Sign in

drewfudenberg.bsky.social

@drewfudenberg.bsky.social
987 followers 159 following 146 posts
PostsRepliesMedia
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 23/09/2026
And a paper on when slowing down AI research can be sustained in equilibrium: economics.mit.edu/sites/defaul...
economics.mit.edu
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 09/09/2026
That’s one possibility. Or there could be a principal with two agents? Or? Even with two agents there’s not a single way to go. Levine and I have a long run agent paying a control cost to influence a short run agent, but you could imagine a two agent model with eg a Nash bargaining solution.
230
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 09/09/2026
More verbosely, two seems enough for most of the things economist have been trying to explain, but there are definitely cases were more than two would be better
130
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 09/09/2026
Parsimony :)
230
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 06/09/2026
Thanks!
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 06/09/2026
Links please?
100
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 06/09/2026
True but she was then the president of her university…the Lavigne comparison is more apt.
020
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 05/09/2026
Fluctuating environments might also favor genotypes that induce stochastic phenotype switching. here is an obscure paper with a very simple model of that. www.biorxiv.org/content/10.6...
biorxiv.org
Rapid evolution can select for fitness tradeoffs in fluctuating environments
In fluctuating environments, the fates of new mutations depend on the fitness tradeoffs they experience across multiple environmental conditions. Selection on these tradeoffs is well-understood when m...
120
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 17/08/2026
That’s great! I’ve been very much out of the loop on this since TE was adopted by the econometric Society
030
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 17/08/2026
It would be great if you did! If you go the TE route of open access, no salaries you still need a source of funds for the website and copy editing . Martin Osborne originally got us free hosting at Toronto and did the copyediting himself, but that’s gonna be a hard set of gifts to replicate.
140
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 16/08/2026
Ted Bergstrom was one of if not the first to sound the alarm on this: www.econ.ucsb.edu/%7Etedb/Jour...
econ.ucsb.edu
0120
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 16/08/2026
I’ve always thought it was too bad that the Game Theory Society gave ownership of GEB to Academic Press when it was founded as opposed to retaining ownership themselves. The fact that it and the JET belonged to commercial publishers the main reason we started Theoretical Economics.
1348
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 10/08/2026
The independent case gets tougher for me. If only some small fraction of people survive not sure what the break point is. One in 1000 seems pretty ridiculous is one 10 OK?
020
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 10/08/2026
Yeah, those pesky IRB’s :-)
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 10/08/2026
I don’t have a very firm sense of how I would like to answer, but my best guess is that if we were really sure that there was no correction and every member unanimously supported taking this risk I would find it hard to oppose them even though it strikes me as creepy
110
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 10/08/2026
Yep! As I say in first year micro, the content of a continuity assumption depends on the space its defined on. Not that that properties particular to continuity for example, also matters for the independence axiom.
120
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 09/08/2026
in principle, yes, and maybe the realized outcome really will be moderate, I certainly horeo. but 80:20 eiter things are wonderful or humanity is destroyed is a strange kind of “moderate “position
220
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 03/08/2026
Moreover, like the tools it’s built on refine keeps getting better. The current version catches errors in unchanged proofs that last year’s version missed
130
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 01/08/2026
As your article says, journals will eventually have to adapt to these changes in the production process. I wonder how much referees tastes and standards will change as well.
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 29/07/2026
here are two recent papers of mine on the economics of AI and a third that’s recently revised economics.mit.edu/sites/defaul..., economics.mit.edu/sites/defaul..., economics.mit.edu/sites/defaul...
economics.mit.edu
150
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 27/07/2026
Hi Ben, nice paper. I especially liked the result about the heterogeneous impact of improvements in AI quality. I suspect some of your results don’t really need to Cobb Douglas assumption how much have you tried to relax that?
000
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 22/07/2026
Increased morbidity gap could be due to increased unhealthy lifetime holding healthy lifetime fixed. Increased healthy lifespan is better than increased unhealthy lifespan, but many people would prefer to be alive but sick as opposed to dead.
011
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 01/07/2026
Will online/supplemental appendices be allowed? If so what can go in them?
