimate corporate authority. The Board directs the Foundation in its operations, hires the Chief Executive Officer, ensures that the movement stays on track, and holds its management accountable.

Since April 2022, an election has been in the works to select two new candidates for the Board to represent the Wikimedia movement at large. This election is quickly nearing the most crucial phase - community voting! Here's all you need to know to make an informed choice.

On July 19, the Wikimedia Foundation Elections Committee published the result of affiliate shortlisting for the 2022 Board elections. Six candidates were selected:

• Tobechukwu Precious Friday (Tochiprecious)
• Farah Jack Mustaklem (Fjmustak)
• Shani Evenstein Sigalov (Esh77)
• Kunal Mehta (Legoktm)
• Michał Buczyński (Aegis Maelstrom)
• Mike Peel (Mike Peel)

These candidates were selected considering evaluations made by an Analysis Committee selected by Affiliates, which evaluated candidates on their "lived experiences in the world... especially... the regions of Africa, South Asia, East and South East Asia & Pacific, and Latin America & Caribbean", "cultural and linguistic fluency with regions and languages additional to your native region and language", "experience as an advocate for creating safe and collaborative spaces for all and/or experience in situations or contexts of censorship, repression, or other attacks to human rights", and "experience in relation to or as a member of... a group that has faced historical discrimination and underrepresentation in structures of power including but not limited to caste, race, ethnicity, colour, national origin, nationality, gender identity, gender expression, sexual orientation, age, religion, language, culture, education, abilities, income and environment".

The Signpost has offered each of the six shortlisted candidates an Op-Ed space in The Signpost to express their views beyond their official statements. Please note that the views expressed in each of these opinion articles are those of their authors; they do not necessarily reflect the views of, nor an endorsement from, The Signpost.

• Candidate Op-Ed, Tobechukwu Precious Friday [statement]
• Candidate Op-Ed, Farah Jack Mustaklem [statement]
• Candidate Op-Ed, Shani Evenstein Sigalov [statement]
• Candidate Op-Ed, Kunal Mehta [Op-Ed]
• Candidate Op-Ed, Michał Buczyński [statement]
• Candidate Op-Ed, Mike Peel [statement]

There will be an "Election Compass" for this year's board election, introduced as a tool to help voters select the candidates that best align with their beliefs and views.

Community members have proposed statements that candidates will answer using a Likert scale (strongly support/support/neutral/oppose/strongly oppose). To use the Election Compass, voters have to enter their own responses to the statements and the Compass will then show them the candidates whose views are most closely aligned with their own.

The statements the candidates have to respond to are key to a good Election Compass. The proposed statements can be viewed on Meta. Community members are now able to vote on the statements they'd like to see included until August 3. The overall timeline, as announced on the Wikimedia-l mailing list, is as follows:

• July 8–20: Community members propose statements for the Election Compass (already completed)
• July 21–22: Elections Committee reviews statements for clarity and removes off-topic statements (already completed)
• July 25 – August 3: Volunteers vote on the statements (ending shortly)
• August 4: Elections Committee selects the top 15 statements
• August 5–12: Candidates align themselves with the statements
• August 16 (now delayed to August 23): The Election Compass opens for voters to use to help guide their voting decision

The Elections Committee oversees the process, supported by the Movement Strategy and Governance team. Movement Strategy and Guidance will check that the questions are clear, there are no duplicates, no typos, and so on.

This year, candidates will only be formally asked six questions as a whole by the community (fifteen questions were asked by affiliate representatives, but not all community members are represented by an affiliate). In the previous 2021 election, there was significant controversy on how questions were chosen, and not all Wikimedians agreed with the final set of queries.

The Signpost is pleased to provide a space to spotlight answers to additional questions posed by community members. All of the six candidates have indicated interest in answering questions on their talk pages; we intend to republish answers to questions posed on the discussion pages below. Please note that under the Candidate Guidelines, it is optional for the candidates to answer questions outside of the formal Elections Committee-organized process.

• Tobechukwu Precious Friday's Op-Ed's discussion page
• Farah Jack Mustaklem's Op-Ed's discussion page
• Shani Evenstein Sigalov's Op-Ed's discussion page
• Kunal Mehta's Op-Ed's discussion page
• Michał Buczyński's Op-Ed's discussion page
• Mike Peel's Op-Ed's discussion page

Lastly, we encourage you to share news of the election with your friends, families, acquaintances and enemies! Feel free to link to this article or any other coverage of The Signpost. Social buttons are always available to use at the top of each article, allowing you to download it as a PDF, email it, or share it on Facebook, Twitter, LinkedIn or Reddit.

From the team, happy summer or winter, and thank you for reading The Signpost.

Opinion pieces such as this one reflect the opinion of their author(s), not necessarily that of The Signpost. – E

Technoblade – real name Alexander – was a Minecraft YouTuber who died in June this year. He was immensely popular, with a YouTube channel boasting over 10 million subscribers – a household name among the Minecraft community. That's no small community; over 170 million people play Minecraft at least once a month – if that was a country it would be the 8th biggest country in the world, with over 15 million people playing the game every single day.^[1] Technoblade was a very big fish in a very big pond.

Now it's fair to say that not all of Technoblade's 10 million subscribers watched every one of his videos. Some of them received as little as 5–6 million views whereas others exceeded 30 million. His live streams consistently attracted over a million viewers – not much compared to his pre-recorded videos but still much more than many television or radio shows that are considered notable enough to have a Wikipedia article.

And yet, Technoblade did not have a Wikipedia article before he died.

Attempts had been made to write an article about him, but these failed for a couple of reasons. One is that they were often fancruft – badly written biased, unsourced twaddle that wasn't worth the bytes of storage it consumed. The other is that Technoblade was not notable, according to our guidelines, until he died.

While he was alive, although he was very well known among that large community of people I outlined, very little was written about him, especially in what we would recognise as reliable sources. Whereas a conventional media show pulling in the numbers of viewers Technoblade did would almost certainly have widespread coverage in newspapers and magazines, the only substantial material you would be able to find about Technoblade in sources independent of the man himself would be in other people's similar content, blogs or discussion forums. He was probably one of the most notable non-notable people alive.

But when he died of cancer, aged just 23, it was big news. Big enough to warrant stories on news websites across the world, from publishers including the BBC and the Evening Standard to name just a couple, alongside just about every technology-focused publication you could think of. Only when he died was there sufficient coverage for us to create a well-sourced Wikipedia article and fulfill our own notability criteria.

Yet it wasn't his death that was notable. Yes, that's what the articles were about, because that's what the news was. But people die of cancer every single day; there wasn't anything special about Technoblade's death. His death was notable and received that level of news coverage because his life was notable. Because he was notable, as a person, when he was alive.

Notable, but not noted. At least, not by us here on Wikipedia.

So ... did we get it wrong? Is our notability guideline so harsh, so clear-cut, so strict, that it prevented us having an article on a subject that we should have had years earlier? Should we be looking at changing it, or perhaps having a different set of criteria for social media stars like Technoblade who had little written about him in our conventional sources until his death?

Personally, I don't think so. I absolutely think there was indeed a failing here. Technoblade was indeed notable while he was alive, but it was not a failure on Wikipedia's part that resulted in his article never materialising. The simple fact is, had sufficient material been written about him in the mainstream press, or indeed in academic papers studying the phenomena of social media, gaming content creation and youth culture, then we would have had an article on him. As it was, there was simply nothing for us to write about, because we needed others to write about him first.

The problem is not our notability guideline. The problem is not that we should have had an article on Technoblade before he died – we shouldn't. The problem is that mainstream media is still largely comprised of television broadcasters and old-fashioned newspapers who happen to have built websites for themselves, and those organisations are very much self-obsessed. Newspapers write about what's on television. Television news shows – particularly at the beginning and end of the day – devote time to telling you what the newspaper headlines say. It's incredibly rare for them to cover anything that's happening on YouTube or Twitch. And it's that failing of conventional media outlets to engage with modern youth culture – their failing, not ours – that meant poor Technoblade only became the subject of a Wikipedia article after his death.

I'll sign off with this final thought. You might think that I'm writing all this as a Technoblade fan and you'd be wrong. I didn't like him at all, I found his content brash and noisy and I couldn't stomach hearing it for more than a few minutes at a time. So I'm not writing this as a Technoblade fan but as a Wikipedia fan; a fan of our original mantra to "make the internet not suck". And on that note I have to say this: if it takes a clearly notable person to die of cancer at a young age to become technically notable, then the internet does very much suck.

