Wouldn't it make more sense to do an "isolation chamber" reaction to the debate by isolating yourself from the candidates and only listening to the barrage of spinners?
Okay, I am being a wiseass, but seriously, unless one of these two men (or God forbid both) demonstrate in a truly undeniable manner that they no longer know which planet they reside on, the real fight will occur after the debate. I predict that that will be more interesting and enlightening as to the values of the two campaigns. Since right wing media appears to have pre-sold a lot of extreme claims about Biden's senility, I expect aggressive gaslighting on that subject.
Which reminds me... I doubt I am the only one around here who gets fundraising texts from both sides of the aisle, including both presidential campaigns. I find the difference in tone startling.
The reds call me Patriot and order me to contribute immediately, otherwise, they will TERMINATE (their caps) me in some form or other -- most recently my House GOP membership... or that I will EXPIRE
The blues tell me a sad story and offer me an incredible financial deal with matching funds to my contribution appearing magically out of thin air.
Based just upon my box of text messages, I would think that the mafia was running against a bunch of online scammers. I'm sure that both succeed in raising money, but is either actually a good image politically?
"...unless one of these two men (or God forbid both) demonstrate in a truly undeniable manner that they no longer know which planet they reside on, the real fight will occur after the debate..."
I would make a slight but important edit. I think if neither of them truly flame out, there will simply be no importance whatsoever to either the debate or the post debate spin. Debates generally don't move the needle much, but I'd say if they do it's because of something in the debate, not the spin after.
Excellent, can't wait! One question: is the model over at 538 totally new and of no relation to your prior work? I had wondered that previously, but my interest was again piqued when you said your upcoming model is the only one with a long term track record. Doesn't really matter...just curious.
My understanding is that the 538 model is brand new, written from scratch. But it seems to follow the gist of Nate's model, with pollster ratings, correlated polling error in different states, elasticity, fundamentals, etc.
Assuming they did a good job of it, the results for the 538 model will likely be pretty similar to Nate's, I would think...
I don't know how Nate handled the model's source code while he was at FiveThirtyEight; if it was "open-sourced" so that the rest of the crew could modify it and see how it worked, or if he black-boxed it to keep his proprietary stuff secret. If it was available to the rest of the team, that would make it easier for them to create a similar model, avoid known pitfalls and edge-cases, etc. I'm not saying they'd plagiarize the model, but simply leverage what they learned from it.
Seems like some of the mathy bits may be substantially different, but at least in terms of inputs and layman description of what's happening it really seems indistinguishable. As such I too expect Nate's results to be very similar.
Not being more than a stats aficionado, it sure looks to me as being overfitted (even though the author specifically argues he has taken measures to avoid this). Also, I believe fundamentals get completely thrown out of Nate's model by Election Day, while they won't in 538's new model. Nate's also tends to take polls results more at face value, I would say...
Therefore (wild-ish bet) I expect Nate's model to be more optimistic on Trump's chances right now (maybe around 60/40?), and also to be fluctuating more.
Anyway, Nate is really tired of "model wars" so I don't expect any answer from him on this, and having followed his thoughts on the issue for years I completely understand him.
Q&A question: How does the model incorporate the actuarial risks for two very old candidates and the seemingly higher risk that a candidate may drop out before Election Day?
Understand that Thursday night will NOT be a debate. It will be a two-person Press Conference.
A TRUE debate, in the Lincoln-Douglas tradition, would have each candidate, in turn, DIRECTLY pose a question (within a given time-limit) to the other on ANYTHING.
The other candidate would then have (within a given time-limit) the opportunity to answer the question, and then ask a question of their own.
This would go then go back and forth for the predetermined length of the 'event'.
Other than enforceing time-limits, the 'Moderator' would have NO role (In fact it would be best for an automatic timer to cut the respective mics at the end of each time-limit with no human intervention).
That's pretty much how Lincoln and Douglas did it, and pretty much how most people would like to see it done.
Q&A Question: Do you feel a larger responsibility now that your model has a large following? I would stop short of saying it has an impact on the election, but it definitely shapes a lot of online and media narratives - which in turn could have some impact.
Obviously I know you're making it as accurate as you can, but how do you grapple with the model potentially having an impact on the very thing it's measuring?
One of the weirder things that’s happened over the last ~5 years, and especially the past ~18 months, is the level of attention paid to statistical models, especially by folks who have never built or paid attention to them before.
Call me naive, but the whole idea of a model “war” seems just silly. The whole point of a model is to try to understand/predict something, not to advocate for one thing or another. It seems like a classification error to “war” about a model because it always seems to come from folks who want to advocate for their position rather than actually try to understand the world better.
Not that there’s anything wrong with advocacy *per se*; that’s just not what modeling is for!
How does your model take into account an Independent candidate's chances of flipping a Republican seat when no Democrat candidate is running? In Nebraska, Dan Osborn is running a decent race (polling seems reasonably tight given the number of undecideds) against an unpopular GOP senator - but the traditional press seems to be overlooking what could be a closer Senate race than in Texas or Florida.
