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

[dunceDude]

One of the worst threads I've ever read. Everyone is answering the easier question of "would you like to dunk on Nate Silver" instead of the real question of whether 538's models hold up mathematically—which they totally do. You can see their defense here, but it boils down to the fact that through their history things happen roughly the % of time they predict (i.e. of all the things they've given a 30% chance of happening, roughly 3 out of 10 actually happened).



To the extent that events do *not* occur the percentage of time that 538 predicts, it's because they are being more cautious in their language (per chart, the things they say are likely actually happen a bit more than they predict, while the things they say are unlikely happen more rarely than they say).

That data journalism gets sh**t on so much is really mystifying to me, though I understand it gets you internet points. Their content is better than 85% of what I see from journalists (even after accounting for the fact that I only follow quality journalists).

[kyc0705]

Everyone is answering the easier question of "would you like to dunk on Nate Silver" instead of the real question of whether 538's models hold up mathematically

I think this is weirdly a part of it. He's so standoffish that nobody wants to be on his side. And then he wonders why nobody wants to cut him any slack. And I say this as a guy who's been reading the 538 live-blog religiously this week. I get what he's saying but I'm not shocked that he's having a hard time communicating his argument to people—and it's not only because the average person doesn't understand statistics or probability.

Frankly, I unfollowed Silver on Twitter ages ago just because he can be so unpleasant when someone disagrees with him (his feud with G. Elliott Morris showed that he is substantially less emotionally mature than someone roughly half his age).

[TiltsAreUnderrated]

Everyone is answering the easier question of "would you like to dunk on Nate Silver" instead of the real question of whether 538's models hold up mathematically

I think this is weirdly a part of it. He's so standoffish that nobody wants to be on his side. And then he wonders why nobody wants to cut him any slack. And I say this as a guy who's been reading the 538 live-blog religiously this week. I get what he's saying but I'm not shocked that he's having a hard time communicating his argument to people—and it's not only because the average person doesn't understand statistics or probability.

Frankly, I unfollowed Silver on Twitter ages ago just because he can be so unpleasant when someone disagrees with him (his feud with G. Elliott Morris showed that he is substantially less emotionally mature than someone roughly half his age).

Morris was even worse in the end, though. He put way too much emphasis on margins and trashed all Emerson polls, even the ones that didn't use MTurk, just because he was angry that some did.

[kyc0705]

Everyone is answering the easier question of "would you like to dunk on Nate Silver" instead of the real question of whether 538's models hold up mathematically

I think this is weirdly a part of it. He's so standoffish that nobody wants to be on his side. And then he wonders why nobody wants to cut him any slack. And I say this as a guy who's been reading the 538 live-blog religiously this week. I get what he's saying but I'm not shocked that he's having a hard time communicating his argument to people—and it's not only because the average person doesn't understand statistics or probability.

Frankly, I unfollowed Silver on Twitter ages ago just because he can be so unpleasant when someone disagrees with him (his feud with G. Elliott Morris showed that he is substantially less emotionally mature than someone roughly half his age).

Morris was even worse in the end, though. He put way too much emphasis on margins and trashed all Emerson polls, even the ones that didn't use MTurk, just because he was angry that some did.

Yes, but I was talking less about the accuracy of their forecasts and just more about the way Silver talked to/about Morris—it was just unpleasant and petty for absolutely no reason at all.

[Ferguson97]

4 years and people still don't understand probability.

Pollsters ed up. Nate is not a pollster. His model was largely accurate. He got what, two states wrong that everyone agreed were tossups?

[MARGINS6729]

Keep dreaming, the polling industry is dead.

[lfromnj]

4 years and people still don't understand probability.

Pollsters ed up. Nate is not a pollster. His model was largely accurate. He got what, two states wrong that everyone agreed were tossups?

>Junk In
>Junk Out

tell me when he actually regrades the pollsters based on not only methods but also their end results.


Marist was still junk in 2018 and they still have an A grade. There is 0 excuse to let them keep it even if somehow their methods are ok.

[Statilius the Epicurean]

I think this is weirdly a part of it. He's so standoffish that nobody wants to be on his side. And then he wonders why nobody wants to cut him any slack. And I say this as a guy who's been reading the 538 live-blog religiously this week. I get what he's saying but I'm not shocked that he's having a hard time communicating his argument to people—and it's not only because the average person doesn't understand statistics or probability.