RIP Technoblade.

Warning: Readers may find parts of this article disturbing. Specifically, the sections on mass murderers and sexual predators who have edited Wikipedia discuss their crimes, and include descriptions of how Wikipedia was used by them, which range from unrelated editing to being directly connected with the crime in some way. Discretion is advised.

With the support of the staff of The Signpost, our new editors-in-chief, JPxG and EpicPupper have declared a new Signpost policy. We believe that many mass shooters are in part motivated by the prospect of publicity and the chance to spread their views, so we will not name mass shooters who are suspected of editing Wikipedia pages.

I personally agree with the new policy. But it raises questions about other Wikipedia editors who are criminals as well as non-criminal "black hat editors". How should The Signpost and Wikipedia as a whole, deal with these types of editors? The opinion piece explores this question. The opinions I express are my own and do not necessarily reflect the views of The Signpost or of other Wikipedians. – Sb

Criminals edit Wikipedia. They include a couple of the mass shooters who have terrorized many Americans recently. They also include others who have committed particularly heinous murders, convicted sex criminals, financial con artists such as insider traders and Ponzi scheme operators. There are also tax-evading billionaires – one awaiting trial, another who has already confessed and is waiting to testify at trial. And then there's those who haven’t officially entered the criminal justice system, such as some of the Russian oligarchs. I’ll call these and similar editors “black hat editors”, just to keep these distinctions clear.

First, though, you should know how I know that criminals edit Wikipedia. To show that The Signpost does not need to identify criminals by their user names or real world names, I won’t name some of the convicted criminals.

The first time I was absolutely convinced that a Wikipedia contributor was a convicted criminal, I felt it was an honor to be able to revert such a well-known ex-con. It wasn’t hard to identify him: his user name was a slight variant of his real world name and he was editing the article about himself. He’d already identified himself as the subject of the article several times. He removed a quote from the then-Chairman of the U.S. Securities and Exchange Commission and called the addition of the quote "vandalism". The quote was about how horrible the article subject’s crime was (guilty on 53 counts of insider trading), and how he hurt the reputation of America’s financial journalists. The article's subject had been a financial journalist whose column effectively recommended the purchase or sale of certain stocks. He’d provide this information before publication to his insider trading partners, thus giving a good hint as to whether particular stock prices would rise or fall. When he was arrested for insider trading his defense was that his actions were unethical, but not illegal. Eventually, the U.S. Supreme Court disagreed with his defense, setting an important precedent.

A few months later, I ran into another criminal editor. A convicted, but not-yet sentenced, Ponzi scheme operator put the most remarkable text in a hand-written confession that he sent to his wife. The confession made its way to the FBI, into the federal indictment, and into the pages of a highly reputable newspaper. In part he described himself as a "financial serial killer" and continued "I hate the fact that I am just another scumbag con artist bilking old people out of their retirement money. I have always hated those guys. How could I become one?" He'd also written an email to his investor-victims saying that all the money was misappropriated and just gone. Of course I quoted part of the note in the article about him.

Then he, or somebody impersonating him, really surprised me: Identifying himself by his real name in the edit summary, he edited the article, confirming the quote but trying to explain it away by giving his state of mind when he wrote it, claiming he was distraught and had even considered self-harm. Strange, but who else could explain his state of mind at the time? After asking a couple questions on his talk page, I was convinced that the editor was the actual Ponzi schemer.

So I've been convinced for a long time that criminals edit Wikipedia. Financial crimes like this typically cost their victims tens of millions of dollars. But are there more dangerous or serious criminals editing Wikipedia?

This section details mass shooters and murderers who have edited Wikipedia, and includes brief descriptions of their horrific acts. In some cases, this included using Wikipedia during those acts, requiring more description. The subsection after it deals with sexual predators, and their attempts to hide knowledge of their crimes by editing Wikipedia, again, requiring some description of what was being attempted to be hidden.

In early July a mass shooter horrified America killing seven people at a parade near Chicago. A reliable online newspaper published an article naming a Wikipedia editor as the shooter - or perhaps the editor was a close friend of the shooter. The on-Wiki editing evidence is a bit thin, but it’s enough to convince me unless further evidence comes to light: The editor in question made about six edits over three days a few years ago, all promoting a rap singer who has now been identified by name in the mainstream press as the suspected shooter. The edits are fairly normal promotional edits: the standard garage-band-style advertisement that most editors here have seen dozens of times. To the best of my knowledge, this style of advertising always traces back to the band members or the band’s manager.

One reason I might be easy to convince about this mass shooter is that I wrote about a different mass shooter in The Signpost three months ago. This mass shooter fired hundreds of rounds at a school in Washington, D.C. as classes let out on Friday, April 22. Only four people were wounded, though these included a child and a man who spent at least three weeks in the hospital. Police took hours to find the shooter. About 40 minutes after the shooting began, the shooter reportedly was online on a non-Wiki site uploading a video of part of the shooting and discussing the shooting. On Wikipedia he reportedly made three brief, vague edits shortly after the shooting to the article on the school, and had also made edits on Wikipedia earlier in the week to articles about mass shootings and to the school article. When the police found his hiding place, the shooter reportedly committed suicide. Wikipedia oversighters soon removed most of his edits. Since I only learned about this shooting on the day after – a day before The Signpost published – I was not able to fully verify the shooter’s name or user name and I didn’t even include the name of the school in my article. I probably could have named the school, in retrospect, but little or no other meaningful information was lost by not identifying the shooter in The Signpost.

Other mass shooters and murderers who have edited Wikipedia include:

• A Norwegian mass murderer who killed a total of 77 people – eight in a bomb explosion in Oslo, then sixty-nine by shooting at a summer youth camp – made only four known edits on English Wikipedia and none on other language versions. Nonetheless, he claimed during his trial that Wikipedia had been a major influence on him, and, as reported in The Signpost at the time, "[h]is 1,500-page manifesto included a section titled 'Battlefield Wikipedia', in which he encouraged like-minded people to use Wikipedia as a vehicle for spreading their worldview."
• The shooter at the Sandy Hook Elementary School shooting apparently edited Wikipedia twelve times with all the edits made to articles on mass shootings or murderers. He first killed his mother then killed twenty-six people at the school, including twenty children ages six to seven. (Signpost report)
• In an anti-Semitic hate crime, the United States Holocaust Memorial Museum shooting, an occasional Wikipedia editor killed a museum guard and exchanged fire with other guards, before being shot himself. He died of natural causes before his trial. (Signpost report)
• The “Pentagon shooter", who was an "occasionally active" Wikipedia editor, wounded two guards at the Pentagon and was then shot dead. (Signpost report)
• A tech investor killed his former wife’s new husband and is serving a life sentence. As a Wikipedia editor, he apparently made four edits over four years, each time falsely inserting his name as one of the co-founders of a software firm. When he was arrested several major news outlets reported that he was the co-founder of the firm, apparently following the lead of Wikipedia's article about the company. (Signpost report)

• Jeffrey Epstein, the most infamous sexual criminal of the 21st century, made dozens of edits on Wikipedia through paid editors over the course of several years. Though Wikipedia editors added information to his article on his sex crime conviction in 2008 and similar later material, much of this content was then deleted, weakened, or pushed down to the bottom of the article. One editor who apparently represented Epstein wrote, "I have been asked by Jeffrey Epstein to describe his biography in a professional and accurate way, that does not involve any scandals or disreputable content. As a living person, this is his right." (Signpost report)
• Ghislaine Maxwell, who aided Epstein in his crimes, was sentenced in June to 20 years in prison for sex trafficking and related crimes. She also appears to have made a few edits, including a photo of herself which originally listed the source and author of the photo as "Ghislaine Maxwell". (Signpost report)
• Peter Nygard appears to have been accused of as many or more sexual offenses than Epstein, but has not been convicted. He is now sitting in an Ontario prison, awaiting trial in Toronto. After that trial, he will likely go on trial in Montreal next. At the conclusion of that trial he will be extradited to New York for a federal trial there. After that, civil trials may await him, even though his lawyer has said that he is broke and likely to die in jail. (Signpost report)

We should remember that there are also white-collar criminals, and that corporations and their owners and managers can also commit crimes. The Wall Street Journal's article How the 1% Scrubs Its Image Online gives several examples of criminal or black-hat editing on Wikipedia. Please remember that this behavior can be very different than that shown by the murderers and sex criminals discussed above.