Very onboard with your plan, Nate. Looking forward to the patience based model, and your debate review (I doubt I can stomach watching it myself).
Ps. I used to live nextdoor to a Sri Lankan foodie, so I think sriracha is pretty mild too. Try some Golden Cadillac wings next time you visit Toronto :)
Question for Q & A: There is a lot of Christianity in Politics right now (Louisiana’s Ten Commandments law taking the state further in its own weird direction). Is this a case of Red states getting redder? Is there likely to be a positive or negative reaction in the key purple states?
For the q and a. What do you think are the best reasons for having/not having kids? Is this really why you chose the way you did? Would love to hear this on the podcast from both of you
Poker Question for the Q&A, how many players in the super high roller ($250k) would you guys have 100% of themselves. What do you think is the skill difference in a $50k/$250k, how rich would you personally feel you'd need to be to buy in to that tournament? Thanks nate, love all the content!
Hey just to clarify a different point, when you say "have 100% of themselves" do you mean nobody else has staked them? I.e. it's all their own money? Sorry, just not quite sure of that turn of phrase.
Q&A Question: during your most recent podcast, you noted that with Global Warming, we'd expect more extreme weather events. You noted that when you move a standard distribution to the right, more of the distribution falls in the range of what was "extreme" in the original distribution. Matt Yglesias has used a similar image in describing how a leftward shift in median public opinion ends up generating more extreme leftists.
Without disputing the specific points above, my question is: why should we assume the entire distribution moves? Couldn't the mean/median move in one direction without moving the entire distribution? For example, if reducing malnutrition raises the median adult height of a population, would we expect there to be meaningfully more 7ft tall people? In the case of Global Warming, wouldn't a shift of the entire distribution imply LESS extreme cold events (which is contrary to my understanding of climatologists predictions)?
More generally, how do you think about when the entire distribution shifts, rather than a shift in the key underlying parameters (mean, median, std dev) etc?
Thank you so much, Nate, for building a platform that is substantially independent. Please do comment on Maine and Nebraska on their curious electoral vote status. Can Maine and Nebraska throw the presidential one way or the other?
Wouldn't it make more sense to do an "isolation chamber" reaction to the debate by isolating yourself from the candidates and only listening to the barrage of spinners?
Okay, I am being a wiseass, but seriously, unless one of these two men (or God forbid both) demonstrate in a truly undeniable manner that they no longer know which planet they reside on, the real fight will occur after the debate. I predict that that will be more interesting and enlightening as to the values of the two campaigns. Since right wing media appears to have pre-sold a lot of extreme claims about Biden's senility, I expect aggressive gaslighting on that subject.
Which reminds me... I doubt I am the only one around here who gets fundraising texts from both sides of the aisle, including both presidential campaigns. I find the difference in tone startling.
The reds call me Patriot and order me to contribute immediately, otherwise, they will TERMINATE (their caps) me in some form or other -- most recently my House GOP membership... or that I will EXPIRE
The blues tell me a sad story and offer me an incredible financial deal with matching funds to my contribution appearing magically out of thin air.
Based just upon my box of text messages, I would think that the mafia was running against a bunch of online scammers. I'm sure that both succeed in raising money, but is either actually a good image politically?
"...unless one of these two men (or God forbid both) demonstrate in a truly undeniable manner that they no longer know which planet they reside on, the real fight will occur after the debate..."
I would make a slight but important edit. I think if neither of them truly flame out, there will simply be no importance whatsoever to either the debate or the post debate spin. Debates generally don't move the needle much, but I'd say if they do it's because of something in the debate, not the spin after.
Excellent, can't wait! One question: is the model over at 538 totally new and of no relation to your prior work? I had wondered that previously, but my interest was again piqued when you said your upcoming model is the only one with a long term track record. Doesn't really matter...just curious.
My understanding is that the 538 model is brand new, written from scratch. But it seems to follow the gist of Nate's model, with pollster ratings, correlated polling error in different states, elasticity, fundamentals, etc.
Assuming they did a good job of it, the results for the 538 model will likely be pretty similar to Nate's, I would think...
I don't know how Nate handled the model's source code while he was at FiveThirtyEight; if it was "open-sourced" so that the rest of the crew could modify it and see how it worked, or if he black-boxed it to keep his proprietary stuff secret. If it was available to the rest of the team, that would make it easier for them to create a similar model, avoid known pitfalls and edge-cases, etc. I'm not saying they'd plagiarize the model, but simply leverage what they learned from it.
Ok I went over there and found their article on the subject, which confirms roughly what you thought: https://abcnews.go.com/538/538s-2024-presidential-election-forecast-works/story?id=110867585
Seems like some of the mathy bits may be substantially different, but at least in terms of inputs and layman description of what's happening it really seems indistinguishable. As such I too expect Nate's results to be very similar.
