Wikipedia:Wikipedia Signpost/Single/2023-10-23

We have had eight successful candidacies for adminship so far in 2023, which is just one more than the worst-ever year for RfA, which was 2021. The number of active administrators started the year 2023 at 500, then took a big dive in mid February for no reason that The Signpost has been able to determine, touched 452 a couple of times in April, and until October was steady around 460.

On October 18, we hit a new record low, going back over a decade, of 448 active admins. To find the last time English Wikipedia had fewer than 449 active admins, we have to go back to 2005.^[a]

The reasons for this are cloudy and have been covered before by The Signpost, for instance at "Administrator cadre continues to contract – more" in January 2022.

In a recent Administrators' noticeboard discussion titled "Twelve fewer administrators", BeanieFan11 noted not a single month this year have we had a net gain of administrators, and so far all but one have had a net decrease, some large - per the admin's newsletter: January: +3, -11, net -8; February: +1, -5, net -4; March: +1, -2, net -1; April: +1, -1, net 0 (only month without negative net); May: +1, -4, net -3; June: +1, -3, net -2; July: +1, -8, net -7; August: +1, -4, net -3; September: +2, -4, net -2; October: +1, -12, net -11; Overall: +13, -53, net -40

At least one disappeared admin was fallout from the WP:ROADS controversy ending in a content fork and some departures, including Rschen7754 who resigned as an administrator and editor. – B

The Wikimedia Foundation has published comprehensive notes from the recent community call about the Knowledge Equity Fund on Meta. The notes include a Q&A. The WMF also highlights that the Fund has helped it to make new connections:

We contacted user groups and connected the grantees with them geographically or thematically, explaining the objects of the fund. We are also trying to create new synergies between Wikimedia user groups and external groups to increase our impact.

A few examples of connections we made are:

• Project Multatuli, which we connected with Wikimedia Indonesia
• Create Caribbean were connected with Noircir, Wiki Cari UG, Whose Knowledge, Projet:Université de Guyane and WikiMujeres
• Black Cultural Archives were connected with Noircir, Whose Knowledge and Wikimedia UK
• Criola were connected with Whose Knowledge, WikiMujeres and Mujeres (mulheres) LatinoAmericanas in Wikimedia and
• Data for Black Lives which we connected with AfroCrowd and Black Lunch Table

Through these connections, we have seen positive synergies within the movement at large

An ongoing English Wikipedia Village Pump Request for Comment on the controversial fund stands at 35:23 in favour of adopting the following non-binding resolution:

The English Wikipedia community is concerned that the Wikimedia Foundation has found itself engaged in mission creep, and that this has resulted in funds that donors provided in the belief that they would support Wikimedia Projects being allocated to unrelated external organizations, despite urgent need for those funds to address internal deficiencies.

We request that the Wikimedia Foundation reappropriates all money remaining in the Knowledge Equity Fund, and we request that prior to making non-trivial grants that a reasonable individual could consider unrelated to supporting Wikimedia Projects that the Foundation seeks approval from the community.

– AK

A request for comment (RfC) to create an English Wikipedia policy or guideline regulating editors' use of large language models (e.g. ChatGPT) was rejected recently. Specifically, the RfC concerned the proposal to elevate the draft page Wikipedia:Large language models (expanded from a much smaller version created in December 2022) to policy or guideline status. As summarized by the closing editor:

There is an overwhelming consensus to not promote. 1 editor would promote to policy, 7 editors prefer guideline, and 30 editors were against promotion. 2 editors were fine with either policy or guideline. [...] The most common and strongest rationale against promotion (articulated by 12 editors, plus 3 others outside of their !votes) was that existing P&Gs [policies and guidelines], particularly the policies against vandalism and policies like WP:V and WP:RS, already cover the issues raised in the proposals. 5 editors would ban LLMs outright. 10-ish editors believed that it was either too soon to promote or that there needed to be some form of improvement. On the one hand, several editors believed that the current proposal was too lax; on the other, some editors felt that it was too harsh, with one editor suggesting that Wikipedia should begin to integrate AI or face replacement by encyclopedias that will. (2 editors made a bet that this wouldn't happen.)
Editors who supported promoting to guideline noted that Wikipedia needs to address the use of LLMs and that the perfect should not be the enemy of the good. However, there was no general agreement on what the "perfect" looked like, and other editors pointed out that promoting would make it much harder to revise or deprecate if consensus still failed to develop.