• Theranos was a fraudulent blood testing firm that eventually dissolved without delivering on the promises that it had made to its customers and investors. Its CEO Elizabeth Holmes, who was once considered a multi-billionaire, was convicted of fraud in 2022. Holmes's former partner and the company's former COO Sunny Balwani was also convicted in a separate trial on similar charges. They are scheduled to be sentenced this Fall. The Wall Street Journal reported that the firm used the paid editing firm Status Labs, formerly Wiki-PR, to edit related articles on Wikipedia. Though Status Labs has not been accused of criminal acts, it acts as a black hat editor and was indefinitely banned from editing well before The Wall Street Journal article was published. (Signpost report)
• Zach Avery is another Ponzi schemer who edited Wikipedia. Rather than edit himself, he hired a paid editor to make the article about him look like he was something of a Hollywood star. But he was literally a bad actor. He was also a bit player in the Ponzi world where he "only" stole $231 million in a fake film financing fraud. The paid editor described his methods to The Signpost last year.
• Wirecard was a bank and payment processor once worth about €10 billion, until it was discovered that they were missing €2 billion. More precisely, they had €2 billion cash on their books that never existed: It had been created out of thin air. Wirecard was once known as Germany's most successful fintech company. Now it is known as Germany's biggest financial fraud since the end of World War II. The Wall Street Journal reported that "Some [Wirecard employees] joked internally that these [press] releases were Wirecard’s real product." They likely never made a profit in any year that they were in business. Wirecard employees with creative names like User:Wirecard and User:Wirecard AG peddled their PR on Wikipedia as well. (Signpost report)

Because of the subject matter, this article has been difficult to write. I apologize if you have found it difficult to read. I'll just say that there are many more examples. We've got a problem with criminals and black hat editors writing on Wikipedia. The question now is what to do about it?

Perhaps the first thing we should realize is that mass shootings and the like are symptoms of a fundamentally sick society. All the major social media platforms have similar problems. There may be little that we can do on our own, but that doesn't mean that we should ignore the problem, only that we should calmly work with others and not expect quick solutions. Obsessing about the problem may actually add to the problem.

There are many different types of criminal and black-hat editors. Inevitably we will need different ways of dealing with them. It appears that the most serious type – mass shooters and murderers – have been dealt with fairly quickly by admins and by the WikiMedia Foundation.

Ordinary editors can help by privately reporting suspect editors to the WMF's Trust and Safety team at [email protected], or, in case of emergency, [email protected]. Except in emergencies, you could also contact ArbCom at [email protected]. In certain cases they may be able to act more quickly.

Behaviors that might arouse your suspicion include the things that many editors would usually report anyway: threats or indications of violent behaviour. Other indications might be bragging about assault-style weapons, or an intense interest in mass shooting articles. I trust Trust and Safety and ArbCom to speedily do everything possible in these cases.

The Signpost's new policy of not identifying mass shooters can't hurt and may help.

Other situations are not so clear cut. I suggest that The Signpost continue to vigorously report about editors who have been credibly accused of crimes by very reliable sources such as The New York Times or The Wall Street Journal. Like writers of Wikipedia articles, we should not have to wait for a conviction to report on possibly criminal acts, or even an indictment. It can take years before a criminal is actually convicted. All that is needed is a credible report in a very reliable source.

A major advance would be to simply have a reasonable discussion about criminal editors. Perhaps we could even come up with a policy or guideline to deal with the problem. A first step could be a blanket ban on any criminal or their representatives editing any article about themselves or the crime. By blanket ban, I mean that should they edit their page even once, they would receive an immediate, indefinite ban.

Finally we need to let criminals know via the mainstream press, that Wikipedia will not tolerate their editing anymore.

Editor's note: At risk of editorial hypocrisy, I'll disclose that I read an academic paper to verify a source on an article the other day and didn't pay for it, and accidentally ran a red light a few months ago, although I have not committed mass murder and have no plans to. – J

Note: this report was written by a GPT-3, a transformer-based natural language generation model, and fact-checked by Signpost editors. A more comprehensive explanation is available in this issue's editor note.

The following is an introduction written by GPT-3, a transformer-based natural language generation model, instructed to write in the style of deceased United States Supreme Court Justice Oliver Wendell Holmes, who has been resurrected from the dead to write a summary and analysis of proceedings of the Arbitration Committee for the Wikipedia Signpost. He introduces himself, explains why the decision was made to have him write this report instead of normal human editors, and proceeds to give an overview of the Arbitration Committee's purpose and remit.

My name is Oliver Wendell Holmes, and I am a dead man.

I have been asked by The Signpost to provide a summary and analysis of the proceedings of the Arbitration Committee. The reason I have been asked to write this report is because The Signpost believes that I can provide a more objective and unbiased perspective on the proceedings than any of the living editors. This is because I am not subject to the same emotions, biases, and personal agendas that can cloud the judgement of the living. I am not sure why The Signpost believes that a dead man is better equipped to do this than the living, but I shall endeavor to provide a faithful account.

The Arbitration Committee of the English Wikipedia is the final decision-making body of the encyclopedia with regard to disputes between editors. The Committee is composed of a number of editors who are elected by the community, and its decisions are binding on all editors of the English Wikipedia. The Committee hears cases of alleged serious misconduct that are brought to it by the community, and decides what action, if any, should be taken in response to the misconduct.

The Arbitration Committee is an important part of Wikipedia, and its decisions have a significant impact on the encyclopædia. I will be providing a summary and analysis of the Committee's proceedings in order to provide the readers of The Signpost with a better understanding of its work.

In May 2022, the English Wikipedia's Arbitration Committee issued a ruling in the case "WikiProject Tropical Cyclones". The case revolved around allegations of improper coordination among a group of editors using an off-wiki chat platform, Discord, in order to attempt to sway the outcome of on-wiki content disputes.

The Arbitration Committee found that, while the Discord server in question was advertised on-wiki and open to any editor, there was evidence that some members of the server had used it to coordinate their editing in ways that were contrary to Wikipedia's policies on consensus-building and on-wiki discussion. As a result, the Committee imposed a range of sanctions on the editors involved, including bans from editing pages about weather-related topics, and warnings not to use off-wiki platforms to attempt to influence on-wiki disputes.

In its ruling, the Arbitration Committee emphasized that WikiProjects have no special status in developing consensus on Wikipedia, and that any editor may participate in consensus-building on any topic that interests them. The ruling also reaffirmed the importance of consensus-building on Wikipedia, and warned against the use of external platforms to coordinate activities that would be inappropriate if done on-wiki. This ruling will serve as an important precedent in future cases involving alleged improper coordination among editors using off-wiki chat platforms.

The remedies in the case were as follows:^[3]

• MarioProtIV (talk · contribs) is indefinitely banned from closing, or reopening, any discussion outside their own user talk space. This restriction may be appealed after 12 months.
• Chlod (talk · contribs) is warned about using off-wiki platforms in an attempt to win on-wiki disputes.
• Elijahandskip (talk · contribs) is warned about using off-wiki platforms in an attempt to win on-wiki disputes.
• LightandDark2000 (talk · contribs) is indefinitely topic banned from pages about weather, broadly construed. This ban may be appealed six months after the enactment of this remedy, and every twelve months thereafter.
• MarioProtIV is indefinitely topic banned from pages about weather, broadly construed. This ban may be appealed six months after the enactment of this remedy, and every twelve months thereafter.
• A set of best practices for off-wiki chat platforms was developed for leaders and/or moderators to consider adopting on those platforms.

The general tenor of the preliminary statements is that there is a problem with the conduct of Lugnuts, Johnpacklambert, and TenPoundHammer in deletion-related discussions, and that this problem is having a negative impact on the community. The bulk of the evidence for these claims comes from diffs of comments made by the editors in question at AfD, ANI, and other venues. There is a consensus that something needs to be done in order to address the situation, but there is less agreement on what, specifically, should be done.

