Thank you for this reflection. While you didn't end up on the right side of maybe this time, your willingness to go out on a limb and make a forecast that went against prediction markets and the crowd based on your own perspective actually contributed to the information ecosystem making everyone's forecasts better.
Thanks, Jonathan! I think that's the right way to think about it. Hopefully we can generate useful insights—and the process is valuable—even when our inferences are wrong.
I think you may be being too hard on yourself regarding the prediction. Giving something a 30% chance means it will happen a lot. There was a lot of noise and not a lot of great signals. As Yogi said “Predicting is hard. Especially about the future.”
Thanks, Lawdog. You're absolutely right that it's not that bad in itself that something I gave a 35% chance happened—and forecasting is very hard. I don't want to be too hard on myself. But other forecasts gave it a higher chance, which is evidence they probably knew something I didn't. In hindsight, the reasons I had for thinking it wouldn't happen look fairly wrong and while I knew Trump could win, I did think a decisive win like this was very unlikely.
I applaud you for owning the error, rather than trying to wriggle out of it. And I agree completely with Lawdog's comment. Your prediction was probabilistic, not absolute, and if your probabilistic predictions weren't sometimes wrong, that would indicate that they were poorly calibrated. So taking the L and moving on is just the right response.
Thanks, Bill! I don't want to be too hard on myself, but I don't think it was a good forecast. As I just told Lawdog, it's not that bad itself that something I gave a 35% chance happened. But what actually did happen makes it look worse. It doesn't look like a close miss and my reasons for thinking Harris would win don't look compelling in hindsight. And although I didn't spell it out in my forecast post, I thought Trump winning increasing his support across the board and winning the popular vote was much less likely than 35%. I don't want to overindex on one case, but I think incumbent unpopularity and cost-of-living are much more important indicators of who will win an election than some of the other stuff I thought would matter.
As someone who spends time deeply studying these topics, I always enjoy reading your analysis and predictions. If it's any solace, not just you, even Allan Lichtman, who has had a (near) flawless presidential election prediction record has been proven wrong.
But, I would like your opinion on my perspective going into this election and (now) its aftermath.
The day kamala announced her candidacy, I believed Trump would. I had the same guess the moment Hillary announced her candidacy.
Both times, there were 2 key aspects in my mind
1. I believe american presidential elections is about voting for a candidate that people can identify with, in some form or fashion. In that sense, on a (presidential election district weighted) average basis, americans prioritize candidates on these hygiene factors -
- Male
- Christian
- White
(Exception to this would be too progressive? radical? un-relatable ?) IMHO, this is not as much a bias as it is a blind spot / unspoken part of american political discourse.
2. In most western nations (and probably eastern ones too), once a society makes a "progressive" move (e.g. electing a black man as a president), said progressive move or even the general progressive direction does not repeat for a loooong time (I believe, for decades).
This is not me being judgey. These are my humble impressions of having read/heard whatever historical material (mostly podcasts). This is not my personal preference.
Past is rarely a predictor of future. However, IMHO, american citizenry, making any correction towards more "practical" of lens for a presidential candidate seems far off.
In said vain, here is my prediction for the 2028 presidential election. If the democratic party fields a candidate that does not meet all 3 hygiene factors above, they will lose. Hands down. Irrespective of who the republican candidate is, as long the republican party does not violate the 3 hygiene factors.
Thanks. I certainly agree racism, sexism, and christian nationalism played a big role in this election. I also think you're right that Democrats are very likely to nominate a white, Christian man in 2028, and that there's likely to be a backlash when someone like Obama who codes as "other" wins. But I don't actually think that Kamala or some other person who isn't a white, Christian male couldn't win. I think in another year, Kamala could have won. Women, people of color, and non-Christians have been elected to other major offices. I think the dominant factor this year was voter unhappiness with the incumbent party. If I'm right about what the Trump administration is going to be like, I think the political thermostat will have swung the other way.