Similarly, on Wikimedia Commons a page collecting guidance about AI-generated media (particularly the use of generative AI models such as DALL-E, Stable Diffusion or Midjourney), likewise created in December 2022, is still marked as "a work in progress page", although it appears to have progressed a bit further already towards reflecting community consensus.

In any case, discussions about generative AI are continuing in the Wikimedia movement, also in off-wiki fora such as the "ML, AI and GenAI for Wikimedia Projects" Facebook group and the "Wikimedia AI" group on Telegram (non-public but with public invite link). At Wikimania 2023, it was the subject of various sessions including two panels titled "AI advancements and the Wikimedia projects" (video) and "ChatGPT vs. WikiGPT: Challenges and Opportunities in harnessing generative AI for Wikimedia Projects" (video). The September edition of the Wiki Education Foundation's "Speaker Series" likewise had the topic "Wikipedia in a Generative AI World", featuring three speakers including Aaron Halfaker (User:EpochFail, a former research scientist at the Wikimedia Foundation and developer of the AI-based ORES system that is still widely used for vandalism detection and other purposes). – H

In its EU Policy Monitoring Report for September, Wikimedia Europe highlights several legislative efforts that are ongoing on the continent. Some of them raise concerns regarding their possible impact on Wikipedia and other Wikimedia projects:

• The EMFA (European Media Freedom Act) is "intended to help a pluralistic media landscape", but also contains problematic provisions, e.g. a requirement for online platforms to warn "media providers, who can be media outlets but also individuals, such as journalists [...] ahead of moderating their content and to give them a fast-track channel to contest decisions. Some lawmakers even suggest that online platforms be prohibited from deleting content by media providers before the provider has had a chance to reply. All this is highly problematic, seeing that disinformation is sometimes produced by media providers." Efforts to exempt Wikimedia projects or at least non-profit "online encyclopaedias" succeeded initially but then were in jeopardy again. However, negotiations are expected to continue into 2024.
• The controversial Regulation to Prevent and Combat Child Sexual Abuse (CSAR) proposed by EU Commissioner Ylva Johansson is reported to have "stalled somewhat" recently. It would cover Wikimedia projects too, "and the Wikimedia Foundation has provided [already in 2022] constructive feedback, outlining some risks and challenges posed by the scanning technologies used. Wikimedia is also criticising the idea to scan direct, interpersonal communication in a general manner and without judicial oversight."
• In France, the proposed Loi SREN "would introduce some provisions on data retention and user identification, in order to not allow already banned users to re-register. That would require the collection of heaps of data and the compulsory identification of all users. Wikimedia projects are squarely in the scope of this proposal." Initial efforts to "take our projects out of the fireline" have failed.

– H

• Annual reports: Wikimedistas de Bolivia, Tyap Wikimedians User Group
• Wikimania videos are up: Two months after Wikimania (which took place in Singapore as an in-person event, for the first time since 2019), video recordings of the conference's sessions have become available on Commons (joining unedited recordings that were already published during the event on YouTube).
• Global bans: Gustin Kelly, since 18 October 2023
• Articles for Improvement: This week's Article for Improvement is Power (social and political). Please be bold in helping improve this article!

In Slate, Annie Rauwerda (of Depths of Wikipedia fame) explains why "On Wikipedia, Anyone Can Be a Model". The article focuses on LittleT889, who created the article thirst trap ("a type of social media post intended to entice viewers sexually") and illustrated with a shirtless selfie of himself that has since been "viewed almost a million times" (although it was recently replaced in the article). What's more, "He adds photos of himself to all sorts of encyclopedically relevant topics, like water bottle flipping, Nae Nae, and my favorite, the Floss (dance) article, where he wears sunglasses indoors and furiously shakes his hips in front of three guitars and a bongo drum", as well as Running man (dance), Dougie, and Naruto run.