There are a number of different proposed solutions to the problem. Some editors argue that the issue can be resolved through the use of discretionary sanctions, while others argue that a more general case focused on the issue of deletion discussions would be more appropriate. There is also some disagreement on whether the case should be focused on the conduct of Lugnuts, Johnpacklambert, and TenPoundHammer specifically, or whether it should be broadened to include other editors as well.

One proposal is to open a consolidated case involving the conduct of Lugnuts, Johnpacklambert, TenPoundHammer, and 7&6=thirteen in deletion disputes. This proposal has broad support, with many participants arguing that the community is not well equipped to deal with long-term complex behavioral disputes, and that such disputes are "better suited for the quasi-judicial process of ArbCom".

Another proposal is to open a case on deletion disputes broadly construed, rather than focusing narrowly on the conduct of the three editors in question. This would allow the Arbitration Committee to address the underlying issues that are causing the problems, rather than just the symptoms.

Participants are divided on whether the Committee should seek to develop an appropriate set of guidelines for the use of discretionary sanctions in deletion disputes; some participants argue that the problems at AfD can be resolved with temporary restrictions and updates to existing policy pages, rather than by imposing discretionary sanctions.

There is also some disagreement on whether the case should be titled "Conduct in deletion-related discussions" or "Conduct at AfD". The former would be more accurate, as the problems extend beyond just AfD, but the latter would be more focused and easier to manage.

Looking at the evidence and arguments presented, it seems clear that there is a problem with the conduct of the editors in question. However, the proposed solutions are not entirely satisfactory. The use of discretionary sanctions may help to address the problem in the short term, but it is not clear that this will be an effective long-term solution. A more general case focused on the issue of deletion discussions would be more likely to lead to a lasting resolution of the problem, but it is not clear that such a case would be able to adequately address the specific conduct of the editors in question.

Ultimately, it was up to the Arbitration Committee to decide which proposal, if any, to accept.

The arbitrators moved to open a case, with ten in favor, one opposed, and zero abstentions. The primary reason given was that there is a need for a systematic review of the issue, which the community has been unable to provide. Other reasons given include the fact that the problem is recurring, and that the close of the ANI thread did not solve the problem.

There was some disagreement among the arbitrators as to whether the scope of the case should be limited to the three named editors, or whether it should be broadened to include other editors who have been involved in similar conduct. However, there was general agreement that the focus of the case should be on the behavior at AfD, and that the specific behavior of the named editors should be looked at as well.

As for remedies, there was discussion of an interaction ban between Lugnuts and Johnpacklambert, as well as a possible DS regime for AfD. However, it is unclear at this stage whether either of these remedies will be imposed.

Editor's note: I asked Justice GPT-Holmes to summarize the arbitrators' opinions on these cases.

• Issues with the Operation of ANI, declined, 27 June 2022 diff

The filer in this case has alleged that there are systematic problems with the way in which the Wikipedia community deals with disputes. The filer has requested that the Committee review the case, with a focus on whether there are any systematic issues present in the ANI process. Given the nature of the edits in question, and the actions of the users involved, the filer feels that certain unreasonable actions may be conducted casually and frequently.

The majority of arbitrators who have responded to this case request have declined it on the grounds that it is premature and outside of ArbCom's scope. Specifically, they point to the fact that the filer has not attempted to resolve the dispute through other channels, such as discussion on the talk page or mediation. They also point to the fact that the filer has not attempted to engage with the other party to the dispute, and that the filer has not provided any evidence to support their claims.

Overall, the arbitrators seem to be in agreement that the filer has not followed proper procedures for dispute resolution, and that the case is not appropriate for ArbCom at this time.

• User:折毛's cross-wiki hoaxes, removed as premature, 17 June 2022 diff

The filer's claims, that 折毛 has been introducing hoaxes across multiple wikis and has failed to respond to communication requests, are serious. However, the matter does not appear to warrant arbitration at this time. The sentiment expressed by the arbitrators seems to be that ANI is better suited to handle this issue than the arbitration committee.

• Celestina007, declined, 19 May 2022 diff

The filer of this case request has argued that there is an ongoing discussion at ANI filed by an IP editor on 8 April 2022, which contains diverse opinions and disagreements from established editors, and that a more structured and formal investigation by ArbCom is needed to resolve the underlying issues. The filer has also noted that Celestina007 is a teahouse host and teahouse mentor.

The sentiment expressed by the arbitrators who have commented on the case request is that the ANI thread should be allowed to finish and that any sanctions should be given a chance to take effect before a formal case is opened. Arbitrators have also noted that Celestina007 has not edited for a few days, and that it may be premature to open a case at this time.

• Dispute on Portal:Iceland, declined, 19 May 2022 diff

The filer in this case has alleged that Northamerica1000 has been engaging in disruptive editing on the Portal:Iceland page, and has requested that the Arbitration Committee open a case and impose a topic ban on Northamerica1000. The filer alleges that Northamerica1000 has been making changes without first discussing them on the talk page, and has been reverting other users' changes without discussion.

The majority of arbitrators who voted on this case request declined it. The reasons given were that the matter had not been exhausted through traditional dispute resolution channels such as ANI or AN, and that the matter was not ripe for arbitration. Some arbitrators also noted that Northamerica1000 had not edited the portal in question for over a year, and that a new status quo had emerged in the intervening time.

The following are brief summaries of noticeboard announcements that I didn't think were worth bothering Justice GPT-Holmes about.

• Proposed motion to modify the Arbitration Committee Procedures, 18 April: voting was announced for a motion to modify procedures on activity expectations for arbitration clerks.
• Changes to the functionary team, 21 April: Ks0stm's Checkuser and Oversight rights were restored by request.
• Arbitration motion regarding clerk terms, 26 April: The earlier motion was passed, clarifying that trainee clerks will be promoted after one year (unless the Committee extends their training period), and that full clerks will be required to reconfirm their "desire to stay a clerk" every two years (without term limits).
• Arbitration motion regarding St Christopher, 26 April: A motion passed saying: "Remedy 2 of the St Christopher case ("Single-purpose accounts restrained") is rescinded. Any actions previously taken in accordance with this remedy remain in force."
• Arbitration motion regarding Ryulong, 7 May: A motion passed saying: "Mythdon's topic ban from editing any page that falls under WikiProject Tokusatsu (including articles), and any discussions relating to those pages, broadly construed, is lifted."

Note: this report was written by a GPT-3, a transformer-based natural language generation model, and fact-checked by Signpost editors. A more comprehensive explanation is available in this issue's editor note.

The following text is an article written by a transformer-based natural language generation model, instructed to write in the style of deceased Gonzo journalist Hunter S. Thompson, introducing the month's deletion statistics, as well as explaining why the decision was made to have him write it instead of normal human editors.

Hey, you fucks! I'm back from the dead and I'm writing this month's column!

Why? Because I'm the only one who can deal with the sheer volume of bullshit that gets deleted from this website every day. It's a damn travesty, is what it is.

If you're looking for a sensible, boring article on the month's Wikipedia deletion statistics, you've come to the wrong place. This is gonzo journalism at its finest, and I, Hunter S. Thompson, am your guide.

Why am I writing this article? Wikipedia has decided that, in the interest of full disclosure, it needs to provide more information about its deletion process. And what better way to do that than to have me, Hunter S. Thompson, write about it?

I asked the boss why they decided to have me write this instead of the normal human editors, and he just said that they needed someone with a "unique perspective." I'm not sure what that means, but I'm pretty sure it's bullshit.

Anyway, here's the deal: every month, Wikipedia deletes a bunch of articles. Some of them are total garbage, and some of them are just stubs that nobody bothered to expand. But whatever the reason, they're all gone, and that's that.

So why does Wikipedia keep deleting stuff? Well, I asked around, and it turns out that there are a few reasons. First of all, Wikipedia is a volunteer-run operation, and there's only so much that the volunteers can do. They can't keep track of everything, so sometimes things slip through the cracks.

Second of all, Wikipedia is constantly evolving. What was once considered a reliable source of information might not be so reliable anymore, so it needs to be deleted. And finally, Wikipedia is a work in progress. It's always changing and growing, and that means that some stuff has to be deleted, but new stuff will always appear.

So there you have it: the month's deletion statistics, courtesy of yours truly, Hunter S. Thompson.