Is that a bad thing? Honestly I like Nate's no-bullsh**t personality and that he's not a partisan hack. 538 integrates uncertainty into their forecasting in a way very few models or pundits do.

I'm sort of confused about the criticism they're getting because the entire reason Biden was favoured 90% was because he could withstand a massive polling error in key swing states and still win...which is what happened.

[BlueSwan]

I'll never ever understand the hate that Nate Silver gets on this forum.

Models SHOULD be primarily focused on handling polling data. All that other sh*t is pretty much BS.

But in handling polling data you have to rely on the actual data that you get. Silvers model usually fares better than other models who project more confidence than Silvers model.

[S019ian Liberal]

No. F---- him.

[S019ian Liberal]

I'll never ever understand the hate that Nate Silver gets on this forum.

Models SHOULD be primarily focused on handling polling data. All that other sh*t is pretty much BS.

But in handling polling data you have to rely on the actual data that you get. Silvers model usually fares better than other models who project more confidence than Silvers model.

This is not how it works. Of course it's easy to add uncertainty -- for instance if I believe Biden to be the favorite I can just give him 51% chance throughout the campaign and leave it there. Then afterward if Trump wins I will say "but hey look, I had him at 49% shot" or if Biden prevails then "I called the right winner correctly" when in fact I was no more accurate or spot-on than anyone else here who gave Biden a 99% to win instead of course as a matter of fact wisely so.

[S019ian Liberal]

Trafalgar was junk guys, did you not see their final map? They had Trump winning all kinds of states he is not going to win. Them getting Florida right on the broken clock principle does not make them a good pollster. IDK why they're getting a pass for being wrong on AZ, Georgia, Michigan, Penn exc while other polls are getting killed for just getting Florida wrong.

They were less wrong in those states than the other pollster were. Case in point ABC WaPo Biden +17 in WI. Never did Trafalgar have Trump +14 there which would actually be less badly so and off than QPac and others were, for instance.

[S019ian Liberal]

Absolutely he did. Many said he was "wrong" when he said Trump would lose in 2016, despite giving him a 30% chance to win. By that logic, since Biden won, he was absolutely correct to give him a 90% chance to win.

The snake states that were wrong had >30% chance of going the other way but it doesn't matter. If you judge based on the result, as many did last time, then he called the election exactly correct!

There's a 100% chance you understand 0% about probabilities.

[𝕭𝖆𝖕𝖙𝖎𝖘𝖙𝖆 𝕸𝖎𝖓𝖔𝖑𝖆]

Am I the only person here who is old enough to remember when Atlas was giving sh**t to Nate Silver because his 2020 model was not bullish enough on Joe Biden?

[S019ian Liberal]

Nobody got this election right.

I did!

This happens in baseball analytics, where Nate started out, allot too. ‘The thing you said had a 30% chance of happening happened, where’s your math now nerd?’

Nate Silver is frankly just a simply obnoxious idiotic snobbish asshole.

[S019ian Liberal]

Nobody has said what you think they said. What a weak useless and utterly deranged defense of an entirely discredited pollster!

[DabbingSanta]

4 years and people still don't understand probability.

Pollsters ed up. Nate is not a pollster. His model was largely accurate. He got what, two states wrong that everyone agreed were tossups?

The margins were *really* wrong, though, often by 5 to 10 points. This counts too. Nate has consistently overestimated Democrat support in the last four election cycles (2014-2020). Sure, the polls have also consistently overestimated Democrat support, but you'd think after the third time you would take these biases into account when making the "perfect model".

[pbrower2a]

Republicans so far are masters of killer ads for their opponents.

The Democrat will take away your guns.

The Democrat will bring socialism.

The Democrat will weaken the police so that crime will run rampant.

So that you can be sure of having a job, you had better support the one who will cut taxes for the 'job creators' -- and don't you dare ask them to say what they will do with their tax cuts.

Trump and the GOP did those late, and they may have worked to defeat some Democratic nominees even if they did not win the election for Trump. The Master Class demands everything, and if it doesn't get everything it will seek to make everyone else pay a steep price.

Trump was about as pure a negative campaigner as there ever was. He offered no solutions other than to make the super-rich even richer and to make people too stupid or scared to see anything wrong with him. Such is his personality.

