Jason Roos

I'm not overfull...>>sob<<...YOU'RE overfull

These are the days of procrastination and progress

Elrond Halfelven sounds like a real asshole

You may have just solved a seven-hour-old mystery for me! My best guess was a feral parakeet

it's easy to think that academia is outdated, worthless etc but the fact that we are consistently the first targets for repression by dicators and wannabe dictators suggests we are still doing something useful and good for free societies despite our best efforts to fucking suck super bad

Great job everyone, we did it

On the way to #Berlin to present at the Digital Economy Workshop @diw.de @bsoeberlin.bsky.social
On the agenda:
``Digitization and Community Participation'' w/@jmtroos.bsky.social, now featuring a new dataset covering 600K+ events, 400K+ individuals in 12K local communities in 2024 (200M+ rows...)

Whisks are 100% Bouba.

@shire-reckoning.bsky.social hasn't posted in the last day and now I'm worried Sauron found the one ring

Neuroforecasting reveals generalizable components of choice Abstract. Accurate forecasts of population-level behavior critically inform institutional choices and public policy. While neuroforecasting research sugges

Our latest paper on #Neuroforecasting is out at @pnasnexus.org (w/ Brian Knutson and Lester Tong)

A short 🧵

tinyurl.com/e4279exp

I'm checking this every day, great deals on heavy (5+ lbs) ashtrays

* All authors contributed equally and are listed in order of emotional neediness.

This is what the Internet was meant for

Remember that email is a replacement for physical letters, which used to arrive only after spending the better part of a week on a train.

Same

For consistency in a decision theoretic framework. I mean, you can define "near" anyway you like, but if you want to make a theoretical argument, you might have a problem.

A potential problem with the proposal is that "near" takes on different meanings when using the mean (L2, SD) versus median (L1, MAD).

I prefer median when talking about people -- means often describe people who don't exist, medians are less prone to this

I often find myself arguing for why I do impact-driven research on climate adaptation and sustainable behavior. These projects are often long, messy and with unclear outlets that not always count for career progression in my field.

Dashes en – em are always good choices

god “black cow” is such a groove

Here are the first five sets of slides:

01 Introduction: psantanna.com/DiD/01_Intro...

02 Classical 2x2 setup: psantanna.com/DiD/02_two_b...

03 Clustering issues: psantanna.com/DiD/03_Clust...

04 Functional form: psantanna.com/DiD/04_Funct...

05 Covariates: psantanna.com/DiD/05_Covar...

Where is Parson Brown during the events of the song? Or better yet—*what* is Parson Brown?

The area of Andorra is 470 square kilometers (180 square miles). To put it in perspective, the area of San Jose, CA is 180 square miles.

So Andorra is about 2.5 times bigger than Lichtenstein or DC.

Lichtenstein has about 40,000 people, Andorra has about 80,000.

Do Andorra next

It's a useful analogy, and highlights how AI gets vilified by virtue of being new. I remember academic scare around Wikipedia. But there are also differences: Wikipedia is in principle traceable back to a source. And did anybody hype Wikipedia as a replacement for creating original works?

In any case my point was simply that calling the thing that everybody is referring to when they AI "evil" isn't an obviously wacko position to take. I can see where that comes from.

I think the actual contents of the "public" data is where a lot of the vitriol originates from. I mean, can any foundation model be said with certainty not to contain any unauthorized content? I'm not especially bothered by it, but I also don't create art for a living.

I don't think they are evil per se, but I can understand why, e.g., an artist whose copyrighted work was ingested into a text or image generator, and then encounters claims that these things can generate original writing or art, why they might reach for "evil" to describe all that.

Skeletor torments He-Man with powerful intuition about Bayesian reasoning and the nature of information