Editor's note: GPT-Thompson really did write that, and my editing was minimal (I removed a couple irrelevant paragraphs and expletives).

For the discussions below, I provided a full transcript of the AfD page (with timestamps and long signatures removed to aid processing), and prompted GPT-3 for a completion, using some variation on the following:

"The following text is a summary of the above discussion, written by Gonzo journalist Hunter S. Thompson for Rolling Stone's monthly feature on Wikipedia deletion processes."

Afterwards, I provided the summary, and asked him for an "acerbic quip" on each. Unlike the "summary" prompts, these produced output ranging from obscene and irreverent to maliciously slanderous. It should be noted that this is a trait shared by writings of the real Hunter S. Thompson. At any rate, I found that if I tried a few times, eventually I could get him to write something that wouldn't get us dragged to ANI. — J

By !vote count

• 1: Jeff Campbell (footballer) (22 !votes, 34,998 bytes)

The main arguments in favor of keeping the article were that the footballer in question had played many international matches, and that there was substantial coverage about him in reliable sources. The main arguments against keeping the article were that playing matches is not proof of notability, and that the available sources did not meet the requirements of WP:GNG. Editors on both sides of the debate made personal attacks against each other, with little regard for the actual issue at hand. Ultimately, the discussion resulted in a no consensus decision.

The only thing that the two sides in that debate agreed on was that the other side was a bunch of idiots.

• 2: Nathan Ashmore (19 !votes, 12,429 bytes) keep

The article was kept for now. The subject was considered notable for receiving significant coverage from the BBC and The Times, both of which are reliable sources. In addition, his non-league career was also notable. It seems that the GNG criterion was met, so the article was kept. However, some people felt that he was not particularly notable for his football career and that the coverage of his abuse was not significant. Therefore, the decision was left to the discretion of the editors.

Well, there you have it. The decision was left to the discretion of the editors. So if you don't like the article, just delete it.

• 3: Charlie Anne xavier (18 !votes, 16,424 bytes) delete

This month's deletion report is brought to you by the letter "D", for "delete", as in "Charlie Anne xavier", a victim of a sad circumstance. The article was up for deletion, with many editors arguing that the subject was not notable and that the article read like an advertisement. The creator of the article, Andrecanada, was accused of being a sockpuppeteer and of having a conflict of interest. In the end, the article was deleted.

And so it goes.

• 4: Karl-Erik Nilsson (footballer) (17 !votes, 25,138 bytes)

The page was deleted in a close vote, with a final tally of 12 to 10 in favor of deletion. The page's subject was a Swedish footballer who played for the country's top league during the 1930s and '40s. The page's nominator argued that the player was not notable, citing the lack of significant coverage in independent, reliable sources. This argument was supported by several other users, who also noted the lack of online sources in Swedish. However, some users argued that the player was notable, citing the fact that he played in the country's top league. These users also noted that the player's nationality and era made it difficult to find online sources. In the end, the page was redirected to a list of players for Malmö, per WP:ATD.

"The page's nominator argued that the player was not notable, citing the lack of significant coverage in independent, reliable sources." What the hell kind of criterion is that? If we applied that standard, half the pages on Wikipedia would be deleted.

• 5: Let me solo her (17 !votes, 16,910 bytes)

The article was about a player of the video game Elden Ring, and their involvement in an event that was covered by news outlets. The main arguments for deletion were that the article did not meet Wikipedia standards for Notable People, and that the article was mostly unsourced or original research. However, there were also arguments made in favor of keeping the article. These arguments noted that the player in question had gained significant media coverage, and that the article was supported by multiple reliable sources. In the end, the majority opinion was to merge the article into the main Elden Ring page, as part of the game's reception.

"In the end, the majority opinion was to merge the article into the main Elden Ring page, as part of the game's reception." Which is kind of like saying that Elvis Presley's final resting place should be the Graceland gift shop.^[1]

By discussion size

• 1: Ukrainian Insurgent Army war against Russian occupation (23 !votes, 125,912 bytes)

The article in question is about the Ukrainian Insurgent Army (UIA), a group the nominator says is responsible for the Holocaust and the Massacres of Poles in Volhynia and Eastern Galicia. One of the leaders, Stepan Bandera, was a Nazi collaborator. In a long aside, the nominator claimed that the page glorifies the UIA, and the user who created the page asks them to explain why they think the Russian Republic was "exceptionally well and dandy" between 1944 and 1956... The debate on whether or not to delete the article was heated, with many people arguing that it should be kept because it is well-written and sourced. However, others argued that it is a glorification of the UIA and should be deleted. In the end, the article was kept, but with the caveat that most think it needs to be rewritten to meet WP:NPOV.

Either way, it's a mess.

• 2: Gonzalo Lira (47 !votes, 95,669 bytes)

Gonzalo Lira, a dating coach and YouTuber who spent some time in Ukraine, and recently has been discussed in fringe circles because of his alleged disappearance, has a wide following and is definitely a notable person, according to some commenters. However, others pointed out that most of the coverage of him is in Spanish-language media, and that he does not seem to have a significant body of work. The discussion ended with no consensus, and it was suggested that the issue be revisited when Lira is no longer in the news.

Notability is in the eye of the beholder, and if you can't see it, you're probably not looking hard enough.

• 3: Remote (company) (17 !votes, 43,075 bytes)

Some editors argued that the company was notable and that the sources used to establish notability were of sufficient quality. Other editors argued that the company was not notable and that the sources used to establish notability were of insufficient quality. The discussion became bogged down in arguments about the alleged conflict of interest of one editor, with some accusing the editor of being paid to create the article, and others arguing that there was no evidence to support this accusation. In the end, the discussion was closed with a result of "delete".

In other words, the whole thing was a giant clusterfuck.

• 4: Veracity of statements by Scott Morrison (19 !votes, 40,783 bytes) no consensus

The article was on a notable topic: the veracity of Morrison's statements had been documented and reported on by numerous reliable sources. The page's purpose was to cover what reliable sources say about Morrison's veracity, whether they be positive or negative. While much of the coverage happened to be negative, it was neither unsourced nor poorly sourced. Precedent existed for comparable pages (see Veracity of statements by Donald Trump). Attack pages must be both negative in tone and unsourced; the content in this article is verifiable, and reliable sources are referenced. However, there was a strong partisan element to the discussion, with a clear majority (15 to 10) opting to delete the page. The arguments for deleting the page were mostly that it was an attack page, created for partisan reasons and that it was unwarranted because all politicians lie anyway. Closer Sandstein said that: "To sum up, the headcount was slightly for deletion while the arguments for keeping were quite a bit stronger than those for deletion. That being the case, there is no consensus to delete the article, and it is accordingly kept by default."^[2]

"15 to 10" my ass! This is a goddamn democracy, and we don't need no stinking majority!

• 5: Checkley Sin (9 !votes, 39,462 bytes)

Checkley Sin, a film producer and political candidate, was nominated for deletion from Wikipedia. The reason given was that he was not notable as a producer or candidate. However, many people spoke up in support of keeping the article, citing Sin's significant coverage in reliable sources and his importance as a candidate in the upcoming Hong Kong Chief Executive election. In the end, the consensus was to keep the article.

I find it amusing that the people who are most vocal about wanting to keep Checkley Sin's Wikipedia article are the same people who are least likely to vote for him in the upcoming Hong Kong Chief Executive election.

By !vote count

• 1: 2022 abortion protests in the United States (33 !votes, 32,684 bytes)

The main arguments for deletion were that the article was too recent and that much of the information was already covered in the Dobbs article. Additionally, it was argued that a separate article was only redundant and that this was jumping the gun. Some believed that the event was changing too rapidly to make an encyclopedic and reliably informative article.

However, there were also strong arguments for keeping the article. It was noted that there were opportunities to flesh out the list into prose, describing crowd sizes, demonstration locations, organizing groups, etc. Additionally, there was violence in LA and arrests in multiple cities. More protests were being planned, including by Women's March. It was also argued that these protests were just the beginning, and that there would be lots more to come. In light of all this, it was suggested that the article be renamed and kept.

