Yum.

Which, incidentally, would have saddened Hugh Kenner…

Never read Timothy Donnelly before. But I’m very much enjoying the loping, looping fluency of poems in The Problem of the Many

This is James J. McCarthy at the Univ of Illinois, instructing the instructors of the deaf in psycholing. advances: "The essential distinction between the mech. communication model and the psycholing. model of lang. function is that the former fails to take into account the 'meaning' of signals"

It'd be fascinating to pursue the history of usage of the term "language model". Not the familiar pedigree of LMs qua probabilistically-constrained-token-sequences (ngrams, audio signal processing, embeddings, transformers), but the variety of efforts called 'LMs'. E.g. this from 1957:

Not so systematic, but here's a literary flourish from Scientific American, 1900, reporting on T.J.J. See's improved estimates of planetary sizes:

Apparently they were still at it in 1876! Here's the gendering on a page from the 1876 Baltimore Sun Almanac: He/him: The Sun, Mars, Jupiter She/her: nature, Venus, The Earth, The Moon Indeterminate: Mercury

John Paul George Ringo

Arrived by post earlier this week! Tiny correction to p168 n14 - the entry on 'mediation' in Tom Bottomore (ed.) 'Dictionary of Marxist Thought' is by István Mészáros. Went there from your v. elegant one-sentence summary. I'd never thought of Aristotelian mean as background here - thinking on it.

Arrived in the post today. One of the odder little tributaries in the history of word embeddings: Richard League’s 1977 Psycholinguistic Matrices. With no little hippy enthusiasm, the author uses Charles Osgood’s 1950s semantic differential embeddings to study semantic satiation.

Some crazy shifts in the global distribution of science over the last decades.

ROFL. Living in the eternal now.

#Wikimania2025 keynote: Joyce Nakatumba-Nabende, Vukasi Marivate, Selena Deckelmann and Job Meaura keynote panel on #AI: how should Wikimedia projects adapt, respond to and shape the AI landscape?

At Wikimania 2025 in Nairobi! A fantastic welter of conference activity here. Listening to Ruby D. brown and Nwonwu Uchechukwu Pascaline present their research on barriers to African women contributing to Wikipedia #AfricanWikiWomen

Most of these women aren't yet in any of the different language wikipedias. Only 237 Malawian women have a wikipedia page (31% of all Malawians who do). But some exciting plans are brewing in Scotland & Malawi to generate more English, Chichewa and Chitumbuka pages about Malawian women.

Short 🧵today on Malawi: population 20.3 million. Wikidata currently knows of 1541 Libyans. 739 (50.93%) of them are women, including one trans woman. This high percentage of women is the result of some recent work, but it takes the long-running 50:50 campaign in Malawi as inspiration.

Each day this July I'll survey Wikidata's coverage of women in a different country. I'll focus on 30 countries, amounting to half the global population, where women are under-represented relative to the country's population. Just Wikidata? Or a more general digital problem? #1Day1WikidataCountry

Similarly. the writer Kawthar Al-Jahmi (كوثر الجهمي) is only there in Arabic Wikipedia, and not yet in any other language wikipedias.

The actress Khadija Sabri (خدوجة صبري) is another Libyan woman who has her own page on Arabic Wikipedia, but doesn't yet have one in English.

For example, the politician Saniya Ghouma (سنية غومة) has a page in both the Arabic wikipedia and the French wikipedia. But there's not yet a page about her on the English wikipedia.

All kinds of remarkable Libyan women are on this redlist, from peacebuilders to powerlifters: academics, activists, actors, athletes, chess players, deminers, doctors, fashion designers, filmmakers, journalists, judges, karatekas, lawyers, local politicians, marine biologists...

🧵today on Libya (ليبيا) : population 7.3 million. Wikidata currently knows of 1096 Libyans, of whom 306 (29.5%) are women. Unfortunately, most of these women aren't yet in any of the different language wikipedias. Only 91 Libyan women have a wikipedia page (12.3% of all those Libyans who do).

Here's a bubble chart of Malawians in Wikidata by gender and occupation. (Men in orange, women & other genders in purple.) Not quite 50-50, but getting there! #Malawi #50-50Campaign #DigitalRepresentation

Yay! I've reached my first #Wikidata global representation goal! Women from the countries shaded red are 25% of global population, but in Dec '24 were only 0.83% of Wikidata bios. I set a target to increase this to 1%. Four months later, with 10,000 women added, they're now 1.01% of Wikidata bios!

🧵reporting February progress in improving #Wikidata global representation. Women from the countries shaded red are 25% of global population. In Dec '24 they were 0.83% of Wikidata bios. This February I added almost 2000 more women, mostly from Pakistan, Yemen, Chad & Bangladesh.

🧵on global digital representation. Countries marked in red on this map contain 50% of global population. But people, especially women, from these countries are digitally under-represented. Women from these countries - a quarter of humanity - were just 0.87% of #Wikidata biographies in Dec 2024.

I've read people alleging that this cybertruck is on fire. Even in the heat of the moment, let's not rush to hasty conclusions! All I can see here is an awkwardly expansive exuberance, an impossible warmth of heart, an oddly dramatic gesture of oxidization.

The Levi-Strauss remark btw (in case anyone else was wondering) is within a sequence of paragraphs where he's trying to rebut imagining criticisms in the introductory 'Overture' to The Raw and the Cooked (p7 of my 1970 Harper Torchbook paperback edition)

I did wonder how old this sense of 'riveting' was, though - so here's the OED. Start's with Dibdin's Complete History of the English Stage, and ends with Law and Justice As Seen on TV...

At the other extreme, 19 countries have fewer than one per 100,000 population. Together, these countries amount to 46.9% of the world's population. But they only provide 3.3% of Wikidata's women. We're a long way from #WorldWideWikidataWomen