Rohan Alexander
LightNews LightNews
banner
rohanalexander.bsky.social
Rohan Alexander
@rohanalexander.bsky.social
2.5K followers 1.3K following 400 posts
I spend too much money on books. And too much time writing them. 🇦🇺🇨🇦.
https://rohanalexander.com
http://tellingstorieswithdata.com
Posts Replies Media Videos
rohanalexander.bsky.social
Very happy to host Chris Barrie tomorrow at noon EST on "Reasoning models and synthetic data generation".
All welcome.
Zoom: utoronto.zoom.us/j/4784708970
Details: rohanalexander.com/tdw.html
Wednesday 26 November 2025, noon (EST) Christopher Barrie, New York University "Reasoning models and synthetic data generation" Christopher Barrie is an Assistant Professor of Sociology, at New York University as well as Core Faculty at CSMaP and a Research Fellow at the Department of Sociology at the University of Oxford. His research interests include computational methods, generative AI, political attitudes, communication, protest, conflict and MENA. He holds a DPhil in Sociology from Oxford.
November 25, 2025 at 2:19 PM
rohanalexander.bsky.social
Cannot make this up.
Screenshot of a paper "What does current genAI actually mean for student learning?" under a popup of Science Direct's AI bot thing.
November 24, 2025 at 1:34 AM
rohanalexander.bsky.social
Allowing billionaires to collude is the most Canadian thing that could happen to the Hudson's Bay charter
Screenshot of an article in the Globe and Mail with the headline "Thomson and Weston families submit $18 million joint bid fro Hudson's Bay charter.
November 15, 2025 at 3:56 PM
rohanalexander.bsky.social
There are three Toronto Data Workshop online talks coming up 12, 19, and 26 November, noon EST. This first is this Wednesday at noon on Zoom: utoronto.zoom.us/j/4784708970. All welcome.
Wednesday 12 November 2025, noon (EST) Georgia Channing, Hugging Face "AI for Scientific Discovery is a Social Problem" Georgia Channing is the ML for Science Lead at Hugging Face and a DPhil candidate at the University of Oxford. Her academic background is in machine learning, and her research has spanned neural architecture search, protein classification, deepfake audio detection, and hyperspectral imaging. She holds an undergraduate degree from the University of Tennessee and a Masters degree from the University of Oxford. She works now on building tools and methods to advance scientific achievement. Wednesday 19 November 2025, noon (EST) Mellissa Meisels, Yale "Strategic Campaign Attention to Abortion Before and After Dobbs" Mellissa Meisels is an Assistant Professor in the Department of Political Science at Yale University, where she is also affiliated with the Institution for Social and Policy Studies and Center for the Study of American Politics. Her research centers on the U.S. Congress, money in politics, electoral campaigns, and political organizations. Much of this work investigates the strategic choices of candidates and financial contributors in congressional campaigns, with an emphasis on primary elections. To do so, she employs original data, natural language processing, causal inference tools, experiments, and structural estimation of formal models. Her research is forthcoming and published in Proceedings of the National Academy of Sciences, Journal of Politics, British Journal of Political Science, and Political Science Research and Methods. Wednesday 26 November 2025, noon (EST) Christopher Barrie, New York University "Reasoning models and synthetic data generation" Christopher Barrie is an Assistant Professor of Sociology, at New York University as well as Core Faculty at CSMaP and a Research Fellow at the Department of Sociology at the University of Oxford.
November 11, 2025 at 2:15 PM
rohanalexander.bsky.social
Happy new Helen DeWitt book day for those who celebrate (which in a just world would be everyone)!!!
A photo of the cover of the book "Your name here" by Helen DeWitt and Ilya Gridneff.
November 9, 2025 at 8:43 PM
rohanalexander.bsky.social
Enjoying these graphs.
Two scatterplots show daily Canadian federal procurement contracts awarded from 2004 to 2025. Both plots include vertical dashed lines marking federal elections and use logarithmic vertical axes. Panel (a), titled Number, plots the total daily number of awarded contracts against date. Most days cluster between 10 and 1,000 contracts, with periodic fluctuations but no clear long-term trend. Panel (b), titled Value, plots the total daily dollar value of awarded contracts over time. Most points fall between $100,000 and $1,000,000,000, with noticeable variability but no consistent upward or downward pattern. Overall, both panels show persistent daily activity across two decades, with no sharp discontinuities corresponding to election periods.
October 30, 2025 at 4:25 PM
rohanalexander.bsky.social
Got my own office coffee set-up (staff at L'Espresso were rude one too many times!)
October 28, 2025 at 3:14 PM
rohanalexander.bsky.social
Number of unique speakers in the Australian parliament, by gender, 1998-2025. By Lindsay Katz and @inessadeangelis.bsky.social.
A graph from 1998 to 2025 with date on the x axis and number of unique speakers on the y axis. There are a lot of dots, but it broadly shows the number of women rising over time, to around 40 in 2025.
October 28, 2025 at 12:00 AM
rohanalexander.bsky.social
GitHub Actions is my passion.
October 25, 2025 at 5:56 PM
rohanalexander.bsky.social
Looks great @vincentab.bsky.social !
October 25, 2025 at 12:54 AM
rohanalexander.bsky.social
Funny that Dan Wang's book came out the same year Mokyr got the Nobel; while the one phrase review of Wang is engineers vs lawyers, the similarities between what he ways about US manufacturing and Mokyr on the industrial revolution is striking.
(Mokyr in the 4th image is from Anton Howes).