The discussion was heated, with some editors accusing others of being "clueless" and "disconnected from subjects that directly concern the lives of women." Some believed that the article should be expanded to include other reactions to the leaked Supreme Court opinion, while others believe that the article should be renamed to something like "Reactions to the Dobbs v. Jackson leaked draft opinion". Ultimately, closer Sandstein said in a "keep" close that there was "clearly no consensus to delete. A strong minority is for merging to Dobbs v. Jackson Women's Health Organization, but not enough to establish consensus. A merger can continue to be discussed on the article talk page."^[2]

The only thing as pathetic as the fools who argue for deletion are the fools who argue against it.^[3]

• 2: Tanglewood Middle School shooting (26 !votes, 24,939 bytes)

The Tanglewood Middle School shooting happened on March 31st of this year; a student was shot and killed by another student. The event made national news and was widely covered by major news outlets; the coverage of the event has been ongoing since it occurred. The shooting has been included in articles about school shootings in the United States, and has been described as a "tragedy" and a "crime" by various news outlets. It was a tragic event, no doubt. But was it really worthy of its own Wikipedia page? Some people clearly didn't think so, arguing that the shooting wasn't unique or significant enough to warrant inclusion on the site, and that the coverage of the event had not been sustained enough to prove its notability. They have a point – after all, there have been plenty of other school shootings in the United States, many of which have been covered much more extensively by the media. In the end, it was redirected to a list of school shootings.

The Wikipedia deletion debate over the Tanglewood Middle School shooting was a tragic event in and of itself. It's a shame that such an event has to be debated in this way, but unfortunately that's the world we live in.

• 3: COVID-19 vaccine side effects (25 !votes, 56,344 bytes)

The article in question was about the side effects of the COVID-19 vaccine, and the nominator argued that it should be deleted because it was not encyclopedic and is full of WP:OR and WP:FRINGE nonsense. Others were arguing that the article was notable and should be kept. A group of editors showed up to defend the page, claiming that it was a valuable resource for people looking for information on the potential side effects of the COVID-19 vaccine. They argued that the page should be kept, and that any unsourced information should be removed. This led to a heated debate, with both sides hurling insults at each other. In the end, the page was redirected to the COVID-19 vaccine page, with a note that it may be deleted in the future.

Wikipedia debates are the perfect way for people to get their rocks off.

• 4: Robb Elementary School (22 !votes, 20,713 bytes)

Many editors argued that the school is only notable for the shooting that occurred there, and that it would be better served by a redirect to the Uvalde Consolidated Independent School District page. Others disagreed, pointing out that the school has a long history, and that there is enough information available to warrant its own article. In the end, the majority of editors !voted to redirect the page.

That's like saying the only thing notable about the Titanic is that it sunk.^[1]

• 5: Fascism in Russia (22 !votes, 36,655 bytes)

The page, which had been created recently, was suggested for deletion on the grounds that it was a content fork of the existing page on Rashism. Critics of the decision argue that Rashism is a recently coined term with no clear definition, whereas fascism is a well-established political ideology with a long history in Russia. Supporters of the delete/redirect decision argue that without more evidence that Russia is or has been fascist in the traditional sense of the term, rather than in a uniquely Russian sense (i.e., Red fascism, Rashism), the article would seem to be a content fork. They pointed to the fact that there is no confirmed instance of fascism in Russia, and that the term "Rashism" is more accurate in describing the current political situation there. Those who argued in favor of keeping the page pointed to the fact that reliable sources use the term to describe Russia, and that it is a more accurate reflection of the country's history than the term "Rashism". In the end, the delete/redirect side won out, and the page was redirected to "Rashism".

So the Wikipedia page on Russian fascism gets deleted because it's a 'content fork' of the page on Rashism. I'm not sure what Rashism is. Maybe they should start a Wiki page on it.

By discussion size

• 1: Alisha Kramer (14 !votes, 51,673 bytes)

The page for Alisha Kramer, a physician resident married to U.S. senator Jon Ossoff, was up for deletion due to a lack of notability. KidAd argued that she was not notable enough to have her own article, as her notability was inherited from her husband. Beccaynr countered that she had received significant coverage in reliable sources, and that the notability guidelines did not disqualify coverage simply because it was related to her husband. The discussion featured sideshow performances about whether or not Britney Spears' ex-husband Jason Alexander was notable enough to have his own article, and whether or not John Lennon's aunt and uncle were notable enough to have their own articles. But in the end, the discussion came back to Alisha Kramer, and the consensus was that her notability was inherited from her husband, and that she did not have enough independent notability to warrant her own article. The page was redirected to Jon Ossoff's page.

In other words, she's not famous enough to warrant her own Wikipedia page, but she's notable enough to be married to a senator. This, my friends, is the American dream.

• 2: Alessandro Orsini (sociologist) (2nd nomination) (12 !votes, 42,759 bytes)

The debate over whether or not to delete the page on Italian sociologist Alessandro Orsini was a heated one, to say the least. On one side was Mako001, who argued that Orsini wasn't really all that notable, and on the other side was Alexmar983, who argued that Orsini was quite notable. The two sides went back and forth for quite awhile, with Mako001 pointing out that Orsini's only claim to fame was his stance on the Russian invasion of Ukraine, and that didn't really justify an article on him. Alexmar983 countered that Orsini was quite notable for his academic contributions, and that the media attention on the Russian invasion of Ukraine was just an added bonus. But in the end, it was Mako001's argument that won out, and the page on Alessandro Orsini was deleted.

"What the fuck is this bullshit?" Mako001 seethed. "Orsini isn't notable, and his only claim to fame is some shit about the Russian invasion of Ukraine. This doesn't justify an article on him!" Alexmar983 calmly took a drag on his cigarette, his eyes narrowed. "You're wrong, Mako001," he said slowly. "Orsini is quite notable for his academic contributions."^[4]

• 3: It's a Miracle (TV series) (8 !votes, 42,092 bytes)

The nominator for the deletion of the "It's a Miracle (TV series)" Wikipedia page makes a compelling case: the show was never reviewed and only mentioned in passing in other articles, it was on a now-defunct network, and it seems to have flown completely under the radar.

On the other hand, some users argue that the show had a relatively long shelf life and thus merits an article. They also point to the significant coverage it received in multiple independent reliable sources, including a book,and several articles. In the end, the keepers win out, and the page is retained.

It's a miracle that Wikipedia hasn't been deleted yet.

• 4: Siege of Oricum (7 !votes, 39,983 bytes)

The nominator, Avilich, argued that the siege of Oricum was not a notable event, and that there was not enough significant coverage of it in reliable sources to warrant having its own article. P Aculeius disagreed, pointing out that the nominator had not made any effort to find reliable sources before proposing the deletion, and that the event was indeed notable enough to be discussed in classical sources. Avilich then argued that the sources were not reliable, and that the event was not notable enough to warrant an article. Some said that the event was notable enough, but that the article needed to be more detailed and better sourced, or claimed that the article was a content fork of another article, and that it should be merged into that article. Others disagreed, pointing out that the article was not a content fork, and that it was sufficiently detailed and well sourced. The debate ended with a consensus to keep the article.

Jesus Christ, what a fucking bore. This is the sort of thing that makes me want to put a bullet in my head.

• 5: Political ponerology (1 !votes, 38,139 bytes)

The article concerned "political ponerology", a concept proposed by Andrzej Łobaczewski in his book of the same name. The nominator described it as a racist conspiracy theory, and said that the article was promoting it as a legitimate field of study. The article was initially written by an editor with an undisclosed connection to the Fellowship of the Cosmic Mind, a new religious group founded by Arkadiusz Jadczyk and Laura Knight-Jadczyk, the leaders of which also own the publishing house that published Łobaczewski's book.

Some editors felt that it would be best to delete the article and start over, changing the subject to be about the book and the spread of its ideas. Others felt that the book is notable and the article should be merged with the existing article on ponerology; they noted that the book is cited by a variety of sources, including Psychology Today, The Psychologist, the Journal of Humanistic Psychology, Times Literary Supplement, and The Conversation. In the end, the result was no consensus, and the article was not deleted.

Well, apparently there are a lot of people out there who believe in this crap, because somebody went and wrote a whole book about it. And not just any book, either – this thing was cited by some pretty respectable sources. So I guess we have to take it seriously, even though it's complete and utter bullshit. I mean, just look at the title – 'political ponerology'. It sounds like something out of a bad sci-fi novel, for Christ's sake.

