Nate Silver: "**** you, we did a good job"

[Asta]

I was an applied math/stats major in college so it may come across biased, but people need to stop chastising Nate Silver. Polling is imperfect due to psychological elements.

There is nearly no one who got every election correct from 2000 to 2020, particularly close to the margins. I only got Georgia wrong this year yet I don't claim to be a better election pundit than him.

Unlike 99% of people who, with their rose-colored glasses, are myopic and self-absorbed in their beliefs that their side will win, at least he is able to provide probabilities buttressed by concrete empirical reasoning.

Not everyone can claim to prognosticate increasing probability of EC/PV split in 2016 or that Obama would have had the edge in 2012 if PV was close like he can.

Deductive reasoning is just as important as correct answers themselves. I would not bet my mortgage on Nate Silver or anyone else for forecasting elections but I sure would rather read his reasoning than "muh 2016".

[Asta]

Right. And that does excuse a fundamentally disoriented, piss-poorly calibrated model.

So what component, if any, of his model, do you disagree with? Don't just say something like "everything". I'm not claiming his models are perfect or imperfect. If you're going to provide a contrarian view, then please specify your breakdown backed by the data.

Good for you. And you might very well be a better pundit than him indeed. Not that that's a terribly high bar to attain, of course.

I sure am not. My basis was that polls did not seem to have adjusted errors in the Midwest regions in 2018. My read on Florida early voting was that Florida is likely gone for Biden. I don't deserve any praise for it.

Very flawed empirical reasoning at that, mind you.

So back to the above, what about it is flawed? Response bias? Shy Trump voters? If that's the case, then that would be a problem with the pollsters, not Silver's models. His job is to aggregate and apply them. If you're not going to back up your models using long-term evidence, then your punditry is as good as bunk science. There is no long-term bias toward one party over another. If polls are going to systematically oversample Democrats, then it wouldn't be Silver's duty to artificially add 10 points to Republicans based on just recent happenings.

I am not going to praise a pundit who gets correct answers at the expense of abandoning scientific reasons; not that it would have a durable effect in the long term anyway.

I would diss Silver if he adds 10 artificial points to Republicans then end up overestimating Republicans in the next election, even if it ends up more accurate than 2020.

Of course, to someone like you who is ready to abandon the logic train, you would likely say I might as well use such methods. Sorry to you mate, but I value principles over short-term predicting abilities.

State vs national polling suggested, if not indicated this. How do you think Silver separated his noise from the signal anyway? People just fail to pat attention individually.

Yea in hindsight, it's easy to claim to have observed this. At the time, nearly no one was talking about this.

who exactly has been saying that? cuz i know that i currently haven't, that's for sure!

Numerous doomers, including myself expected Trump to outperform his expectation, not necessarily based on 2016.

[Asta]

Right. And that does excuse a fundamentally disoriented, piss-poorly calibrated model.

So what component, if any, of his model, do you disagree with? Don't just say something like "everything". I'm not claiming his models are perfect or imperfect. If you're going to provide a contrarian view, then please specify your breakdown backed by the data.

Good for you. And you might very well be a better pundit than him indeed. Not that that's a terribly high bar to attain, of course.

I sure am not. My basis was that polls did not seem to have adjusted errors in the Midwest regions in 2018. My read on Florida early voting was that Florida is likely gone for Biden. I don't deserve any praise for it.

Very flawed empirical reasoning at that, mind you.

So back to the above, what about it is flawed? Response bias? Shy Trump voters? If that's the case, then that would be a problem with the pollsters, not Silver's models. His job is to aggregate and apply them. If you're not going to back up your models using long-term evidence, then your punditry is as good as bunk science. There is no long-term bias toward one party over another. If polls are going to systematically oversample Democrats, then it wouldn't be Silver's duty to artificially add 10 points to Republicans based on just recent happenings.

I am not going to praise a pundit who gets correct answers at the expense of abandoning scientific reasons; not that it would have a durable effect in the long term anyway.

I would diss Silver if he adds 10 artificial points to Republicans then end up overestimating Republicans in the next election, even if it ends up more accurate than 2020.

Of course, to someone like you who is ready to abandon the logic train, you would likely say I might as well use such methods. Sorry to you mate, but I value principles over short-term predicting abilities.

State vs national polling suggested, if not indicated this. How do you think Silver separated his noise from the signal anyway? People just fail to pat attention individually.

Yea in hindsight, it's easy to claim to have observed this. At the time, nearly no one was talking about this.

who exactly has been saying that? cuz i know that i currently haven't, that's for sure!

Numerous doomers, including myself expected Trump to outperform his expectation, not necessarily based on 2016.

absolutely and complete a total lie. i have been the only one defending the polls when others were all in favor if skewing them by artificially inflating the number of disadvantaged party's margin and then repeatedly getting burned for it each time. including trump and took so much flak for it.

for you to talk about having "principles" or taking a ride of the "logic train" is frankly insulting and very rich coming from a person like analyzing this. really? this is wrong and you should know better than to peddle such nonsense. very offended by your untrue and blatantly false attacks, wherein you keep getting things wrong but blame it on others, and then project your long-term track record of failure onto "luck" where it can't even explain those same errors!

I apologize if that came across insulting. I thought that you're mocking Silver, are you not?

You saying I might as well be a better pundit gave me the impression results are the only things that matter. I did not read all 4 pages. You still haven't explained this. I'm frankly perplexed by your seemingly willingness to attack his methodology.

Btw, keeping all NJ buddies was a self-trolling job that I intentionally did to amuse myself after seeing consecutive NJ voters in one thread asking about if they voted. I take no offense to it so keep your signature at your own discretion.