Ryan Briggs

@ryancbriggs
489 Followers
204 Following
161 Posts
Raising kids & bread & grant money. Cleaning data & diapers & fish. International development & EA. Social scientist. Working at the University of Guelph, living in Toronto.
Websitehttps://www.ryancbriggs.net/
@atpfm y’all might find it interesting that Apple is already using LLM tech in spellcheck. It’s apparently roughly GPT-2 based https://jackcook.com/2023/09/08/predictive-text.html
A look at Apple’s new Transformer-powered predictive text model

I found some details about Apple’s new predictive text model, coming soon in iOS 17 and macOS Sonoma.

@nic @owenozier a bit out of date, but maybe useful for that use case https://alexandercoppock.com/Green-Lab-SOP/Green_Lab_SOP.pdf
@owenozier @nic those are my standard picks. Depending on your goals, I’d love if you included things on workflow or at least pointed people to that stuff: here is how you organize your folders and code, etc. I think people mess that up around as often as they mess up statistical issues.

New article in World Development Perspectives! With the amazing co-authors Tom Wein & Torben Fischer, we provide a framework for thinking about #dignity in the design, delivery, and donation-seeking for aid, as well as some initial empirical results that suggest we have much more work to do do.

https://t.co/BfGtPWb0aZ (and https://osf.io/rcwv2/)

cc @ryancbriggs

Checking out @ivory!
So the problem with improving your understanding of research methods is that you're angry all the time

So far this year under the new NAFTA:
* Canada won its case against the US on automobiles,
* won a case on solar panels,
* and lost a case that requires us to open up some of the dairy market.

So we kinda won three times.

@owenozier I don't remember “bad controls" language that well, but I recall them as being things that are on the causal path from x -> y. The table 2 fallacy can happen even after you correctly adjust for confounders. The issue is that people have a strong impulse to interpret the coefficients from the set of adjustment variables, even though they usually aren't identified.
@AndrewCharlesBaker teaching prep can squeeze out all research time if you let it. Do not let it.