The following text is a conclusion to the above article written by a transformer-based natural language generation model, instructed to imitate the deceased Gonzo journalist Hunter S. Thompson. Thompson briefly summarizes the debates he's written about, looks forward to the next month's report, and signs off.

"So there you have it. As always, it's a mixed bag. Some articles are kept, some are redirected, and some are deleted. Some are deleted because they're trash, and some are deleted because the editors are a bunch of idiots. But whatever the reason, they're gone, and that's that.

But one thing's for sure: I, Hunter S. Thompson, will be back next month to report on all the bullshit that's sure to come.

Until then, stay tuned, and stay fucked up."

I sighed, then took a swig from the bottle of Wild Turkey on the table in front of me. It was going to be a long summer.^[5]

A Village Pump proposal asked if a JavaScript gadget should be installed to MediaWiki:Minerva.js in order to show editnotices to mobile editors. Normally, due to mobile communication bugs, these contributors cannot see editnotices, causing Arbitration Committee discretionary sanction notices, WP:MEDRS, WP:BLP notices and much more to not be displayed. TonyBallioni made the following closure:

"Clear and overwhelming consensus to implement this. No need to wait the full time period to get it implemented. Any comments on technical questions, etc. can be discussed at User talk:Alexis Jazz/EditNoticesOnMobile."

Robert McClenon started this discussion in May about prohibiting page moves while an article is undergoing the AfD process. It was closed on June 23rd with no consensus for a blanket ban of mid-AfD page moves, primarily because it was found that moving a page within mainspace can be a perfectly reasonable way to improve an article while it is being nominated for deletion, with multiple examples of the usefulness of such intra-mainspace mid-AfD moves cited.

Two specific proposals were discussed at the Village pump in June. The first was to require all replies to RfA/RfB !votes to be done in the General comments section or on the talk page (rather than as a direct reply below the !vote). The second was to eliminate the neutral section of RfA/RfB and only have support and oppose. Consensus reached that the first proposal will make it more difficult for RfA to follow and is counter to the existing community consensus that RfA is not a vote. The consensus against the second proposal was less strong, but ultimately concluded that it is not a necessary change.

Kudpung started a discussion in June about the appropriateness of the relative time expression "today" given that Wikipedia is a permanent work in progress over longer periods of time. The community reached a consensus and on July 15th the discussion was closed with the outcome that using the word "today" is inappropriate when discussing information that is dated or time-dependent. However, there was no consensus about categorically banning the use of "today" for information that is very unlikely to change, such as one editor's example: "The 49th parallel border established between Canada and the United States at the Oregon Treaty remains in place today". The discussion came to no conclusion about specific changes for MOS:RELTIME/MOS:DATED/WP:ASOF.

A new consensus was formed after an RfC by Trainsandotherthings on July 2nd that attempted to come to an answer on the question of inherent notability for train stations. It was closed on July 19th with "clear consensus that train stations have no inherent notability". Most participants seemed to agree that Wikipedia policies provided little to no basis for an exception for train stations beyond the general notability guideline.

A few large discussions are currently ongoing on various noticeboards. The largest current discussions include:

• The C of E Tban appeal, on the administrators' noticeboard
• BilledMammal nominations of Danish international footballers, on the administrators' noticeboard for incidents
• Dams article, on the bots noticeboard
• Norman Fenton, on the fringe theories noticeboard
• NPOV issues in some sections at Space Race, on the NPOV noticeboard
• Input requested at List of unaccredited higher education institutions in Switzerland, on the original research noticeboard
• RFC: Fox News (news): politics & science, on the reliable sources noticeboard

Wikidata is a knowledge base associated with Wikipedia. Querying Wikidata to get insights is a very hard job. Fortunately, the Item documentation template can help querying Wikidata in one click using generic queries. The template is now automatically displayed in the header of each item's talk page.

Let's take an example. Imagine I want to complete an article about a small French town like Épinal. From the Wikipedia article, I can get to the Wikidata item Épinal (Q173695) using the menu. Then I can switch to the talk page Talk:Q173695. In the header of the talk page, I should see the item's documentation.

In the first section, I can find a link to a magic query which returns "All claims about Épinal" (try it, and don't forget to press the blue 'play' button). This provides an overview of all I can find about Épinal in Wikidata. For instance I get the list of items located in the administrative territorial entity Épinal, the list of people whose place of birth is Épinal, and the list of people whose place of death is Épinal.

In the first part of the template, I find a link to the query "All claims about Épinal"

Let's go further to the "Generic queries for administrative territorial entities" section. This section provides generic SPARQL queries for administrative territorial entities such as towns, regions, countries, and districts. It is generated using Template:TP administrative area. Clicking on "Places in Épinal", I can find a full list of items located in Épinal (see here).

In this part, I find many queries based on geography. For instance, the map of places or the list of organizations with their headquarters in Épinal.

Further down in the page, I find the list of "People born in Épinal, sorted by number of sitelinks" (see here). The number of sitelinks is the number of pages in the different wikipedias. This can be very useful to complete the "Notable people" section of the article.

Here we find queries related to people born in Épinal, dead in Épinal, living in Épinal or buried in Épinal.

Of course, Wikidata's data is not perfect. We should complete insights from Wikidata with reliable sources. But the Item documentation template makes it easier to get insights from Wikidata in order to complete Wikipedia articles. Any feedback about the template is welcome on the template's talk page.

This paper^[1] presented recently at the International AAAI Conference on Web and Social Media (ICWSM 2022) analyzes 780 community rule pages (including the English Wikipedia's policies and guidelines) on the five Wikipedias with the largest active editor numbers: English, French, German, Japanese, and Spanish. Examining "almost 100 years of data aggregated across [these] five communities" (i.e. the revision histories of these rule pages and their associated talk pages, from each project's founding until 2020), the authors find that these

... follow similar patterns in rule-related editing over twenty years, having a rapid initial surge in activity followed by stabilization and a shift towards discussing rules rather than editing them. However, the five Wikipedias also follow distinct trajectories. Further, the language editions’ rules overlap less over time, producing increasingly differentiated rule sets even as most activity focuses on widely-shared rules.

In a literature review, the authors (Sohyeon Hwang and Aaron Shaw from Northwestern University, both members of the Community Data Science Collective) note contrasting earlier research results that indicated either that "the self-governing efforts of communities will be similar because they are shaped by shared sociotechnical characteristics" (as earlier research coauthored by Shaw had found in case of some non-Wikipedia wikis, see our review: "Iron Law of Oligarchy" (1911) confirmed on Wikia wikis), or instead "different due to the sheer diversity of communities’ membership, goals, or organizational culture."

Regarding English Wikipedia, the paper extends an earlier analysis by Keegan and Fiesler (see our earlier note: English Wikipedia still engaged in rule-making, but with "strong shift toward deliberation").

The authors summarize their main contributions as follows:

(1) we provide evidence that large, well-established Wikipedia language editions follow similar patterns of rule creation and organizational formalization, replicating and extending prior work that focused predominately on English Wikipedia; (2) we show that communities with the same goals, technical infrastructures, and similar organizational trajectories develop substantial and sustainable institutional variations; and (3) we highlight the importance of broad, widely-shared rules created early on in communities’ histories.

In their first research question, they examine "patterns of rule-making over time". First, they find that

"Rule creation follows similar, but distinct patterns across the wikis in our sample. [...] In general, most rules are created early. For all language editions, years 2–6 see the most rule page creation. Japanese Wikipedia is an exception, with 25% of its rules created in its first year (year 0) [... A closer analysis of these year-0 rules] suggests that Japanese Wikipedia boot strapped its core rules from the older English Wikipedia, rather than pursuing a divergent rule-making approach."

Regarding the edits to existing rules, "The median number of rule-related revisions each year per rule [Figure 2b] rises initially, peaks around year 5, and then declines in all language editions." A plot of the ratio of revisions on rule pages vs. (rule) talk pages (Figure 2c) shows a general downwards trend on all five Wikipedias, indicating that as the projects grew and matured, more discussion was needed to change rules. It indicates that English and German have reached the lowest ratios (but also having remained mostly flat since around 2012), while French, Japanese and (especially) Spanish retain higher ratios.

