Don't blame the historians. They have been trying, but we wouldn't listen.

I propose that, in anticipation of legislation forbidding impersonation of humans, AI chat bots shall use the following singular, first person pronouns:
*AI*: nominative
*meh*: accusative
*mai*: dependent
*aine*: independent genitive
*aiself*: reflexive

PSA: for your mental health and sanity do not read Fortune commentary pieces by a 23 year old CEO of a startup that sets other people's money on fire and claims he is not needed for the "technical work", but still wants to get paid for software he can vibecode, but you cannot, for some reason.

Explainability, especially when used in the same context as "trustable".

That would have been so much better as a small multiple of rug plots though...

I just completed my first course (thanks @frank.computer) from the Open Visualization Academy (OVA). I am in awe of the quality of the labor and care that went into building this free repository. What a beautiful gift. 📊

What a wonderful group of people! Wow!

Screenshot of the OVA website, showing all courses

The Open Visualization Academy (OVA) IS LIVE!
openvisualizationacademy.org

Thread follows #dataViz #infographics #dataJournalism #dataVisualization

df <- data.frame( Bio = c(30, 10, 5), Fossil = c(55, 60, 70), Electrons = c(15, 30, 25), Group = c("India", "China", "United States") ) ggtern( data = df, aes(x = `Bio`, z = `Fossil`, y = `Electrons`, color = Group) ) + geom_point(show.legend = FALSE) + scale_T_continuous(breaks = seq(0, 1, .2), minor_breaks = NULL) + scale_L_continuous(breaks = seq(0, 1, .2), minor_breaks = NULL) + scale_R_continuous(breaks = seq(0, 1, .2), minor_breaks = NULL) + theme_rotate(radians = -pi / 3) + theme_clockwise() + scale_color_manual( values = c( "India" = "orange", "China" = "red3", "United States" = "dodgerblue3" ) )

a very simple version with code in the alt text

{{ggtern}}

A line chart shows changes in “net support” from 2020 to 2026 for seven demographic groups: Nonwhite, Age 18–29, Not midterm voters, All registered voters, Midterm voters, Age 65+, and White. The horizontal axis marks years 2020, 2024, and 2026. The vertical axis ranges from –50 to +10 points, with a horizontal zero line. Each group has three data points connected by a smoothed line. All groups except White have negative net‑support values across all years. White voters show small positive values that decline slightly over time. By 2026, the labeled endpoints show: Nonwhite around –46 Age 18–29 around –45 Not midterm voters around –16 All registered voters around –13 Midterm voters around –11 Age 65+ around –5 White around +4 Overall, most groups trend downward between 2024 and 2026, with White voters remaining the only group above zero.

More like this would have been better then? I like that they're trying and ignoring people who want them to "just do normal line and bar charts". Not enough that I would resubscribe, but still...

Binge-watched the entire series. Loved every minute. A thousand thanks.

Part 1: How do LLMs work? YouTube video by Andrew Perfors

I just created a series of seven deep-dive videos about AI, which I've posted to youtube and now here. 😊

Targeted to laypeople, they explore how LLMs work, what they can do, and what impacts they have on learning, well-being, disinformation, the workplace, the economy, and the environment.

You couldn't have given them a graphing calculator?

Regulators ordered The Therac‑25 taken out of service immediately. Distribution of unsafe blood products was halted before the Krever investigation. Why is this government treating AI differently, when a system on the market today can be used to generate illegal and harmful material, including CSAM?

Yup. Table is a chart type.

Behold: the first-ever list of news outlets that have banned generative AI in their reporting. As of today, this is literally information that you cannot find on Google.

My goal is to fill the starter pack, so please send over suggestions with supporting evidence!

go.bsky.app/8cn1XfT

I love OWID and that they give you the datasets with their charts is just so absolutely the-right-thing to do. And yeah, that also means anyone can quickly remake any chart they publish.

Un-stacked area chart showing the share of the total value of merchandise imports from each partner region, expressed as a percentage of total import value, from 1949 to 2024. X axis spans years 1949 to 2024 with decade markers; Y axis runs 0% to 100%. Points and text annotation at maximum and latest value. Facet labels show regions: North America, Europe, Oceania, Africa, South America, Asia (excl. China), and China.

I find small multiples a bit easier to interpret:

My eyes! The color #ff00fe should be restricted to things that need to be removed prior to publication. Designers, please calibrate your monitors? Or turn off Night Shift when you're working? Tx,

"ein platter, geistloser, ekelhaft-widerlicher, unwissender Scharlatan, der, mit beispiellosen Frechheit, Aberwitz und Unsinn zusammenschmierte, welche von seinen Anhängern als Weisheit ausposaunt und von Dummköpfen richtig dafür genommen wurde"

A fresh layer of snow dampens sounds. For people who find the world too loud, that can be wonderful.

Went to the library to pick up a hold. A man behind me asked what the book was about and why I was reading it. I joked I should leave a note in each library book that says: “if you read this, you might want to join my book club”. He had interesting suggestions for further reading. #5minutebookclub

I mean that you must always measure the same thing, so you cannot have state-of-the-art benchmarks, you must use consistent benchmarks.

Except that "metrics such as accuracy and speed, compared to state-of-the-art benchmarks" are not stable over time, so fair comparisons are not possible.

5 bars with relative lengths 25, 11, 4, 1 and 0

Lib Dems are really bad at making charts though.

Trans News Initiative A database and analysis of news coverage affecting trans communities.

This is superb #dataviz.📊 transnewsinitiative.org

The printing press would never taken off it it hadn't been for the introduction of paper in Europe, a few centuries earlier. But nobody ever says "Paper democratized knowledge."

There was never any need to translate the term. The word is the same in French and English. Grain, as Bertin used it, referred to the variable roughness of paper that results in varying levels of light/dark. We still still call that paper grain.

tellis chart of preference for literary genre by party sorted by percentage from high to low. Reform prefers Horror, Conservatives Travel, Greens Romance and Labour self-help.