Does Matt actually think the working families party putting out statements is affecting the polling average? Hard to see how anyone intelligent could sincerely believe that. He's starting to sound really silly on the leftists are annoying point. Like a caricature of a caricature.
I think a more charitable interpretation of what Matt was saying is that he was using WFP statements as an example to illustrate that in order for the Dems to generate positive vibes and earned media (which in turn affects the polling average), the messaging needs to come from the top (i.e., Harris) because the left-of-center rank and file won't fall lockstep behind anything the administration does, unlike Republicans and conservative media vis-à-vis Trump.
I do think that Harris and Walz need to find more mileposts to keep being in the news, and being omnipresent on TV and radio will help. Tomorrow (9/12) challenging Trump to more debates would certainly qualify as a good thing to do.
That said, I think a lot of white pundits and journalists on Twitter have maybe missed the recent interviews she did with Rickey Smiley (black radio host based in Atlanta) and "Angel Baby" (latina radio host based in Phoenix). In both hits, she laid on thick her economic message in very populist terms. Those are the kind of highly-targeted media hits that people are calling for her to do. But then they don't really notice when she actually does it.
So anyway, she should do a mix of these things. More interviews in all mediums would be helpful just to keep driving the conversation.
Nate Silver has real ontological confusions around his models, and has for years. I work with statistical models all the time, and he is particularly bad about this—perhaps because he does not hang out in academic circles where people are so often corrected on such confusions.
1) When the medical professional talks about the chances of survival, they have a LOT of data and rather less theory. The question of ontology (i.e., is the die already cast) vs. epistemology (i.e., are our tools and data just not good enough yet) is VERY hard to disentangle. Obviously, it's some of each, but they don't know how much is ontological uncertainty vs. epistemological uncertainty.
2) That is NOT the case with most models, and certainly not formal models designed for prediction. There's this idea of "parsimony," in modeling. Make it as simple as you can and still get good results. That means making a lot of simplifying assumptions. That makes the model more flexible, avoids overfitting to old data, but DEFINITELY gets in the way of precisely matching future outcomes. It aims to avoid really catastrophic modeling mistakes, but those simplifying assumptions mean that the model is NOT doing what reality is doing. It's predictive model, but not an ambitious/tight simulation.
3) This kind modeling really gives up on leaning into reducing ontological uncertainty. It doesn't try to match reality.
4) But then Nate makes this really unusual ontological mistake. He refers to "The Model," all the time, rather than "My model," or "this version of my model." He talks as though the current version of his model for election prediction as though it has high epistemological certainty, as though the reality of who will win is really so unfixed as to be THAT malleable to each news cycle. But, as Matt suggestion, no the race is not shifting that much that often. It's just that Nate's model is that sensitive to minor polling shifts.
5) Diving into the substance, I suspect that a HUGE driver of election uncertainty is whom can be convinced to turn out and vote. I STRONGLY suspect this. I don't just mean GOTV efforts. I mean the slightly leaning folks who don't necessarily care THAT much about voting....people who are. NOT good GOTV targets because of their weaker candidate preferences. Can the annoyed just enough or inspired just enough to actually go vote? That's a crazy uncertain thing, and very delicate. But I don't think it shows up in polling or voter models.
So, prediction models look in the other places because that's where the light (or data) is better, not because that's where the keys (or real variance) are. In fact, there is a lot of epistemological uncertainty there, in addition to real ontological uncertainty.
6) Nate likes to brag about how well calibrated his models are. What I would like to see—and I don't that he has ever published this—is week-by-week or month-by-month examinations of calibration. I would expect that the calibration gets better as we approach the predicted event. I just don't believe that his models are as good as he want to believe they are. I don't even think we should be trying to run them before Labor Day, and I certainly do not think that they are precise enough to give meaning information when run every day a month or two out.
Well they could've done that, _either_ by voting for somebody else in the primary, or by having him step aside before the convention, and picking somebody else there. At this point, it's too late to swap him out on the ballot, the party has made its choice and there wouldn't be any legal mechanism to make a new one and ask the states to list somebody else.
I suppose in some fantasy world where Trump wasn't a malignant narcissist, he could at least say that if elected, he will resign immediately, and that a vote for him is a vote for JD Vance. Not that Vance is exactly "a more popular one".
2. It likely makes zero difference (brian you should read this and really take it to heart because it fundamentally interferes with your basic thesis analysis about media…crushing Trump when their is like 60m people watching and it makes zero difference means maybe media isn’t the major issue)
Matt, dude, you got to stop being contrarian about everything a dem does and bitching about activists. Dude, all you do is criticize dems and dem voters. Which is fine because a lot of them are annoying and a lot of their policies are stupid…but talking about throwing rocks in glass houses. You do realize you act exactly like an activist right?
Brian I am shocked you haven’t realized that it is impossible for dems to work the refs like republicans. Until you own half the media ecosystem that is basically a propaganda network, it work work. Why? You have (not you in particular but you as in dems) no power to actually work the refs. Where the fuck else are these people going to go (dem pundits and politicians)? Fox News? That is the difference. Oh and dems can’t do propaganda for all the other reasons that matt actually points out…dems can’t agree about anything. To do propaganda you have to actually pretend things are great. Do you see that happening?
My god man, stop talking about Arlington. No one cares. You do realize Trump got more popular when Dems did what you wanted (charging him with OBVIOUS crimes). I am just lost how you think your strategy works when you have real world examples at the highest level (Trump trials) and all that did is make Trump MORE popular than before 2016. At some point, you have to show your work. You can’t just say “see it works for republicans!!” No shit. They have a homogenous voting block..dems don’t.