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 11/06/2026
Or ignored edge cases, which actually is formally a mistake
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 11/06/2026
To be fair, none of the errors involved changing the assumptions of the results. It was just that some of the arguments were too loose.
000
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 11/06/2026
Two recent revisions have shown me how much LLMs have improved over the last year- Claude and Chat have both spotted errors in these two papers that neither they nor Refine spotted last year.
282
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 11/05/2026
Yes, that’s one effect that we didn’t anticipate when we wrote down the model. It would be cool if this sometimes actually happens, I wonder what people could do to make that more likely?
030
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 11/05/2026
Andrew Koh just told me about this cool experiment that seems the perfect motivation for our paper: arxiv.org/abs/2601.20245
arxiv.org
How AI Impacts Skill Formation
AI assistance produces significant productivity gains across professional domains, particularly for novice workers. Yet how this assistance affects the development of skills required to effectively su...
020
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 10/05/2026
drive.google.com/file/d/1JK8H...
drive.google.com
Investment_and_Self_control-49.pdf
252
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 26/04/2026
Wow! I’ve been to Stockholm in December 3 times and never seen so much ice
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 25/04/2026
From my experience, winter is actually fine. Provided you have a flexibility to be outside for an hour or two between 10 and 3. Not that cold or windy just not many hours of daylight.
130
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 17/04/2026
edit : I think I’ve also seen them in June? There are no conferences in Paris in August.
110
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 17/04/2026
really? Even in August? I know some French economist with fairly fancy shorts that they wear in urban settings.
100
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 04/04/2026
That’s exactly the issue!
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 04/04/2026
No, I haven’t, but thanks for suggesting it.
140
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 04/04/2026
Two thoughtful essays about the impact of LLMs on graduate education: ergosphere.blog/posts/the-ma... economics.mit.edu/sites/defaul... From personal experience I think the self-control problem mentioned in the first essay is very real.
ergosphere.blog
The machines are fine. I'm worried about us.
On AI agents, grunt work, and the part of science that isn't replaceable.
36621
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 30/03/2026
Andor is even more political and arguably better. (Written by Gilroy not Lucas)
020
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 26/03/2026
Thanks!
000
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 26/03/2026
I like this study because it confirms my prior. 😅 do you know what if anythinghas been done in the way of replication either in this species or other others?
100
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 19/03/2026
I agree, I am deluged with people offering to upgrade my page for a fee and I’d guess that you get a lot more of those offers and unlike me could list them as a business expense.
000
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 25/02/2026
Great choice!
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 23/02/2026
P.s. OK maybe cancel out was an overstatement but I think the headline and post overstated the difference between what people are looking at and what they should be looking at
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 23/02/2026
But MET is based on heart rate, so everything would be fine if people use their watches to compute their MET. The bigger problem is the third one the article points out, which is that smart watch heart rate readings aren’t necessarily that accurate – much better to use a chest strap.
110
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 23/02/2026
It seems to me that this article makes three points, two of which cancel each other out: 1. People talk about their VO2 max when they are really extrapolating it from heart rate data and not getting it tested in a lab with a mask. 2. What has been shown to matter for longevity is actually MET. 1/2
100
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 21/01/2026
Yes, it’s a great explanation. But as Doctorow points out in pluralistic.net/2026/01/01/3..., the EU has an even better response: repeal Article 6 of its copyright directive.
pluralistic.net
Pluralistic: The Post-American Internet (01 Jan 2026) – Pluralistic: Daily links from Cory Doctorow
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 17/01/2026
He makes some excellent points, but the essay would’ve been better if it were half as long
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 18/12/2025
Nice article. But the American preference for personal trucks is driven in large part by their break on CAFE standards.
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 20/10/2025
Really glad right now that overly mostly uses Google cloud
010
drewfudenberg.bsky.social @drewfudenberg.bsky.social · 10/10/2025
Way to go Sally! links.mit.edu/nl3/N5Jy2cKq...
links.mit.edu
Regarding the Compact
070
Reposted by @drewfudenberg.bsky.social
Jason Furman @jasonfurman.bsky.social · 10/10/2025
The other day a student asked me about the prevalence of insider trading in prediction markets. I now have an answer.
8627162