Nic Weber

"You like School House Rock, right? Ok, this is Math Rock. Basically the same thing."

This is was essentially Tom Griffiths' keynote this morning at COLM

This is good. Like really good.

This is wild coincidence... I was just pushing Claude on category theory and it went very sideways theorizing about morphology.

Also v good bsky.app/profile/rach...

This is v good

...declining state support, threata and real losses in grant funded research, demographic shifts, demonization of liberal classrooms, etc ... AI is an accelerant for a sector already on fire, but it's insane to say this is a "students future" without any consideration for our present.

If all of this forced higher ed to reconsider its methods of content delivery and retrain it's workforce - I think a lot of us (students, professors, support staff) would be thrilled - BUT ITS NOT. Because... Well everything. Most employees in HE are living under a shadow of terror introduced by...

Another broadly discussed tactic is "oral exams"...ok, but with absolutely no consideration for what this testing method measures, and no support for student preparation for a new paradigm of evaluation? It also does not scale without some serious reorientation of in-person instruction norma

Most instructors are trying the same tactics to theaet overreliance on AI, but it's all so meek and most of it is failing. "Ask gpt and then critique it's response" is the equivalent of a programming instructor saying "in this class we only read code you can find on GitHub" - it has v limited value

Now replace Tesla with OpenAI; and driving with literally any subject in higher education. This is college instruction in 2025

Next year, all 50 state DMVs report widespread failure for road tests. The rate of licensed drivers in the US population decreases steadily and then falls off a cliff. Politicians and print media blame driving instructors - why can't they just update their curriculum to meet student needs?

'Examining the effects of birth order on personality' in PNAS -> across multiple controls in three countries (UK, Germany, USA) "...we conclude that birth order does not have a lasting effect on broad personality traits outside of the intellectual domain." www.pnas.org/doi/10.1073/...

Prompt Stability Scoring for Text Annotation with Large Language Models Researchers are increasingly using language models (LMs) for text annotation. These approaches rely only on a prompt telling the model to return a given output according to a set of instructions. The ...

I really like @cbarrie.bsky.social's prompt stability work for text annotation arxiv.org/abs/2407.02039

Simulacra is so back!

This is a very cool paper - curious about doing research with a HS student - is this part of a program or one off scenario?

This tracks with political science theory! Eitan Hersh argues that political hobbyism depends in part on the perceived threat of the party in power www.eitanhersh.com/uploads/7/9/...

today we will all read imbens 2021 on statistical significance and p values, which is a strong contender for having the best opening paragraph of any stats paper

pubs.aeaweb.org/doi/pdf/10.1...

Every department chair rn:

Moodys ratings are probably a bad strategy for effecting public policy change - but showing overtly 'law and order' voters that moral failings also have broader economic consequences might be pursuasive

Guzman and Clark estimate that the larger the share of non-white victims of police violence the lower the cities’ credit ratings ... causal attribution is hard here, but I think the inquiry is smart. use externally valid fiscal measures to quantify the impact of excessive force

Estimating the Negative Credit Impacts of Police-involved Fatalities - Tatyana Guzman, Benjamin Y. Clark, 2025 George Floyd's killing in Minneapolis spotlighted the role of police-involved fatalities that cause immeasurable harm to victims’ families and communities,...

Police brutality is bad business journals.sagepub.com/doi/abs/10.1...

"Voters strongly support policy experimentation and particularly so when they do not hold a strong opinion about the policy."

In follow-on survey experiment "we find that politicians conform their expressed opinion about policy experiments to what we tell them the actual opinion of voters is."

Who's Afraid of Policy Experiments? Abstract. In many public policy areas, randomised policy experiments can greatly contribute to our knowledge of the effects of policies and can thus help t

Randomized policy experiments are often assumed to be hard, expensive (financially and politically), and disliked by voters...

"Collecting survey data representative of the Dutch electorate, we find clear evidence contradicting this view."

doi.org/10.1093/ej/u...

Trump’s first-week strategy: ‘Flood the zone.’ Repeat. It was all part of a plan to begin with a bang and follow a detailed policy blueprint, although key administrative posts remain stuck in a hiring bottleneck and many of his directives have lacked imme...

TBF - I don't fault anyone covering this beat ... the whole point of this administration's chaos is to "flood the zone" ... Its hard to compete for time and attention when literally and figuratively things are on fire elsewhere...

This is an economic story as much as it is a story about civics, education, etc.

Imagine if every employee at Ford Google and Target was fired, and Amazon reported 0 profit in 2025...