Not being more than a stats aficionado, it sure looks to me as being overfitted (even though the author specifically argues he has taken measures to avoid this). Also, I believe fundamentals get completely thrown out of Nate's model by Election Day, while they won't in 538's new model. Nate's also tends to take polls results more at face value, I would say...
Therefore (wild-ish bet) I expect Nate's model to be more optimistic on Trump's chances right now (maybe around 60/40?), and also to be fluctuating more.
Anyway, Nate is really tired of "model wars" so I don't expect any answer from him on this, and having followed his thoughts on the issue for years I completely understand him.
Q&A question: How does the model incorporate the actuarial risks for two very old candidates and the seemingly higher risk that a candidate may drop out before Election Day?
Understand that Thursday night will NOT be a debate. It will be a two-person Press Conference.
A TRUE debate, in the Lincoln-Douglas tradition, would have each candidate, in turn, DIRECTLY pose a question (within a given time-limit) to the other on ANYTHING.
The other candidate would then have (within a given time-limit) the opportunity to answer the question, and then ask a question of their own.
This would go then go back and forth for the predetermined length of the 'event'.
Other than enforceing time-limits, the 'Moderator' would have NO role (In fact it would be best for an automatic timer to cut the respective mics at the end of each time-limit with no human intervention).
That's pretty much how Lincoln and Douglas did it, and pretty much how most people would like to see it done.
Q&A Question: Do you feel a larger responsibility now that your model has a large following? I would stop short of saying it has an impact on the election, but it definitely shapes a lot of online and media narratives - which in turn could have some impact.
Obviously I know you're making it as accurate as you can, but how do you grapple with the model potentially having an impact on the very thing it's measuring?
One of the weirder things that’s happened over the last ~5 years, and especially the past ~18 months, is the level of attention paid to statistical models, especially by folks who have never built or paid attention to them before.
Call me naive, but the whole idea of a model “war” seems just silly. The whole point of a model is to try to understand/predict something, not to advocate for one thing or another. It seems like a classification error to “war” about a model because it always seems to come from folks who want to advocate for their position rather than actually try to understand the world better.
Not that there’s anything wrong with advocacy *per se*; that’s just not what modeling is for!
Question for your Q&A:
How does your model take into account an Independent candidate's chances of flipping a Republican seat when no Democrat candidate is running? In Nebraska, Dan Osborn is running a decent race (polling seems reasonably tight given the number of undecideds) against an unpopular GOP senator - but the traditional press seems to be overlooking what could be a closer Senate race than in Texas or Florida.
Very onboard with your plan, Nate. Looking forward to the patience based model, and your debate review (I doubt I can stomach watching it myself).
Ps. I used to live nextdoor to a Sri Lankan foodie, so I think sriracha is pretty mild too. Try some Golden Cadillac wings next time you visit Toronto :)
For subscriber questions: what is your news diet? How do you stay up to date while also doing things besides reading the news?
Question for Q & A: There is a lot of Christianity in Politics right now (Louisiana’s Ten Commandments law taking the state further in its own weird direction). Is this a case of Red states getting redder? Is there likely to be a positive or negative reaction in the key purple states?
For the q and a. What do you think are the best reasons for having/not having kids? Is this really why you chose the way you did? Would love to hear this on the podcast from both of you
Poker Question for the Q&A, how many players in the super high roller ($250k) would you guys have 100% of themselves. What do you think is the skill difference in a $50k/$250k, how rich would you personally feel you'd need to be to buy in to that tournament? Thanks nate, love all the content!
Guess not guys.. sorry for the typo
Hey just to clarify a different point, when you say "have 100% of themselves" do you mean nobody else has staked them? I.e. it's all their own money? Sorry, just not quite sure of that turn of phrase.
Yes sir, that's exactly right. They use there own money for the buy in (but can still cross book action)
Excited for next week! Don't work too hard this weekend.
Q&A Question: during your most recent podcast, you noted that with Global Warming, we'd expect more extreme weather events. You noted that when you move a standard distribution to the right, more of the distribution falls in the range of what was "extreme" in the original distribution. Matt Yglesias has used a similar image in describing how a leftward shift in median public opinion ends up generating more extreme leftists.
Without disputing the specific points above, my question is: why should we assume the entire distribution moves? Couldn't the mean/median move in one direction without moving the entire distribution? For example, if reducing malnutrition raises the median adult height of a population, would we expect there to be meaningfully more 7ft tall people? In the case of Global Warming, wouldn't a shift of the entire distribution imply LESS extreme cold events (which is contrary to my understanding of climatologists predictions)?
More generally, how do you think about when the entire distribution shifts, rather than a shift in the key underlying parameters (mean, median, std dev) etc?
>For example, if reducing malnutrition raises the median adult height of a population, would we expect there to be meaningfully more 7ft tall people?
Yes. All the potential 7fts who were malnourished and didn't make it would.
Thank you so much, Nate, for building a platform that is substantially independent. Please do comment on Maine and Nebraska on their curious electoral vote status. Can Maine and Nebraska throw the presidential one way or the other?
Will you ever drop transcripts of the podcasts?