Trump didn't get any more votes than 2020 so you were right there. Harris got about 14MM fewer than Biden. What I struggled was the misogyny and racism in the country.
I have to go with Nate Silver's model, forecasting 53.8% for Trump (though just 25.7% for popular vote). I perceive you as having a greater pro-blue bias than Nate, and of course, a dramatically less detailed model.
You're basically reasoning about what the electorate "really" thinks, but I would guess that if your model is halfway-decent at understanding the electoral college, and if you shift it by roughly 1% of Americans (in swing states), then the probability ends up looking pretty much like Nate's.
And I notice that what you haven't done here is explain why ~47% of voters *did* vote for Trump in the past. I think if you don't have a solid, visceral understanding of why people vote for Trump (or, for that matter, against Harris or Hillary), then you'll probably get the vibes wrong and perform worse than a quantitative model like Nate's.
To me, the worst thing about Trump is that he lies and bullshits constantly, with his criminal behavior and total immorality a close second. The thing is, most voters (who have ever considered voting for him) have either never really believed that he lies and bullshits constantly, or, at least, distrust the "establishment" so deeply that it seems like Trump/Harris are almost equally bad. The reason Trump was so popular right off the gate in 2016, I believe, is that most of his lies were already popular memes on Facebook. He could simply take whatever nonsense he saw on Facebook or Twitter at face value and repeat it as gospel truth, and this made him seem very trustworthy to conservatives because that's what most conservatives were already doing. He outperformed other conservatives because he wasn't in one of the usual high-brow conservative clubs, he was in a low-brow "Facebook conservative" club like most of the primary voters were.
Even if, for example, voters understand that his claims of innocence are self-serving and therefore give those little weight, once Trump became nominee in 2016, a huge entire media ecosystem pivoted to convincing people that whatever Trump said was true ― no matter what it was. I noticed this starting with the "big groups of Muslims in New Jersey really did cheer 9/11" of 2016, and I assume this has continued with "the election was stolen" in 2020 and "trials against me are a witch hunt" of 2024. I haven't paid much attention to that ecosystem, but I noticed it a little after "they's eating the dogs, they're eating the cats, they're eating the pets of the people that live there". Trump doubled down, so conservative media either doubled down too, or looked the other way.
Hell, do you ever see people bother to mention that Trump's favorite phrase "fake news" was originally the term used to describe the Macedonians writing completely fake pro-Trump anti-Hillary "news articles" to earn ad revenue (in an article that IIRC was published on election day 2016 or the day after)? It seemed to me like Trump had won control of that phrase within two weeks, just by nakedly accusing the mainstream media of "fake news" with zero evidence. This reflects a broader strategy of people like Trump: take whatever you're accused of and throw it at your opponents. Suppose Trump kills a Democrat by appointing a series of pro-Trump generals, colonels and sergeants and giving the order―an act for which he'd be immune from criminal prosecution, as you know. I don't really think he'd do it, but if he wanted to, he might start Putin-style, by seeding stories about all kinds of imaginary plots by Democrats to kill Republicans. If by chance an important Republican should die of natural causes, Trump would immediately launch an investigation into the "foul play". By the time he assassinates the Democrat, then, it all seems like a wash, with Trump looking no worse than those "killer Democrats".
To be clear, the fundamental problem here is epistemology: most people's is terrible. Even on LessWrong, when I proposed a grassroots web site for hashing out the truth based on evidence[1], it got a cool reception. One commenter repeated Russian propaganda; another disagreed that truthseeking was difficult, saying 'To be clear, I'm saying that [true claims inherently "sound better" than false claims]': my proposal was unnecessary because this person believed they could tell whether a claim was true just by _reading it_.
And on the flip side, Matt Yglesias and Nate Silver have done a good job of highlighting the weaknesses of liberals and the Harris campaign relative to conservatives, e.g. conservative media tends to have the pragmatism not to talk about their pro-billionaire anti-health-care anti-abortion stances, while Harris hasn't really explained how her views apparently changed since 2019 (so the changes may seem insincere) and liberal groups will often blast Biden/Harris for having views they don't like, even when Trump is even worse. (And why wasn't Josh Shapiro nominated as VP?) Scott Alexander, too, recently endorsed Harris but highlighted things he *really* didn't like about liberals and "woke" Democrats, and it's pretty obvious that those sentiments aren't unusual in the general public. So given all this, a Trump victory would not be surprising.
