My humble contribution to the inflation discourse:

@leopd @dbaek__ Have you seen It seems like it's down to a very fiddly hyperparameter.

@OfSymbols @JakeMGrumbach I wouldn't use "tricked" but yeah, I would say that people's perception of the economy is not very correlated to our best objective metrics. Open to changing my mind on this point if someone can present compelling data that's actually correlated in a non-trivial way.

@OfSymbols @JakeMGrumbach I do not think there is any evidence of that at all in the data. I think you are trying to find a way to shoehorn your preferences (people are unhappy because things are so bad and if dems were progressive economically they would reward them) into the data.

@OfSymbols @JakeMGrumbach Even if you think that's a more accurate measure - that measure does not correlate to public sentiment about the economy at all. Why were people happier about the economy when that particular measure was even worse?

RT @PatrickKidger: 🚀 It's time for a new JAX ecosystem library! This time quite a small one: ESM2quinox. A #JAX + Equinox implementation o…

@teortaxesTex On the one hand, it's a useful public service to correct these influence grifters. On the other hand, it's super useful to watch them try to sound competent on things they don't understand immediately label them as grifters.

@RiccardoTrezzi @jasonfurman The part about construction costs is fascinating, but, I'm still trying to wrap my head around how come the conclusion about wage growth at the bottom: - is this an issue of using different references? The absolute numbers look fairly similar.

@MaxJordan_N @Eazy_Bake Sure, as I said, if you want to vote to "punish" your coalition for insufficiently catering to your interests, you are free to do so, the expected consequences were fairly obvious though!

@typedfemale having to auto complete your posting is far harsher punishment, heh

@MaxJordan_N @Eazy_Bake Also, among Biden 2020 voters who did not vote for Harris, Gaza was their #1 issue. Can you get me a source for that? Are you excluding people who switched their vote for Trump?

RT @hmmlowe: The contact hypothesis is a leading theory in social psychology--the idea is that interpersonal contact between groups, under…

@MaxJordan_N @Eazy_Bake The argument that when the choice is between Harris & Trump, Harris isn't sufficiently progressive so you might as well sit it out or vote Trump makes zero sense to me if evaluated under any kind of utilitarian framework.

@littmath You are going to get so many people posting o3 output in your replies, oof

@JStein_WaPo @jasonfurman This is quite different from @arindube's numbers. I wonder what is the different in methodology that produces those differences - or is it just about window? The trend (highest growth in lowest quantile) is consistent though.

@leopd @bremen79 IE: Adagrad/Adam/RMSprop, etc all make different assumptions to estimate curvature and clearly some of those assumptions work better on some kind of models. Can we use that to characterize the properties of the loss landscapes more formally?

@salonium Yes, but, given raw counts, you can use to estimate the uncertainty in the binomial parameter, no?

@TimSweeneyEpic @SohrabAhmari I do not think debanking was actually a thing that happened, or not in the way it was claimed it did @patio11 explained this more at length: & (apologies for the length).