"So on an intellectual level I'm kind of like, you know, well, fuck it, we do the best work that we can. On a kind of emotional level, it's never fun to read criticism about you...."
Says Nate Silver, who made his name being the guy who is right, quoted in Politico in "Nate Silver is happy to be wrong... Blowing Trump’s primary win made me humbler, smarter."
Subscribe to:
Post Comments (Atom)
Posts
23 comments:
Nate Silver seems to be right up there with Lena Dunham in the Althouse pantheon, much to commenter's chagrin.
Nate can survive his Trump primary oopsie. He'll have trouble if Trump -- who Silver only gives a 20% chance -- wins the general.
I like him, generally, and believe in the value of data-driven approach, but intuition and the gut have a role to play too and Silver hasn't figured out how to factor that in.
I'd call BS, except Silver is a genuinely somewhat modest guy who does his best and makes calculated predictions. The prediction percentages may be crap, but he actually tries to make solid predictions off of incomplete data.
He's very proud when it works out, but he also acknowledges mistakes and tries to track their root cause down and improve.
Ultimately, I think it's somewhat of a fools errand - but I can admire he's made a niche out of it for himself.
My take is that as a business venture fivethirtyeight puts Silver in a position which he decries as a statistician: that of making early predictions. It doesn't matter how many times he repeats the fact that early forecasts are weak. Everyone wants a forecast.
No matter how much "science" you wrap around predictions, it is still only a guess.
Like a weather forecast, it's valuable to have a data driven prediction, but you can still get your picnic ruined. It's better than nothing - which is a high standard.
Inasmuch as his "predictions" are simply his desires, in the world of statistics and probability, Silver is all hat and no cattle.
The fact that Silver does make predictions and holds himself accountable when they go awry puts him way ahead of most pundits. I would say most pundits, like Chuck Todd or Chris Matthews are liberals who yearn to run for office one day themselves. They often careen from failed prediction to failed prediction, without assessing their own track record.
At least Silver has a track record and analyzes it.
It’s not Silver’s fault that he was wrong about Trump in the primaries. It’s the polling itself, which seems to predict poorly in any race with Trump in it. Trump does not poll accurately.
Not that I’m unhappy about it. This Trump supporter wants Hillary to poll a little higher than Trump until after the Democrat convention. Once she’s nominated it would be extremely difficult to replace her. We Trump supporters want our man to run against Hillary. We know now, if we didn’t before, that Lynch will NEVER indict Hillary.
The basic problem seems to be the categories they must poll, “registered voters” and “likely voters.” Both categories are of people who either voted in the last election(registered voters) or have voted within the last few years(likely voters).
But a significant amount of Trump voters in the primaries have never voted – ever; they are brand new voters and thus are not counted in the polls.
There may be other factors at work as well. There’s a total MSM negative narrative about Trump supporters that might make a few of them reluctant to declare their Trump preference – especially if they normally vote Democrat. There could be a few disillusioned Bernie supporters who cross over to Trump.
He's going to be really smart after the general election.
Bay Area Guy - I agree with you about pundits and no accountability about their predictions. It is worse than sports pundits. For both groups, tomorrow is another day. Nice gigs.
Silver has usually been right about analyzing the polls themselves, and the only thing he got wrong about the primaries was not a misreading of the polls (which had Trump ahead) but rather a mis-prediction about what would happen as competitors dropped out. That's not really a mistaken analysis so much as a mistaken guess as to future behavior which he couldn't have known either way.
Plus, Silver seems to be far more concerned with getting his information correct rather than the wishful thinking that partisan hacks have indulged in (let's not re-live Karl Rove's 2012 election night meltdown).
Yeah, right? Right?
My take on Silver's error in predicting the Primary:
Silver didn't realize that there were so many angry conservatives out there that were just waiting for a Donald Trump to come along. He and many others failed to accurately read the mood of the disaffected populous who already had latent racisct, bigoted. misogynistic, phobic tendencies. Silver and others failed to see that there were so many of them. Silver and others thought that reason would prevail and even the disaffected would eventually see through Donald Trump, he and others were wrong. Trumpism resonated with far more than anyone could've predicted. In the general, we add different groups of people into the mix.
"He and many others failed to accurately read the mood of the disaffected populous who already had latent racisct, bigoted. misogynistic, phobic tendencies."