Also, something I haven't heard Nate talk about is what happens if the election is very close in the swing states. What may happen in that case is that something important goes before the supreme court ― the same court with three Trump appointees that awarded Trump not only "absolute immunity" from criminal prosecution for "official" acts, but also an evidentiary shield that prevents "official" presidential activities from being used as evidence of a non-presidential crime.
Balancing that, Nate's latest projection is that Trump is down to 51.5%. He notes in his latest articles, though, that there's a lot of "herding" going on, and several things that "are hard to reconcile in the polls"[1]:
> - Why likely voter models help Harris in national polls but hurt her in the swing states.
> - Why some high-quality pollsters like Selzer and NYT/Siena, who are notorious for not herding, are so different than the others.
> - Why demographic shifts are extremely pronounced in the crosstabs, especially Harris losing ground with Black and Latino voters, but produce relatively small swings in the toplines.
> - Why so few polls in states like Wisconsin — less than 10 percent! — dare to show anything outside the range of Harris +2 to Trump +2, even though this is incredibly statistically unlikely.
> - Why there’s a palpable gap between Harris and Trump’s favorability ratings, but she doesn’t consistently lead him in the head-to-heads. There are some other good explanations for this one, but it could reflect herding — pollsters might not bother to tinker as much with their favorability ratings as with their head-to-heads.
> - And why national polls have swung quite a bit toward Trump over the past several weeks but battleground state polls have moved only slightly.
The effects of post-Covid sticker shock cannot be underestimated. It has gotten incumbents of every political stripe kicked out in national elections worldwide by a wave of otherwise unlikely voters and this is also going to happen in the US. And the scary thing is, so many people on the Democratic side (and pollsters) are not accounting for this effect at all.
Trump didn't get any more votes than 2020 so you were right there. Harris got about 14MM fewer than Biden. What I struggled was the misogyny and racism in the country.
I have to go with Nate Silver's model, forecasting 53.8% for Trump (though just 25.7% for popular vote). I perceive you as having a greater pro-blue bias than Nate, and of course, a dramatically less detailed model.
You're basically reasoning about what the electorate "really" thinks, but I would guess that if your model is halfway-decent at understanding the electoral college, and if you shift it by roughly 1% of Americans (in swing states), then the probability ends up looking pretty much like Nate's.
And I notice that what you haven't done here is explain why ~47% of voters *did* vote for Trump in the past. I think if you don't have a solid, visceral understanding of why people vote for Trump (or, for that matter, against Harris or Hillary), then you'll probably get the vibes wrong and perform worse than a quantitative model like Nate's.
To me, the worst thing about Trump is that he lies and bullshits constantly, with his criminal behavior and total immorality a close second. The thing is, most voters (who have ever considered voting for him) have either never really believed that he lies and bullshits constantly, or, at least, distrust the "establishment" so deeply that it seems like Trump/Harris are almost equally bad. The reason Trump was so popular right off the gate in 2016, I believe, is that most of his lies were already popular memes on Facebook. He could simply take whatever nonsense he saw on Facebook or Twitter at face value and repeat it as gospel truth, and this made him seem very trustworthy to conservatives because that's what most conservatives were already doing. He outperformed other conservatives because he wasn't in one of the usual high-brow conservative clubs, he was in a low-brow "Facebook conservative" club like most of the primary voters were.
Even if, for example, voters understand that his claims of innocence are self-serving and therefore give those little weight, once Trump became nominee in 2016, a huge entire media ecosystem pivoted to convincing people that whatever Trump said was true ― no matter what it was. I noticed this starting with the "big groups of Muslims in New Jersey really did cheer 9/11" of 2016, and I assume this has continued with "the election was stolen" in 2020 and "trials against me are a witch hunt" of 2024. I haven't paid much attention to that ecosystem, but I noticed it a little after "they's eating the dogs, they're eating the cats, they're eating the pets of the people that live there". Trump doubled down, so conservative media either doubled down too, or looked the other way.