Rauwerda also managed to get in touch with other selfie contributors, such as a "20-year-old Russian university student [who said that] once he took a picture of his eye so astonishingly beautiful that an Instagram post wasn’t enough — he needed to put it on Wikipedia," and "a retired biology teacher in Germany [who] realized that Wikipedia had no good photos of female fingers [and] uploaded a snap of her own hand to Wikimedia Commons", which now illustrates the article finger. In general, the article observes that Wikipedia's "photos have an unvarnished feel and an unmistakably human charm" and that it's "immediately obvious that Wikipedia’s models are real people, not actors." It also recalls earlier media coverage of similar examples, such as a couple who has graced the high five article since 2008, and last year - now married with kids - recreated the shots for an online magazine (see earlier coverage: "The king and queen of the high 5").

• And of course we're showing it:
• 
  "Up high."
• 
  "Down low."
• 
  Victim misses.
• 
  "Too slow!" (with finger-guns)

Lastly, Rauwerda calls on Slate's readers to consider contributing themselves: "And even though Wikimedia Commons hosts more than 100 million pieces of media, it has some stunning gaps. There’s a big list of requested images, and some of the items are shockingly quotidian, like 'half-up hairstyle,' 'business women shaking hands,' and 'tripping' (go ahead, fall on your face for the sake of free knowledge)."

"TechNewsWorld" reports that "Craigslist, Wikipedia, and Zillow are the fastest-loading U.S. websites on the internet, according to a study released Monday by web design company DigitalSilk." Wikipedia came second "with an average load time of 1.40 seconds (1.6 mobile, 1.2 desktop)", well ahead of sloths such as Instagram ("The site on mobile takes a whopping 6.7 seconds to load and 4.2 seconds on desktop") or Google ("While it had a respectable mobile load time of 1.1 seconds, its desktop time of 4.6 seconds bloated the search giant’s overall performance").

This success is certainly in part due to the longtime work of the Wikimedia Foundation's recently disbanded Performance team, but also, according to one of the study's authors, due to Wikipedia's simpler design: "It’s interesting to see how websites like Wikipedia and Craigslist, which have barely changed their design and have remained largely text-based, topped our list, and the popularity of these sites shows that sometimes simplicity can work."

(The TechNewsWorld article doesn't link to the actual study and doesn't provide much detail about its methodology. But in a similar study featured by ZDNet earlier this year, DigitalSilk had used an online tool called Pingdom Website Speed Test.)

Nature News notes the publication of "Improving Wikipedia verifiability with AI" in Nature Machine Intelligence. "Wikipedia lives and dies by its references" the news article states, "but sometimes, those references are flawed." The neural network-based system described in the academic study, called SIDE, looks at whether Wikipedia references support the Wiki text, and proposes replacements for the weaker refs.

The paper is open access, licensed under the Creative Commons Attribution 4.0 International License, and can be downloaded at the above links.

The Washington Post reviews the book Dictionary People by Sarah Ogilvie, a former lexicographer for the Oxford English Dictionary. The review, titled The most influential crowdsourcing project happened long before Wikipedia, focuses, like the book, on the roughly 3,000 OED contributors who sent in quotations showing words in use in printed texts. Only a dozen or so are actually named in the review, but 97 of these unpaid volunteers are recorded at Wikipedia's List of contributors to the Oxford English Dictionary.

Eadweard Muybridge hasn't yet made our list but two other murderers, Sir John Richardson (naturalist) and William Chester Minor do. Margaret Murray, who later became an Egyptologist and wrote The Witch-Cult in Western Europe, contributed 3,800 quotations from the Douay–Rheims Bible while growing up in India. There are many women among the 3,000 including Karl Marx's daughter Eleanor Aveling and a lesbian couple who wrote under the name Michael Field.