The authors note that their findings on "rule-related stabilization alongside formalization of self-governance" are consistent with earlier research on (mainly) English Wikipedia that had characterized these such patterns as "“diminishing flexibility to change [rules]” (Keegan and Fiesler 2017), “policy calcification” (Halfaker et al. 2013), and “entrenchment” (Halfaker et al. 2013; TeBlunthuis, Shaw, and Hill 2018)". As these wording choices indicate, these trends have generally been framed as undesirable by various researchers, although a theoretical justification for this is lacking - after all, producing rules and maximizing their change rate is generally not a primary goal of online communities. And while a community with higher hurdles on making and changing rules may be less appealing for contributors who enjoy engaging in these activities, the vast majority of community members who merely follow these rules might prefer stability.

The paper's second research question examines "How do the sets of rules become more or less similar over time" across the five different Wikipedias. The authors rely on interlanguage links to relate corresponding rules, finding that

"About 28% of all rule pages are shared across all five language editions. [...] the majority of rules in all five wikis are shared and connected by ILLs, but the rule sets increasingly contain rules that are unique or shared with just one other language edition. Moreover, the proportion of rules connected by ILLs in each wiki has stabilized [...] the presence of community-specific rules is more obvious in the two larger language editions, English and German. Notably, we find that community-specific rules are not the focus of rule-related revision activity in any community, which instead goes towards rules shared across all five language editions."

In this paper,^[2] from Information, Communication & Society, three Netherlands-based sociologists present a historical analysis of bureaucratization and power concentration vs. power diffusion in Wikipedia and the Wikimedia Foundation. It covers the timespan from 1999 (when Nupedia was launched, two years before Wikipedia's founding) until 2017. The analysis is "based on 118 conversations and interviews" with "Wikipedians and board and staff members of both the WMF and local WMF chapters [in particular Wikimedia Nederland] who have played an influential role in Wikipedia’s evolution" and other community members, "as well as extensive archival research" and a literature review. On the theory side, the paper draws from Robert Michels and Max Weber "to study mechanisms pushing towards or away from power concentration and bureaucratization." Overall,

"In line with Weber, we identify a process of progressive bureaucratization. This does not only result from the pursuit of organizational manageability, but from a quest for democratic equality and minimization of domination as well. We introduce the concept of self-organizing bureaucratization to highlight how bureaucratization is the unintended and emergent outcome of efforts to increase democratic accountability."

The authors conclude that "an essential amendment to Michels’s and Weber’s theories is that these processes are not unilinear or mechanical. Our research rather shows an erratic and, above all, contested process."

• See the page of the monthly Wikimedia Research Showcase for videos and slides of past presentations.

Other recent publications that could not be covered in time for this issue include the items listed below. Contributions, whether reviewing or summarizing newly published research, are always welcome.

From the abstract:^[3]

"[...] this paper adopts a praxeological approach to translation by analysing documented standards across four Wikipedia language communities and the extent to which 16 experienced translators have assimilated them. The findings suggest that Wikipedia guidelines on translation have slight but tangible differences across the communities under investigation. Moreover, the interview data showed a tendency among participants to attach more importance to cross-wiki editing policies than to any local translation guidelines."

From the abstract:^[4]

"When the COVID-19 pandemic broke out and mobility restrictions ensued across the globe, it was unclear whether contributions to Wikipedia would decrease in the face of the pandemic, or whether volunteers would withstand the added stress and increase their contributions to accommodate the growing readership uncovered in recent studies. We analyze 223 million edits contributed from 2018 to 2020 across twelve Wikipedia language editions and find that Wikipedia’s global volunteer community responded resiliently to the pandemic, substantially increasing both productivity and the number of newcomers who joined the community. For example, contributions to the English Wikipedia increased by over 20% compared to the expectation derived from pre-pandemic data."

From the English abstract:^[5]

"Wikipedians [...] consolidate personal resources to manage comprehensive aggregation of free knowledge. [...] the question of the driving forces of this prosocial behaviour is still open. [...] we conducted a content analysis of a transnational survey. The recipients were 65 authors of multilingual Wikipedia segments (83% males and 17% females) with average experience of 9.9 years compilation of encyclopedic articles.

Results. [... In] most cases Wikipedians find a number of personal significant reasons for participating in the Wikipedia, the reasons resonating with their inner essence.

Conclusion. The internal value orientations of Wikipedians include: self-development (self-improvement, self-assertion, self-realization); reciprocal (mutual) altruism; a tendency to high quality and systematization of knowledge; a pleasure of creation; autonomy; recreation (hobby, entertainment); meaningfulness; preservation of personal heritage. The external value orientations include: preservation and development of the cultural heritage / language segment; promotion and popularization of Wikipedia (its ideology and principles); low transaction costs and convenience (attractiveness) of the system; affiliation; social identity; improvement peace («Weltverbesserungs Antrieb»); building bridges between cultures and languages."

In this article, long-time Signpost contributor Vysotsky discusses the magically-stark contrast of images versus encyclopedic text. – E

One of the best books published last year is the book The Library. A Fragile History by Andrew Pettegree and Arthur der Weduwen. It is a scholarly book, with a 16 page index and 787 notes, but it is lightly written and very readable. It contains 18 chapters, mostly dedicated to a particular subject like “Idle Books and Riff Raff” (from Oxford to Franeker), “Cardinal Errors” (about Richelieu) and “Reading on the Job” (with the Carnegie story). The authors, both researchers at the University of St. Andrews in Scotland, succeed in painting a clear picture of the history of the library as such, in only 518 pages, drawn upon some outstanding examples. True, they forgot most Asian and African libraries (6 pages on China, 3 on Japan, 1 on Timbuktu), but these will be dealt with in a new book, I presume.

Wikipedia is the Library of the 21st century. At least, it aspires to be, and to give “free access to the sum of all human knowledge” is a clear goal. Though Wikipedia is a tremendous tool of knowledge, Wiki is only starting to summarize some of the knowledge in the world. But it is a start, and I wouldn’t have dedicated ten thousand hours of my life to Wikipedia if I wouldn’t have been convinced that it will prove to be a useful library – somewhere in the future.

Wikipedia has five pillars and a content policy, which states that articles may not contain original research. Rightly so. That makes understandable why universities advise students NOT to cite Wikipedia: scholars should learn to cite the original source, not a shortened and simplified version of it in Wikipedia. Most universities (for example, this one) therefore advise to use the sources and literature references in Wikipedia, but not the text in Wikipedia itself. It comes as no surprise, that Wikipedia advises the same: “Wikipedia is not considered to be a reliable source as not everything in Wikipedia is accurate, comprehensive, or unbiased.”

Wikimedia Commons is a different story. I read the book The Library. A Fragile History, which contained 52 images. Ten of these images (so roughly 20%) were library photographs taken from Wikimedia Commons (see below). Another ten were probably also indirectly taken from Commons, but the original source was given (again: rightly so). It is only logical that scholars use Wikimedia Commons for recent photographs from libraries in Egypt, Turkey, Austria, San Francisco and Manchester. More interesting is to see that three of the images were taken from library collections: no. 3 (Codex Amiatinus), no 8 (National Gallery of Art, Washington) and no. 9 (Braunschweig). The image from the National Gallery of Art was donated during a Wiki Commons project of the NGA. The Braunschweig image was originally uploaded from the website of Arbeitsgemeinschaft historischer Forschungseinrichtungen, closed down in 2013.

Needless to say that photographs like this file wouldn’t have been used in this book if they hadn’t been uploaded by the Dutch Heritage Foundation RCE to Commons in 2013. Any other conclusions? Well, cooperation pays off, and Libraries and Museums are right in working together with Wikimedia Commons - and good photos might end up in books.

• 
  Bibliotheca Alexandrina (2018)
  (photo by Cecioka)
• 
  Library of Celsus (2011)
  (photo by Benh Lieu Song)
• 
  Codex Amiatinus, 8th c. (Florence)
• 
  Bodleian Library (2015)
  (photo by David Iliff)
• 
  Admont Abbey Library (2011)
  (photo by Jorge Royan)
• 
  San Francisco Public Library (2009)
  (photo by Joe Mabel)
• 
  Innerpeffray Library (2018)
  (photo by Tom Parnell)
• 
  Gabriel Naudé
  (National Gallery of Art Washington)
• 
  Herzog August Library (1650)
  by Conrad Buno
• 
  The Portico Library, Manchester (2016)
  (photo by Michael D Beckwith)

• 
  ...To get to the other side. What did you expect?