October 20, 2025 at 12:09 PM
rohanalexander.bsky.social
Sunday morning trifecta: espresso, croissant, and NSERC.
Photo of a croissant, an espresso and a computer with an NSERC page.
October 19, 2025 at 12:27 PM
rohanalexander.bsky.social
Amen
An actor in the dark lit by maybe a campfire, talking to a little girl and the caption reads "I'm 39. That's not middle-aged. I am a very young adult.". IDK what movie it's from but it's a still from Amazon Prime.
October 17, 2025 at 1:22 PM
rohanalexander.bsky.social
Yay, a Nobel partly for econ history! (implicitly just a lot of careful data work)!
October 13, 2025 at 10:10 AM
rohanalexander.bsky.social
Sure it's just one word in the README, but still did something to help ghostty - ghostty.org - the best terminal emulator out there!
October 12, 2025 at 2:31 PM
rohanalexander.bsky.social
Great book 🇨🇦🥅🍁
October 11, 2025 at 10:41 AM
rohanalexander.bsky.social
Fall's here
October 10, 2025 at 6:26 PM
rohanalexander.bsky.social
Hilariously the senate and the reps disagree on when Richard Hartley Smith Abbott died! And you people wonder why I have data trust issues.
Screenshot showing Senate saying he died on 28 Feb and the reps saying he died on 27 Feb.
October 8, 2025 at 10:02 PM
rohanalexander.bsky.social
Sure you might be a pedant, but are you a "finds mistakes in Hansard" level of pedant? (Feeney wasn't in parliament on this day). Lindsay Katz FTW!
Screenshot of Hansard, with "Mr Feeney interjecting" highlighted
October 8, 2025 at 3:40 PM
rohanalexander.bsky.social
"Overcome Imposterism" "Manage Procrastination" "Tame Perfectionism". Real "tag yourself" moment in terms of which workshop you sign up to.
A list of four academic workshops under the heading “Workshops.” The sessions include: “Give Effective Academic Presentations” on Monday, October 6, from 4:00–5:30 pm; “Build Academic Confidence and Overcome Imposterism” on Thursday, October 9, from 12:00–1:15 pm; “Manage Procrastination” on Monday, October 20, from 12:00–1:15 pm; and “Tame Perfectionism” on Thursday, October 23, from 1:00–2:15 pm. Each workshop title is a blue, underlined link.
October 7, 2025 at 1:46 PM
rohanalexander.bsky.social
Data scientists: “OMG we had to learn a new programming language for this project”
Anthropologists: “We literally learnt two languages”
Photo of a book with the following quote circled: “Learning to speak !Kung was a laborious task. No !Kung-English speaker was available, so one had to work with two interpreters. Onesi-mus Mbombo translated my questions from English to Setswana, and Gakekgoshe Isak (the son of Headman Isak and a fluent !Kung speaker) from Setswana to !Kung. The answer came back like a broken telephone from Kung to Setswana and Setswana to English. This cumbersome system spurred me on to learn! Kung quickly, but in fact during the early months I learned more Setswana than !Kung. By March 1964 I could use Setswana in interviews and to elicit more !Kung vocabulary. The !Kung word can be described as an explosion of sound surrounded by a vowel. The bundle of clicks, fricatives, and glottal stops that begins most words makes !Kung a difficult language to record, let alone learn to speak; and the difficulty is compounded by the extremely rapid-fire delivery of !Kung speakers. Nevertheless, month by month my facility progressed until it was possible to dispense with the services of the English-Setswana interpreter a year into the fieldwork.”
October 5, 2025 at 12:25 AM
rohanalexander.bsky.social
How early is too early for children to watch Margin Call?
Two children watching Margin Call.
September 24, 2025 at 11:05 PM
rohanalexander.bsky.social
And now there are two on the ice!
September 20, 2025 at 4:37 PM
rohanalexander.bsky.social
Going to be upset if William Sealy Gosset doesn’t get a guernsey.
September 20, 2025 at 4:28 PM
rohanalexander.bsky.social
Preprint of “Prompting the Professoriate: A Qualitative Study of Instructor Perspectives on LLMs in Data Science Education” available: arxiv.org/abs/2509.12283

By Ana Elisa Lopez-Miranda (an exceptional undergrad!), @tiffanytimbers.bsky.social, and me.

Comments appreciated!
Prompting the Professoriate: A Qualitative Study of Instructor Perspectives on LLMs in Data Science Education Ana Elisa Lopez-Miranda, Tiffany Timbers, Rohan Alexander Large Language Models (LLMs) have shifted in just a few years from novelty to ubiquity, raising fundamental questions for data science education. Tasks once used to teach coding, writing, and problem-solving can now be completed by LLMs, forcing educators to reconsider both pedagogy and assessment. To understand how instructors are adapting, we conducted semi-structured interviews with 42 instructors from 33 institutions in 10 countries in June and July 2025. Our qualitative analysis reveals a pragmatic mix of optimism and concern. Many respondents view LLMs as inevitable classroom tools -- comparable to calculators or Wikipedia -- while others worry about de-skilling, misplaced confidence, and uneven integration across institutions. Around 58 per cent have already introduced demonstrations, guided activities, or make extensive use of LLMs in their courses, though most expect change to remain slow and uneven. That said, 31 per cent have not used LLMs to teach students and do not plan to. We highlight some instructional innovations, including AI-aware assessments, reflective use of LLMs as tutors, and course-specific chatbots. By sharing these perspectives, we aim to help data science educators adapt collectively to ensure curricula keep pace with technological change.
September 18, 2025 at 10:33 AM