Ethan Bueno de Mesquita

@ethanbdm.bsky.social

Studying accountability, national security, electoral accountability, political economy, tech & society, applied game theory @HarrisPolicy .

We live in an age of technological wonders. And yet, it is a bi-partisan consensus in the American government that the only way to protect the traveling public from the national security threats posed by a small tub of cream cheese is through TSA confiscation. That's just the price of freedom.

I see a lot of folks posting about that NYTimes article. So I'm just going to say something I know you all already know, but is important. You cannot infer the effect on unemployment of majoring in one topic vs another by comparing the unemployment rates of folks who chose those majors.

I asked AI whether this was safe and it reassured me by pointing out that an important paper by Fearon and Powell (1979) proves that AI-driven nuclear command and control with a human on the loop guarantees lower risk of nuclear war in equilibrium.

WIRED@wired.com · 12mo ago

Human judgement remains central to the launch of nuclear weapons. But experts say it’s a matter of when, not if, artificial intelligence will get baked into the world’s most dangerous systems. www.wired.com/story/nuclea...

My son is home from college for the summer. He truly has been indoctrinated by high ed ideology. Anyway, after one too many disagreements, I had to lay down the law. I told him that as long as he is living under our roof he will respect our values and be a linguistic prescriptivist.

I've been using ChatGPT (both o3 and 4.0 depending on task) to help copy edit a new paper. While improving writing, it routinely introduces mistakes into previously correct proofs. But it also finds errors and logical holes I missed. It contains multitudes.

Leave aside how endowments work, I'd like people to just understand that if the overhead rate is 60%, that means the university is collecting 35% of the research $ for admin costs, not 60%. Overhead on $1 million grant is $0.6m. So total award is $1.6m and admin percent is .6/1.6 = 35%.

Current stats education for medical/other clinical students is bad and misaligned to what they really need to know to function. It teaches them to produce research in 1993, when we should be teaching most of them to be informed consumers today. We *should* teach something like this:

BildBildBildBild
Zach Binney@binneyz.bsky.social · 2y ago

You could do it in one semester. The course I teach at Emory to first-year undergrads could work with some relatively moderate modifications! The problem is we currently waste the one stats semester we require so many clinicians to use on less worthwhile things like t-tests and linreg assumptions.