Article from a few weeks ago, but now that G. Elliott Morris is taking over without Nate’s models, I’m curious what lemmy’s think about political polling analysis from FiveThirtyEight?
I trust it more, I’d say.
I knew of Nate Silver back when his claim to fame was as a sabermetrician and the creator of a statistical model used to predict how baseball players would perform in the future based on present and prior statistical data. That was PECOTA. I actually liked PECOTA. In the long run I think you’d call it a useful failure. But Nate’s baseball takes were actually very good and quite objective in nature. And he obviously was very good working with statistics.
I got amped up when I learned he was taking his skills into the arena of political analysis. If you remember the early years had a mix of success and failure but was usually good enough to draw onlookers. But something went wrong with all that after a few years – Silver started showing bias in favor of candidates that he had consulting deals with. The objectivity just wasn’t there, he was acting as a paid spokesman would. And the quality of his predictions suffered, as did his demeanor after a while. It was disappointing.
I regard the guy as someone with a deep understanding of political statistics and data who can help paint a very detailed picture, but he displays too much bias to be trusted to remain objective when it matters. It’s kinda like having a defense lawyer. You always know in advance whose side they will take.
Whoever the new guys is, I guess we’ll see whether he will remain a statistician, or follows Silver into trying his hand at becoming an influencer.
I did not trust 538 before. I think a few major US election voter turnouts suffered because too many people thought the 538 results meant that their candidate was a sure thing.
Based on the article, it looks like this change would make that a lot worse, since the main point of contention between Silver and the new guy seems to be that the new guy’s models are a lot more certain of the results too early. So candidates are going to look like the sure thing far more now most likely.
That’s not good. Specifically, I believe Trump beat Clinton because of 538.
That’s not really accurate. The polls were going back and forth for over a month leading up to the election, and the FBI announce they were re-opening the case into Clinton’s emails 3 days before voting day. The polls just didn’t have the time to reflect that change.
Was it 538, or just the utter idiocy of the average American? Just because the polls say he has a 33% chance of winning, doesn’t mean stay home, it means the opposite.
Thems worse odds than Russian roulette.
Weird take.
538 was one of the few groups saying that Trump had a decent chance. They were widely mocked for it. Anybody who believed 538 should have been motivated to get out the vote for Clinton.
And they turned out to be right.
How is that “Trump beat Clinton because of 538”?
Maybe I misremembered 538 for another popular poll at the time.
I think releasing polls to the public is a bad thing for this reason. We ban releasing boat totals before the polls close for this exact same reason.
That’s a problem that could emerge with any system used to predict the outcome of any election.
If you make a prediction, you’re arguably telling people not to vote.
I think it was the elitist confidence that the media including 538 applied to 538.
So confidently predicting an outcome is the problem?
538 is just data-backed fortune-telling.
I didn’t trust Nate Silver anyhow. Maybe it’ll be an improvement?
Why did you not trust him? A history of invalid models, or personality?
My trust in him dropped after being exposed to his activity on twitter. It seemed like he had a habit of trying to apply his knowledge to things he didn’t understand or where he was missing the point. I’m no longer on twitter and can’t remember specific examples, so it may have been more that that format did not work for him.
I remember him having some weird COVID takes. Statistically we should just get out there because only a few of us would die. He seemed to think it was odd that people were staying in for no reason. That completely missed the point that if your family is the statistic, it just fucks your whole life up.
Heh, I remember thinking, statistically, he’s probably not wrong for a population but that completely missed the prisoner’s dilemma of making sure you’re not the statistic.
Agree! This describes my experience exactly. On a related note about format, I thought Jody Avirgan, Five Thirty Eight’s podcast producer early on, was a good foil for him, drawing him into conversation and occasionally pushing back on points that sounded wild or tone deaf.
The old 538 pod was great but after they were bought up and the other regulars were fired it went downhill. Nate is much better in that medium than on Twitter or Meet the Press or whatever.
yeah, outside of the stuff i brought up they’ve also just churned a lot of their best and most visible reporters to other places and opportunities. it’s unfortunate but that’s a natural part of journalism, especially with the whole field being pretty sparse with opportunity to climb the ladder.
