drewfudenberg.bsky.social
60 posts
824 followers
153 following
Getting Started
Active Commenter
comment in response to
post
Your description already seems pretty concise as it is
comment in response to
post
"Inada: It's not a condition, it's a way life"
comment in response to
post
Those who aren’t already familiar with Philipp’s work, and would like to learn more without reading the papers, might be interested in a survey/summary of it that I wrote for the JEP: economics.mit.edu/sites/defaul...
comment in response to
post
The linear algebra class I had was completely devoid of applications. Two of my sons had different first classes at different schools and they too were very dry and abstract. I would’ve liked it a lot more if it had been linked to some applications.
comment in response to
post
Nice. Note that when you take finite approximations of A. Here 0 is a best response to a full support belief, but the required weight on 0 converges to 1 as the smallest positive grid point goes to zero. So the full support beliefs that rationalize 0 converge to the point mass.
comment in response to
post
Discarding sounds wasteful. What about relaxing? It worked OK for some of my clothes :-)
comment in response to
post
My to do list is probably much smaller than yours so I’m able to remember it -at least most of the time. The rare time something slips can be painful, but so far I have resisted this logical step.
Maybe next time I mess up, I will at least think about implementing it .
comment in response to
post
It also wants me to replace $
With \,
Is that really the new standard and if so, why?
comment in response to
post
For this reason, I’m worried about the trend to impose page limits on online appendices as well as the main text.
comment in response to
post
in addition to the worry that long proofs are hard to absorb, there’s also the problem that clear and reader-friendly detailed proofs -e.g. with both the high-level summary and then details proved in lemmas-can be hard to reconcile with journals’ page limits.
comment in response to
post
for what it’s worth I think you regained a lot of dignity by having the courage to warn everyone about what happened. I had not heard of this type of scam before.
comment in response to
post
What to make of this? To me, the take-away is we need a much better understanding how the information environment maps onto people's mental representation of the decision problem.
Attention/memory has a huge role to play, as they will interact with context to determine this representation.
fin.
comment in response to
post
I agree with all of this, but I probably would not have been able to say it as well. :-)
comment in response to
post
It’s a super interesting experiment, but I think it is much too early to draw any sweeping conclusions as it’s not the case that behavior in these deterministic mirrors is always the same as in the probabilistic version.
comment in response to
post
I've watched that movie but had forgotten this scene.
One moral is that deductions based on common knowledge of payoffs are fragile.
comment in response to
post
Here's another game theory related clip: youtu.be/lErlHLCNM_s?...
comment in response to
post
This one is with @bartlipman.bsky.social and Aldo rustichini.
comment in response to
post
Here's another Dekel paper I enjoy teaching: www.jstor.org/stable/29985...
comment in response to
post
But now that we know you could do it, you can stop, right? 😁
comment in response to
post
It’s a cool paper and I teach it. But I don’t think it’s that descriptive, and I view it mostly as a cautionary tale about iteratedweak dominance arguments.
comment in response to
post
And I’d feel the same way as a speaker. Do you have the same preference on this as a speaker or audience member?
comment in response to
post
For me it depends on the number of people in the audience and the norm about asking questions. In a big room with no questions I’d prefer shorter talks but not at e.g. Cowles.
comment in response to
post
I only recently learned about the conformal inference literature, I think some other economic theorists might not yet know about it and find it interesting.
comment in response to
post
can see this as extending conformal prediction from exchangeable observations to exchangeable domains.
economics.mit.edu/sites/defaul...
comment in response to
post
bounds on the error when transferring an estimated model to a different domain, e.g. on eliciting certainty equivalents for the same lotteries from participants in different countries, or eliciting certainly equivalents for different sets of lotteries from the same subject pool.
comment in response to
post
h/t @aaroth.bsky.social
comment in response to
post
Agreed! There are some people who I’d like to hear from sometime, but who in my opinion over post
comment in response to
post
Thanks for the update! my knowledge of this was fairly old, so I’m not surprised that it’s obsolete.
comment in response to
post
is it still true that excellent chess players using chess programs do better than either the best chess players or the best programs? I think that was the outcome of some tournaments a while ago.
comment in response to
post
How about a one page extended abstract?
comment in response to
post
I though Maestro was ok, but I really liked Tar.
comment in response to
post
A common state space seems needed for common knowledge, but people could end up with common beliefs about the probability distribution of a publicly observed outcome without having the same state space or indeed without even being Bayesian.
comment in response to
post
Yes it’s easy to formalize. But I thought it was worth reminding people that an additional assumption is needed.
comment in response to
post
This doesn’t quite follow from Savage alone. A dynamic consistency condition to conclude that behavior after the initial period maximizes SEU with the updated beliefs
comment in response to
post
I don’t think there’s a strong case for a common prior in the “original position” that aumann advocated. I think the question is when learning from a common database will lead people with different priors to have a common posterior.
comment in response to
post
But while that’s what I’m used to I think I’d be able to adapt to any consistently use scheme
comment in response to
post
But in repeated games (or more generally multi stage games with observed actions) we used h_{t} to be the commonly known history of play up till time t. This is a sufficient statistic for all of the individual players’ information sets h_{i,t} at that point.
comment in response to
post
In general extensive form games, the terminal node describes the path of play. I think KW use z for this, I know FT did.
comment in response to
post
Almost common knowledge is enough but I don’t think it frequently holds.
comment in response to
post
What sorts of robustness are you alluding to?
comment in response to
post
My work with Levine used the fact that If a set of payoffs is self-generating and compact, its elements are all PPE payoffs. We did not characterize the exact equilibrium set for any fixed discount factor.
comment in response to
post
The B.E. Journal of theoretical economics used to ask referees to assign papers to one of four tiers, and then publish the paper along with its rating. I haven’t been following them, but just now I checked their website and I see they have switched to a two tier system.
comment in response to
post
interesting. How many reviews would you envision your centralized system asking for ?
and how broad set of journals , both in terms of topic and perceived quality?
comment in response to
post
Agreed that this is elegant and cool, but I’m not a fan of the simplifying assumption that the error in distinguishing two candidates doesn’t depend on how close they are. I
The basic insight seems more general, I wonder how much it can be extended?
comment in response to
post
The current gain may indeed be small. But the cost would seem to be small as well.