-- That analysis is worse than Silver's.
"Silver didn't realize that there were so many angry conservatives out there that were just waiting for a Donald Trump to come along. He and many others failed to accurately read the mood of the disaffected populous who already had latent racisct, bigoted. misogynistic, phobic tendencies. Silver and others failed to see that there were so many of them. Silver and others thought that reason would prevail and even the disaffected would eventually see through Donald Trump, he and others were wrong. Trumpism resonated with far more than anyone could've predicted. In the general, we add different groups of people into the mix."
I don't know if it's so much not realizing so many conservatives were angry, but misjudging the direction their anger would take them. For years, we heard from conservatives that what they didn't like about McCain, Romney et al was that they weren't sufficiently conservative, so you would think they'd flock towards Ted Cruz (or Rand Paul, if they were more libertarian). Trump from the beginning ran against a lot of conservative orthodoxy (abortion, taxes) so a lot of pundits didn't figure those voters would go to him.
Silver also looked at how in past elections the "somewhat conservative" voters usually teamed up with moderates to pick a more moderate choice (i.e., someone they figured could compete better in the general election) and the establishment lined up for that candidate. He figured the same pattern would repeat once the field narrowed down. What he missed was that a lot of those "somewhat conservative" voters were fine with Trump, and the rest of them did not settle on any one candidate. On top of that, few candidates dared attack Trump because they hoped to win over his supporters, so he benefitted from taking the media attention away from the dozen remaining candidates. By the time it was down to him and Cruz (who only the very conservative voters liked) it was inevitable that he'd at least have the delegate plurality.
But there were no polls that said "Trump can't get the nomination", only speculation that some other candidate would add the other voters together enough to beat him. In other words, Silver wasn't wrong in doing what he normally does (analyzing evidence) but rather he was wrong in going out on a limb (assuming something would happen because it usually did).
Conventional wisdom would say that the Republicans picked their worst candidate to be their nominee. Conventional wisdom would say the Democrats have picked their worst candidate to be their nominee. Given that Mr. Silver has a problem: its hard to make predictions, especially about the future.
I wonder what percentage Mr. Silver would have put on that election between the Loyalists and the Patriots.
"Conventional wisdom would say that the Republicans picked their worst candidate to be their nominee. Conventional wisdom would say the Democrats have picked their worst candidate to be their nominee. Given that Mr. Silver has a problem: its hard to make predictions, especially about the future."
Analyzing the past is no picnic either! Had Trump lost the nomination to Cruz, and Cruz was struggling against Hillary, we'd be hearing from Trump fans about how Trump would be way ahead of her by now. There's no way to prove how things might have turned out.
While I think both parties picked terrible candidates (who would also make terrible presidents), I'm obviously not in either party's plurality view. Enough Dems somehow thought Clinton was their ideal standard bearer, and enough Repubs thought that about Trump, and whoever disagreed could not decide on an alternative. So here we are, in what some of us think is a lose-lose election.
Silver is wrong about baseball all the time. He's used to being wrong
Silver has been trying to spin his goof in a positive way since he made it. I understand why he's doing it, but the categorical rejection of what ended up happening is more than just being wrong. It's ridiculous. It's like Bagdad Bob level ridiculous. He didn't just say Trump wouldn't win, he said it in a way which totally dismissed the eventual reality.
His credibility will forever be associated with Trump and Home Alone
If he's a "liberal" (and by that I mean of course "tax-happy, coercion-addicted, power-tripping government sniffer and State fellator") then he hasn't had much practice being right. Modern "liberalism" is the new Gang that Couldn't Shoot Straight; or, The Gang That's Wrong About Nearly Everything.
I don't think there's any positive spin possible on how wrong Silver was on the primary.
On the other hand, a lot of people were very wrong.
You'd have to advance a speculative theory about people lying to pollsters, more or less, in order to have given Trump a favorable chance (early on, at least) of winning the primary.
Nate Silver was, due to a run of good luck *while he already had visibility*, overrated. He had extra credibility he didn't earn. He'd been lucky in the past. His best model put out a ludicrously low probability of Trump winning the primary and he went with it - no caveats. Now, he's properly rated. He should get used to it.
A 20% bet on Trump winning is not a terrible bet. I'd bet Trump wins at those odds.
Post a Comment