The implicit comparison of OED contributors to Wikipedians in the book review's title might seem exaggerated at first glance. After all, some of the OED contributors are quite unusual. But Ogilvie does play at comparing OED contributors with Wikipedians in the book's Introduction, which I just had to read on Amazon after reading the review. I'll have to read the rest of the book before drawing any firm conclusions. –S

• What do "Ant", "Orca", and "Erie" have in common?: Christina Iverson's husband (archive, October 10) tells her that she is "a professional Wikipedia reader." Actually, she is an Associate Puzzle Editor at The New York Times and has often visited our articles on Ant, Orca and Erie due to the words' usefulness in puzzles. More important is her belief in lifelong learning. Do you know who gave the eulogy at boxer Muhammad Ali's funeral? Or when Cheez-Its were first sold?
• Wikipedia and Bible paper were both useful to Jeff Turner, "the Picasso of counterfeiters", according to Insider via YouTube.
• Maryana Iskander at the UN: WMF's CEO spoke at a Ministerial meeting of the Freedom Online Coalition at the UN as recorded on a YouTube video and transcript.

Rank	Article	Class	Views	Image	Notes/about
1	India at the 2022 Asian Games		2,382,812		Hangzhou received the continental games that ended on October 8, and our South Asian friends that always shape this report had their best performance, surpassing 100 medals and standing fourth in the total count. Among the podiums are over 20 in both shooting and athletics, three for those who medaled in Tokyo 2020 (javelin thrower Neeraj Chopra, seen to the left, boxer Lovlina Borgohain, and the field hockey team), and both gold medals in the national sport (see below).
2	2022 Asian Games		2,120,017
3	2023 Cricket World Cup		2,102,848		India is a natural to host this, as they even have a cricket ground for 132,000 people. The tournament began last week, with a rematch between the finalists of the last edition which ended in the champions' defeat. The group stage of this edition alone will last over a month, so this article will take a while to leave.
4	David Beckham		1,654,841		Netflix released Beckham, a documentary miniseries about one of the faces of England's late 90s football resurgence, who won a lot with Manchester United, Real Madrid and LA Galaxy, with the stint in the last one helping Beckham get an expansion team from Major League Soccer, Inter Miami FC, who currently employs an even bigger football star than Beckham ever was.
5	Jawan (film)		1,248,683		After becoming the biggest Indian movie of 2023 and the second-biggest Hindi film of all time, this Shah Rukh Khan-starrer continues to bring in the crore in the last lap of its run. Meanwhile, a certain Canadian believes that Hollywood could learn a thing or two from it.
6	Travis Kelce		1,034,857		The Kansas City Chiefs are going off well in their post-Super Bowl season. But most views regarding their tight end regard how he and Taylor Swift might be in a case of "You know I love the players, and you love the game".
7	Cricket World Cup		1,034,149		#3 led to readers seeking out the article on the quadrennial event whose trophy still remains the biggest honor in cricket.
8	Kevin McCarthy		928,377		He was the US Speaker of the House for 10 months, starting in January this year and ultimately ending this month after he was axed. This makes him the third shortest serving US speaker in history, and the first in history to be axed.
9	Deaths in 2023		921,298		For you, it's goodbye For me, it's to cry For whom the bell tolls...
10	Dick Butkus		860,882		The American football linebacker who played for the Chicago Bears between 1965 and 1973 has died aged 80. He was considered the most feared tackler of all time. After his playing career, Butkus became active in sports commentary and philanthropy through his charity the Butkus Foundation.