Hell, do you ever see people bother to mention that Trump's favorite phrase "fake news" was originally the term used to describe the Macedonians writing completely fake pro-Trump anti-Hillary "news articles" to earn ad revenue (in an article that IIRC was published on election day 2016 or the day after)? It seemed to me like Trump had won control of that phrase within two weeks, just by nakedly accusing the mainstream media of "fake news" with zero evidence. This reflects a broader strategy of people like Trump: take whatever you're accused of and throw it at your opponents. Suppose Trump kills a Democrat by appointing a series of pro-Trump generals, colonels and sergeants and giving the order―an act for which he'd be immune from criminal prosecution, as you know. I don't really think he'd do it, but if he wanted to, he might start Putin-style, by seeding stories about all kinds of imaginary plots by Democrats to kill Republicans. If by chance an important Republican should die of natural causes, Trump would immediately launch an investigation into the "foul play". By the time he assassinates the Democrat, then, it all seems like a wash, with Trump looking no worse than those "killer Democrats".
To be clear, the fundamental problem here is epistemology: most people's is terrible. Even on LessWrong, when I proposed a grassroots web site for hashing out the truth based on evidence[1], it got a cool reception. One commenter repeated Russian propaganda; another disagreed that truthseeking was difficult, saying 'To be clear, I'm saying that [true claims inherently "sound better" than false claims]': my proposal was unnecessary because this person believed they could tell whether a claim was true just by _reading it_.
And on the flip side, Matt Yglesias and Nate Silver have done a good job of highlighting the weaknesses of liberals and the Harris campaign relative to conservatives, e.g. conservative media tends to have the pragmatism not to talk about their pro-billionaire anti-health-care anti-abortion stances, while Harris hasn't really explained how her views apparently changed since 2019 (so the changes may seem insincere) and liberal groups will often blast Biden/Harris for having views they don't like, even when Trump is even worse. (And why wasn't Josh Shapiro nominated as VP?) Scott Alexander, too, recently endorsed Harris but highlighted things he *really* didn't like about liberals and "woke" Democrats, and it's pretty obvious that those sentiments aren't unusual in the general public. So given all this, a Trump victory would not be surprising.
[1] https://www.lesswrong.com/posts/9KAskejZdhskyriHk/let-s-make-the-truth-easier-to-find
Also, something I haven't heard Nate talk about is what happens if the election is very close in the swing states. What may happen in that case is that something important goes before the supreme court ― the same court with three Trump appointees that awarded Trump not only "absolute immunity" from criminal prosecution for "official" acts, but also an evidentiary shield that prevents "official" presidential activities from being used as evidence of a non-presidential crime.
Balancing that, Nate's latest projection is that Trump is down to 51.5%. He notes in his latest articles, though, that there's a lot of "herding" going on, and several things that "are hard to reconcile in the polls"[1]:
> - Why likely voter models help Harris in national polls but hurt her in the swing states.
> - Why some high-quality pollsters like Selzer and NYT/Siena, who are notorious for not herding, are so different than the others.
> - Why demographic shifts are extremely pronounced in the crosstabs, especially Harris losing ground with Black and Latino voters, but produce relatively small swings in the toplines.
> - Why so few polls in states like Wisconsin — less than 10 percent! — dare to show anything outside the range of Harris +2 to Trump +2, even though this is incredibly statistically unlikely.
> - Why there’s a palpable gap between Harris and Trump’s favorability ratings, but she doesn’t consistently lead him in the head-to-heads. There are some other good explanations for this one, but it could reflect herding — pollsters might not bother to tinker as much with their favorability ratings as with their head-to-heads.
> - And why national polls have swung quite a bit toward Trump over the past several weeks but battleground state polls have moved only slightly.
[1] https://www.natesilver.net/p/a-shocking-iowa-poll-means-somebody
You have to be thorough in your post mortems https://www.perplexity.ai/search/how-many-votes-did-harris-lose-onXDDTnITb2YPHcvBYQbCA#0
What about Terence Tao?
Just don’t really matter with these predictions because without the senate she has lost anyway
The effects of post-Covid sticker shock cannot be underestimated. It has gotten incumbents of every political stripe kicked out in national elections worldwide by a wave of otherwise unlikely voters and this is also going to happen in the US. And the scary thing is, so many people on the Democratic side (and pollsters) are not accounting for this effect at all.
Please encourage people to get out the vote.