Thank you for this reflection. While you didn't end up on the right side of maybe this time, your willingness to go out on a limb and make a forecast that went against prediction markets and the crowd based on your own perspective actually contributed to the information ecosystem making everyone's forecasts better.
Thanks, Jonathan! I think that's the right way to think about it. Hopefully we can generate useful insights—and the process is valuable—even when our inferences are wrong.
I think you may be being too hard on yourself regarding the prediction. Giving something a 30% chance means it will happen a lot. There was a lot of noise and not a lot of great signals. As Yogi said “Predicting is hard. Especially about the future.”
Thanks, Lawdog. You're absolutely right that it's not that bad in itself that something I gave a 35% chance happened—and forecasting is very hard. I don't want to be too hard on myself. But other forecasts gave it a higher chance, which is evidence they probably knew something I didn't. In hindsight, the reasons I had for thinking it wouldn't happen look fairly wrong and while I knew Trump could win, I did think a decisive win like this was very unlikely.
I applaud you for owning the error, rather than trying to wriggle out of it. And I agree completely with Lawdog's comment. Your prediction was probabilistic, not absolute, and if your probabilistic predictions weren't sometimes wrong, that would indicate that they were poorly calibrated. So taking the L and moving on is just the right response.
Thanks, Bill! I don't want to be too hard on myself, but I don't think it was a good forecast. As I just told Lawdog, it's not that bad itself that something I gave a 35% chance happened. But what actually did happen makes it look worse. It doesn't look like a close miss and my reasons for thinking Harris would win don't look compelling in hindsight. And although I didn't spell it out in my forecast post, I thought Trump winning increasing his support across the board and winning the popular vote was much less likely than 35%. I don't want to overindex on one case, but I think incumbent unpopularity and cost-of-living are much more important indicators of who will win an election than some of the other stuff I thought would matter.
Robert,
As someone who spends time deeply studying these topics, I always enjoy reading your analysis and predictions. If it's any solace, not just you, even Allan Lichtman, who has had a (near) flawless presidential election prediction record has been proven wrong.
But, I would like your opinion on my perspective going into this election and (now) its aftermath.
The day kamala announced her candidacy, I believed Trump would. I had the same guess the moment Hillary announced her candidacy.
Both times, there were 2 key aspects in my mind
1. I believe american presidential elections is about voting for a candidate that people can identify with, in some form or fashion. In that sense, on a (presidential election district weighted) average basis, americans prioritize candidates on these hygiene factors -
- Male
- Christian
- White
(Exception to this would be too progressive? radical? un-relatable ?) IMHO, this is not as much a bias as it is a blind spot / unspoken part of american political discourse.
2. In most western nations (and probably eastern ones too), once a society makes a "progressive" move (e.g. electing a black man as a president), said progressive move or even the general progressive direction does not repeat for a loooong time (I believe, for decades).
This is not me being judgey. These are my humble impressions of having read/heard whatever historical material (mostly podcasts). This is not my personal preference.
Past is rarely a predictor of future. However, IMHO, american citizenry, making any correction towards more "practical" of lens for a presidential candidate seems far off.
In said vain, here is my prediction for the 2028 presidential election. If the democratic party fields a candidate that does not meet all 3 hygiene factors above, they will lose. Hands down. Irrespective of who the republican candidate is, as long the republican party does not violate the 3 hygiene factors.
Do let me know what you think.
Thanks. I certainly agree racism, sexism, and christian nationalism played a big role in this election. I also think you're right that Democrats are very likely to nominate a white, Christian man in 2028, and that there's likely to be a backlash when someone like Obama who codes as "other" wins. But I don't actually think that Kamala or some other person who isn't a white, Christian male couldn't win. I think in another year, Kamala could have won. Women, people of color, and non-Christians have been elected to other major offices. I think the dominant factor this year was voter unhappiness with the incumbent party. If I'm right about what the Trump administration is going to be like, I think the political thermostat will have swung the other way.