Rank	Article	Class	Views	Image	Notes/about
1	Gaza Strip		5,344,460		The Israeli–Palestinian conflict has been ongoing since 1947, and centers on the fact that both Jews and Arabs claim the Holy Land for themselves. It is one of the longest lasting conflicts that continue to this day. Saturday, October 7, was a day of Sabbath, a day of holiness and rest observed by Jewish men and women. This was the date on which Hamas began Operation Al-Aqsa Flood, breaching the Gaza–Israel barrier, murdering over 1,200 Israelis and taking hundreds hostage. Israel began Operation Swords of Iron. Prime Minister Benjamin Netanyahu (#15) began retaliatory strikes and pledged to "turn all the places where Hamas is organized and hiding into cities of ruins". On October 8, Israel declared a state of war. The Israel–Hamas war is currently in its second week. Thousands on both sides have been killed, and more than a million have been displaced.
2	Hamas		5,213,111
3	2023 Israel–Hamas war		4,371,096
4	Israel		4,062,258
5	2023 Cricket World Cup		3,829,668		Only a nation of a billion can prevent a top 5 centered on the war, as India is hosting the world championship of its favorite sport. Last week saw the hosts win their first three matches, and five-time champions Australia lose their first two matches.
6	State of Palestine		3,813,694		#3 brings in the partially-recognised country in an ongoing sovereignty dispute with #4, that conflict having itself a spot on the list of longest continuing conflicts.
7	Israeli–Palestinian conflict		2,921,208
8	David Beckham		2,553,561		Let's liven up with sports. In the world's most popular, football, the Netflix documental miniseries Beckham chronicled the English striker who currently is a co-owner of Inter Miami FC. In another that certainly ranks high on global popularity helped by the huge populations of India and Pakistan, cricket, the former is hosting the world championship (#5).
9	Cricket World Cup		2,370,824
10	West Bank		1,684,896		There are two territories claimed by the State of Palestine in the geographical state of... Palestine. The first is the West Bank, occupied by Israeli military, and is split into Palestinian enclaves and a contiguous Israeli occupied area. The second is the Gaza Strip (#1), ruled by Hamas (#2) as an oppressive de facto one-party state.

• These lists exclude the Wikipedia main page, non-article pages (such as redlinks), and anomalous entries (such as DDoS attacks or likely automated views). Since mobile view data became available to the Report in October 2014, we exclude articles that have almost no mobile views (5–6% or less) or almost all mobile views (94–95% or more) because they are very likely to be automated views based on our experience and research of the issue. Please feel free to discuss any removal on the Top 25 Report talk page if you wish.

This article was originally published on Diff on October 10 2023 (licensed CC-BY-SA-4.0). It shows how to generate and use QR codes that link to sound files that will read the introduction section of Wikipedia articles. The QR code on the right links to a reading of an article mentioned in the story below. Just scan the code with your phone and click the link given there. We'd be glad to hear your reaction to the reading, posted in the comments section below.

A new tool to bridge the gap between our physical surroundings and the digital realm came to life under the name of Sawtpedia. This tool enhances our exploration of items documented on Wikipedia by enabling users to listen to Wikipedia articles through the simple act of scanning a QR code generated by the tool itself. Sawtpedia opens up a fascinating audio dimension to the world of knowledge discovery.

In this post, we'll delve into Sawtpedia, exploring its origins, capabilities, and the potential it holds for knowledge dissemination.

Before we dive into Sawtpedia, it's essential to recognize QRpedia, the precursor of Sawtpedia. QRpedia, introduced in April 2011, is a mobile web-based system that employs QR codes to deliver Wikipedia articles to users in their mobile's language. Conceived by Roger Bamkin and coded by Terence Eden, QRpedia utilizes Wikipedia's API to determine whether a specified Wikipedia article is available in the language used by the mobile device used to scan the QR Code.

QRpedia has found its niche in various settings, including museums, GLAM (Galleries, Libraries, Archives, and Museums) institutions, historical cities, and even cemeteries.

Fascinated by this tool, I tried to introduce it to my country Tunisia. During Wikimania 2014 I initiated the idea of a new project in the medina of Tunis (UNESCO world heritage city) that I named Medinapedia. This project was executed by the Tunisian NGO Carthagina and with the collaboration of Wikimedia Tunisia and the Tunisian Association of the Preservation of the Medina – Tunis (ASM Tunis).

• 
  Medinapedia's leaflet at the Wikimania 2014 Community Village (User:Mrjohncummings CC-BY-SA-4.0)
• 
  QRpedia in the Medina of Tunis (User:Yamen CC-BY-SA-4.0)

However, after installing the QRPedia codes on the monuments within the Medina of Tunis, I realized that it would be more practical to listen to the Wikipedia articles rather than reading them while strolling through the historic city.