As soon as Clare was laid off I no longer was interested in 538 (atleast the podcast end). She had the most interesting perspective in my opinion. As far as the models go, most of the ones out there all seem similarly accurate so it is what it is.
Where did she go?
On the models, do you have any sources talking about the models and their accuracy?
No I do not 😂. I’m sure there’s something out there but I’m thinking about it from the perspective that if anyone had been way off recently there would have been a big hubbub about it. After Trump got elected most of these analysts changed their models and they’ve been more accurate since.
Clare writes for a couple different sites now, The New Yorker and NY Magazine… Basically free lancing. I believe she’s also working on a book.
Article from a few weeks ago, but now that G. Elliott Morris is taking over without Nate’s models, I’m curious what lemmy’s think about political polling analysis from FiveThirtyEight?
GEM is generally fine from what i’ve seen; it’s also hard to feel sympathetic for Nate even though 538 is his baby. a lot of his colleagues don’t like him (because he apparently presided over a very bro-ey, douche-y culture) and he’s kind of a dipshit disaffected libertarian who loves to speak out of his ass on things he doesn’t understand via Twitter.
I don’t buy the libertarian thing. I used to listen to his political podcast on the regular, it was pretty clear that he was no fan of right wing politicians.
it’s possible he dislikes many current right-wing politicians, but he explicitly described himself in 2012 with the following quote and hasn’t really shown much indications he’s deviated far from this belief: “I’d say I’m somewhere in between being a libertarian and a liberal. So if I were to vote, it would be kind of a Gary Johnson versus Mitt Romney decision, I suppose.”
I think he has kind of an inherent distrust of politicians/politics in general so that does kind of track. I’m not 100% sure his personal beliefs but he has never really expressed a progressive viewpoint that’s for sure (and not saying it’s a bad thing).
Honestly, no. Well, not immediately. If the new guy can prove himself then sure, but we are well past the point of just trusting anyone’s political reporting without a strong history of it.
Wait and see.
I’d agree with the history of valid reporting before trust is established.
For me one of the reason this is challenging is there is very little out that seems to be good political reporting, at least on polling predictions.
What is he, a religious leader now?
He was a good modeler and handicapper. His model happened to work well for the 2016 election. That’s it. He’s not a fucking oracle.
Nate Silver was a big figure for his time, but I think that many folks are viewing polling with a deserved skepticism at the same time certain organizations come to understand that “manufactured” polls can sell the narrative they need to survive and maintain the status quo.
His departure is one of mixed feelings for me, on one hand I felt like he was at least principled. I don’t exactly know the predecessor, but I imagine the network wanted more flexibility in crafting a narrative than doing real forecasting. On the OTHER hand, his name was starting to get attached to everything and I think it led to an outsize ego that ultimately was going to slowly poison things. IMO - it was only a matter of time before we saw someone new. It’s just that the context here gives me pause as to if the replacement is better or worse.
I think I’ll have to wait and see…
Looking at he the feud, I’d say this new guy is going to be wrong more often simply because he argues for certainty early on. And that’s going to just make people hate polls and hate math that much more. Nate wasn’t perfect but he was better than most.
I will wait and see. I treat all polls as biased to a degree, and just historical snapshots.
Morris has also built robust polling analysis models, so it will be interesting to see how (or if) his models manifest themselves on 538.
As to Nate, he can well and truly go fuck himself. His arrogance is staggering, as demonstrated by his behavior at the onset of COVID in early 2020. He seemed to argue with every epidemiologist that crossed his path early in the pandemic, simply because he believed HE knew better about the impacts of a global pandemic than they did. Because as we all know, being a statistical analysis guru makes you an expert in epidemiology (yes, I’m being sarcastic).
The dude is so far up his own ass that the only way he can see outside is through his belly button.
Sounds like he thinks he’s a Hari Seldon figure. Someone should have told him Foundation was science fiction.
I have 538 bookmarked but I never find it all that helpful. By the time the charts start to make sense to me, it’s too late. Maybe I don’t give it enough of a chance/credit but that is how it seems to me.
So I guess I am saying, I didn’t really trust it before.
I didn’t trust it WITH him!
