Feel free to borrow/steal---the source is here: github.com/pos5747/notes The goal is a *still relevant* course on parametric models that can sit alongside a course on more agnostic methods for causal inference. *These are in-progress and written in the pre-Claude Code era.

You guys @carlislerainey.bsky.social has a free textbook online and it seems really useful pos5747.github.io/notes/

probably a good time to share this again: asteriskmag.com/issues/10/ca...

A massive seven-year project exploring 3,900 social-science papers has ended with a disturbing finding go.nature.com/4bZ9k0W

📄Published Today in Nature: 500 researchers reproduced 100 studies across the social & behavioral sciences to assess their analytical robustness (led by @balazsaczel.bsky.social & @szaszibarnabas.bsky.social). Article: www.nature.com/articles/s41... Preprint: osf.io/preprints/me... TLDR: 1/11

🧵1/ Our first meta-science paper (with 350+ coauthors) is published today in Nature. It presents one of the largest-ever reproducibility projects in economics & political science. Here’s what we found 👇

Today is the day that a lot of social scientists get their first Nature publication

I know how to solve the marriage and birth rate crisis and the good news is that it’s delicious

You can just make scones

My attempt at articulating why I think LLMs are a clear net positive for all research, and it's a mistake to ignore it despite private concerns: paulgp.com/2026/03/16/r...

This is a good example of the kind of anti-semitism that is common enough in leftist circles that I imagine most people don’t even notice it, but it’s really quite vile.

Each frontier AI model seems to use a little under a year's worth of a square mile of farmland's water to train. I think about this as the country having 4 square miles of farmland sectioned off to grow some of the most popular consumer products in history.

I forget who first pointed out that the pre-2020 internet is going to be thought of like low-background steel, but I think about that a lot.

We’re so lucky that these issues only affect the papers that smart people look at very deeply.

I will always re-post Toronto mango content

Money is useful for 1. Buying illegal drugs 2. Hiring hitmen 3. A medium of exchange that facilitates modern technological civilization Which one do you use it for?

This is a millennial problem, though we greatly improved overall coffee quality so you’re welcome. Gen Z cafes have plants and ornamentation and way better vibes.

I’m the lead of the Quarto project. We work hard to make it easy for people to use. It *thrills* me that LLMs got so good that they can take over much of the labor required. We’ll always work hard to make it easy for people to use Quarto. But less total human frustration is good, actually!

It's quite wild that the "LLMs are stochastic parrots" thing is still a meme on here given that at for at least the last 2ish years (since o1) so much LLM progress has come from reinforcement learning on things like math and coding problems (not next token prediction).

In a flurry of spending before the fiscal year ran out, in one month the DoD spent more than the annual budget of USAID, whom this administration had accused of waste that never materialized

The most downloaded paper on EconStor in Feb. 2026 was: "Briggs, Ryan C.; Mellon, Jonathan; Arel-Bundock, Vincent (2026) : It must be very hard to publish null results, I4R Discussion Paper Series, No. 281, Institute for Replication (I4R), s.l." hdl.handle.net/10419/336819 @i4replication.bsky.social

lol sometimes this place is like Portlandia but for 2021

This skeet is only two days old, but since then two more synagogues were shot at.

More people should be putting orange peel and dried chillies in their braises.

This is all great

Weirdly little known fact but you can hire grad students in a subject to tutor your kids on that subject and it’s cheaper and often much better than what is commercially available (Russian school of math or whatever).

this but for scientists who have p-hacked

This week a synagogue in Toronto was shot at, 3 Jewish stores were vandalized with swastikas in Montreal, and a student in Halifax was targeted for being Jewish and bullied out of high school.

I gave Zoë Kravitz & Penn Badgley a lunch recommendation in Paris (they asked) & then a bunch of people around us asked to take a photo with them & my mom didn’t know who they were & misunderstood the social situation & so asked if those people also wanted to take a photo with me & my now wife.

Opus 4.6 solved a problem Don Knuth (author of Art of Computer Programming) had been working on for weeks. www-cs-faculty.stanford.edu/~knuth/paper...

This makes sense to me, but I think we need to weight the machine’s ability to produce truth vs the human’s. In some cases, neither *guarantees* the truth but the machines can be shown to usually do better.

Mom, come pick me up, the best lack all conviction and the worst are full of passionate intensity.

Not sure what "entirely generated" means, exactly, but I feel we need new norms around this. Here's a blog I like about "etiquette", "consent", and what should be considered "rude" when sharing AI-generated content. Disclosures after the fact do not cut it, IMO. distantprovince.by/posts/its-ru...

Quantitative social scientists will have long, nuanced, passionate, (useful!) discussions of omitted variable bias like in the quoted post and then publish research based on whether their noisy tests are significant or not.

I wrote a short blog post that describes selection on significance in plain language and then proposes and criticizes two alternatives (selection on precision and registered reports). This is me working through my thoughts online, so feedback is very welcome. ryancbriggs.net/blog/the-pro...

Typos are good because they suggest I wrote this myself and even read over it and made edits. No AI posting here. Just authentic, typo-ridden human generated text

Starting to feel like an important part of my value add in my department is comes from me generating AI-written responses to assessments that profs use so that people can have informed conversations.

“It’s kind of like Halloween except with more vengeance and it’s own special cookie. Oh and also you’re supposed to get really drunk.”

Kids make all holidays like 1000x better.

As a Canadian, I’m not exactly shocked that this government would arbitrarily attempt to change a deal and then start hurling deranged threats when you don’t fold.

So does Bluesky like LLMs now?

I gave up trying to make a physical board and instead made a virtual one ryancbriggs.net/blog/galton-...

I wrote a short blog post that describes selection on significance in plain language and then proposes and criticizes two alternatives (selection on precision and registered reports). This is me working through my thoughts online, so feedback is very welcome. ryancbriggs.net/blog/the-pro...

Next up: mandatory not sharing data

Takeaways from social science foo camp: 1. There is so much interest in AI in this group (fair!) 2. Academics are pessimistic about higher ed (not just due to AI) 3. The unconference format can be really good

I have ~5 hours of open time in mountain view (near Shashi hotel). Recommendations? Google visitor center and then computer history museum? Eat at Zareen’s?

These colors don't run

In roughly zero minutes of my own brain time I got Codex to write a podcast transcriber + diarization (identifying who says what) tool for a podcast I like that lacks transcriptions (cooking issues). It'll take a while to run on all the back episodes, but this is just silly.

The p<0.001 thing is important because a lot of smart people think lowering the p value threshold will save us. I’m very skeptical of this.