[emailking]

Absolutely he did. Many said he was "wrong" when he said Trump would lose in 2016, despite giving him a 30% chance to win. By that logic, since Biden won, he was absolutely correct to give him a 90% chance to win.

The snake states that were wrong had >30% chance of going the other way but it doesn't matter. If you judge based on the result, as many did last time, then he called the election exactly correct!

There's a 100% chance you understand 0% about probabilities.

I use both probability and stats daily. I've said many times Silver's model cannot be either right or wrong. But many decide to judge the model as either right or wrong anyway. If it got 2016 wrong (it didn't), then it got 2020 right (it didn't). Happy?

[lfromnj]

I'll never ever understand the hate that Nate Silver gets on this forum.

Models SHOULD be primarily focused on handling polling data. All that other sh*t is pretty much BS.

But in handling polling data you have to rely on the actual data that you get. Silvers model usually fares better than other models who project more confidence than Silvers model.

Except my problem is that he accepts garbage data as good and refuses to change his pollster grades. Why did Marist still keep an A rating after missing 4 major races in 2018 by their MOE?

[Alben Barkley]

Pretty sure his "snake" came out accurately, which is pretty impressive given how inaccurate the polls he was working from turned out to be.

Yup, the snake got 48/50 states “right” in the sense that who it favored to win, won.

[Fuzzy Bear]

To be fair, it's hard to build an accurate model off polling if the polling is inaccurate. Again, Silver had the best model, giving Trump much better odds to perform as well as he did than the Morris' model. On a separate note, it always felt like Morris let his partisanship affect his modeling decisions. He excluded what turned out to be valid pollsters like Trafalgar and Emerson on the grounds of having unreliable data. I feel like this had to do with them giving good results for Trump. This was defensible, but I feel like he would not have excluded junky polls if they were giving good results for Democrats.

Trafalgar wasn't a valid pollster.

People need to stop saying that all the Republican hack pollsters were good just because they were closer to the actual result.

I could make a map where I just give Trump every single state by 5 points, and then point to FL and say "look I had Trump winning FL by 5 and he won it by 3!  I beat all the fake news pollsters!"  or "look I had Trump winning WI by 5, and he only lost it by 1, that's way better than the fake news pollsters who had Biden +9!"

The fact of the matter is that the mainstream pollsters were wrong, but they were mostly wrong directionally.  Trump got way better turnout than expected so you basically shift every poll 4-5 points in Trump's favor and you're close to the real results.  That's why Nate's snake map ended up being just about right, with GA the only outlier.

On the other hand, Trafalgar was just predicting a Trump win in every swing state based on nothing other than quantified "gut instinct" about "shy Trump voters."  So you had totally wrong numbers like Trump +2 in MI, +2 in PA, +1 in NV, +5 in GA, +3 in AZ, and so on.  Those are wrong numbers, and there's not any rhyme or reason to their wrong-ness.  They're just random Trump-friendly numbers that Trafalgar got not through any scientific methodology, but rather by starting with the final result they wanted, doing a poll, and then fabricating priors to massage the results until they got to their desired outcome.

Why does this matter?  Because if all polls were like Trafalgar, polling would be worse than useless.  It means all Trafalgar basically is is one guy's gut instinct translated into numbers.  It's no more useful than an SN2903 map.

Trafalgar does count; they did the best job, as does Rasmussen.  It just bugs people that Robert Cahaly, an eccentric conservative with a bow tie and unflattering mustache and hairstyle, is building a track record of accuracy in Presidential elections. 

As the media was completely in the tank for Biden this year, why should one be surprised to think that the pollsters were complicit as activists posing as observers?

[Sbane]

Nobody got this election right.

King Sabato? (Aside from NC. And he screwed up on a bunch of House and Senate races)

NC will be pretty damn close as well once all the votes are counted. Florida is the state that surprised me the most and those who stuck with Trump winning Florida should get full credit.

[Sorenroy]

Nobody got this election right.

I did!

I've seen you post this before, but I'm not seeing where this correct prediction is. Your official prediction (the one below your avatar when you post) had Biden winning 350-188. You've already missed Florida and there's a good chance you miss North Carolina as well (not even getting into percentages). If you made something more recent, I don't see it, and I scrolled back before that prediction just in case you posted it in a thread somewhere and forgot to update the official one.

[South Dakota Democrat]

Silver SN2903 is an absolute joke