And that's how the concept of Sawtpedia dawned upon me!

So, what exactly is Sawtpedia? Sawtpedia, a fusion of "Sawt" meaning "sound" in Arabic and Swahili, and "pedia" as a reference to Wikipedia, serves as a tool that produces QR codes for Wikipedia articles. These articles, spanning various topics like monuments and museum collections, offer users the convenience of listening to them in the device's language.

Upon scanning a QR code, Sawtpedia fetches the Wikidata item for the associated article. If an audio file is available on Wikimedia Commons for that Wikipedia article in the user's mobile language via the Wikidata property P989, Sawtpedia plays it. In cases where such audio recordings are not available on Commons, Sawtpedia steps up by generating an audio version using the article's lead text through the gTTS Text-to-Speech System (under MIT License).

The concept was initially introduced at the WikidataCon 2021 and later put into action during the Hack4OpenGLAM when my friend Houcemeddine Turki stepped-in and assumed responsibility for the coding part and the deployment on Toolforge.

• 
  Yamen Bousrih (CC-BY-SA-4.0)
• 
  Houcemedine Turki (CC-BY-SA-4.0)

Sawtpedia is the result of collaborative efforts between Wikimedia Tunisia, Wiki World Heritage, and the Data Engineering and Semantics Research Unit from the University of Sfax.

1. Enhanced User Experience: Sawtpedia seeks to provide a better experience for visitors who scan QR codes. It bridges the gap between written content and auditory accessibility.

2. Foster GLAM partnerships between Wikimedia affiliates and museums: The tool can serve as a strong motivator for museums and local affiliates to collaborate closely. Museums can derive numerous benefits from Sawtpedia. (check this article: 7 reasons why museums should use Sawtpedia!).

3. Content Creation: The tool indirectly aims to foster content creation within the Wikimedia ecosystem. This includes the expansion of Wikipedia articles, Wikidata items, and audio files on Wikimedia Commons.

4. Promote accessibility and inclusivity: Sawtpedia can be used to promote accessibility and inclusivity by providing information to those who may have difficulty reading or accessing text-based content. For instance, the tool has the capability (work in progress) to provide the possibility to listen to content in local languages/dialects that are not configured on mobile devices (e.g. in some countries like India, most of the phones are configured in English while there are many local languages/dialects).

In conclusion, Sawtpedia is more than just a tool; it's a leap forward in making Wikipedia accessible to a broader audience. By combining the power of QR codes, Wikidata, and audio resources from Commons, it transforms the way we engage with Wikipedia articles. Whether you're exploring a museum, visiting a monument, or simply curious, Sawtpedia offers a new dimension to the world's largest open-access knowledge repository.

After a long Request for Comment at Help talk:Citation Style 1, consensus was reached for creating a new template to allow robust and standardized citations for revelations and epiphanies experienced by Wikipedia editors. The template, {{Cite altered state}}, offers a number of parameters representing a broad range of philosophical, religious, and pharmaceutical experiences.

{{Cite altered state
| thought          =
| author           =
| date             =
| state            =
| substance1       =
| dose1            =
| substance2       =
| dose2            =
| background-audio =
| audio-author     =
| url              =
| access-date      =
| medium           =
| like-whoa-man    =
}}

Examples:

• "God explains little and decides nothing." Revealed to me in a dream. Sleep: Selected Ambient Works 85–92, Aphex Twin. 2023-03-19.

• "Each human soul is a temporary fragment of the eternal, an ephemeral window from which the Absolute can be glimpsed." Psilocybe cubensis: 4 grams. August 1994.

• "An alligator with wings would just be a dragon." Whiskey: 6 shots. Subsequent substances: not clearly remembered. Like ten years ago. Like, whoa, man.

• "The government is bullshit." Really thinking about it: A few weeks. September 2011.

It's hoped that this template, and the accompanying citation guidelines, will allow us to finally put into articles the revelations, fully-baked ideas, cosmic truths, and toaster inventors that Wikipedia has always struggled to properly include.