Matt can be insufferable like that. I read him to know how DC Dems increasingly think. He won't be satisfied until they're all in lockstep with him. And then he'll find something else to be contrarian about. It's the way of the pundit. Get clicks and listens and reads.
Your last part is what is so frustrating to me. On the one hand he will do what he always does, bitch and moan about dems, and then say “dems need to do x.” Then dems for some unknown reason do x. Next day, he complains about dems doing c
I also find the Arlington stuff exhausting -- I'm not sure why anyone thinks that the 18th controversy about Trump "Disrespecting The Sainted Troops" would be any more politically efficacious than the first 17. But I also understand why pro-Dem pundits want to push whatever buttons they have in front of them.
Yep. Look I hate Trump. It would make me feel GREAT. However there is no evidence this works. In fact acting like a victim seems to be trump’s mutant power. If anything the fact he is being tried for openly committing crimes and it’s made him more popular kind of disproves Brian’s basic premise
Agree completely. But what I don’t understand about Brian’s opinion is that it just won’t happen. Although it works for republicans it won’t work with dems. Republicans are homogeneous group: white men. Dems are just too diverse to do this effectively
The most important part that you highlight is that the very left is very much like maga. Not in policy or in temperament but in the need to own/control the party. This is fundamentally what the freedom caucus is doing. They couldn’t give a shit about policy. They just want to control the Republican Party. I’m not even sure they want their insane policies to pass.
On media, it’s definitely about the voters and integrity but it’s also that Dems, in general, want to bitch about Dems. This doesn’t mean they like maga but they definitely don’t care as much about just hating on maga. If you look at the most leftist media (the nation or pod save America…maybe not leftist but partisan), they complain all the time about Dems.
Matt fails in my mind because he HAS to be be contrarian. He physically can’t agree with a policy of either party, even if he likes it. He will always complain about something (specifics, tradeoffs, or process). That’s why it’s hard to take him seriously on “what to do.” He is smart and thoughtful but just awful politically
There are two-way polls that only ask about about Harris and Trump. I've not seen that these polls tightened less since RFK Jr.'s dropout/endorsement than more open polls.
I'm not that sophisticated--I really meant it as a question, not a statement of fact. If you've looked at it and seen the two-way polls and open polls track in lockstep, that likely suggests the tightening is more reversion to the mean than event-driven. Though I'd still think the presence of an active 3rd party candidate in the race influences the two-way poll results, as does RFK Jr.'s endorsement of Trump on his way out the door?
Looking at Nate Silver's polling averages (I'm eyeballing this, not doing any detailed analysis): On the day RFK Jr. dropped out, the polls were Harris 48.0% (in the midst of her convention bump), Trump 43.7%, Kennedy 3.9%. Today, they show Harris 48.7% (+0.7%, though down from her peak by ~0.5%), Trump 46.8% (+3.1%). That looks to my untrained eye like it could plausibly be driven largely by RFK Jr. voters coming off the two-way sidelines and skewing Trump.
Body language said it all. She walked across the stage to shake his hand; he did everything but hide behind his podium. She spoke with passion; he was pinch-faced and angry. All the white eye makeup in the world couldn't make his eyes seem open.
I think it’s inaccurate and shortsighted to say that all of those 55% that say they want major change from the Biden policy agenda are inherently saying they won’t vote for her, or even wouldn’t have voted for Biden. I would love to see major changes from many of his policies AND i would’ve voted for him and will vote for her because they are the lesser of two evils.
It was bizarre that during the question on Israel, Trump pivoted to Ukraine and Russia. Israel is a good issue for Trump as Democrats are deeply divided on the issue, while Republicans are not. Ukraine is the opposite, where Democrats are united but Republicans are divided. So why take precious time on the Israel question to talk about Ukraine?
Per the discussion of Nate's forecast, y'all forgot to factor in "response bias" into consideration. After the "trauma" of the "will he / wont he?" July, there was no doubt a "response bias" in exactly who was answering the phone and responding to the polls themselves such that Democrats were more likely to answer the phone and take a call and Republicans were less likely due to the dismay of Trump's struggles post Biden leaving the race.
I would suggest some of the "slip" in polls picked up in the forecast was associated with a mean reversion in this bias as excitement tempered a bit post Convention... (also probably why Harris's bump was so small historically speaking).
It's debatable whether a response bias correlates with any real shift in actual turnout or not...
ah well it was basically brought up in the wrap up as "factored in" by Brian but i am not sure how it could be factored in if we spent the first 90% of the podcast looking for a reason that the mean reverted without bringing up response effects post swap / convention...
Does Matt actually think the working families party putting out statements is affecting the polling average? Hard to see how anyone intelligent could sincerely believe that. He's starting to sound really silly on the leftists are annoying point. Like a caricature of a caricature.
I think a more charitable interpretation of what Matt was saying is that he was using WFP statements as an example to illustrate that in order for the Dems to generate positive vibes and earned media (which in turn affects the polling average), the messaging needs to come from the top (i.e., Harris) because the left-of-center rank and file won't fall lockstep behind anything the administration does, unlike Republicans and conservative media vis-à-vis Trump.
I do think that Harris and Walz need to find more mileposts to keep being in the news, and being omnipresent on TV and radio will help. Tomorrow (9/12) challenging Trump to more debates would certainly qualify as a good thing to do.
That said, I think a lot of white pundits and journalists on Twitter have maybe missed the recent interviews she did with Rickey Smiley (black radio host based in Atlanta) and "Angel Baby" (latina radio host based in Phoenix). In both hits, she laid on thick her economic message in very populist terms. Those are the kind of highly-targeted media hits that people are calling for her to do. But then they don't really notice when she actually does it.
So anyway, she should do a mix of these things. More interviews in all mediums would be helpful just to keep driving the conversation.
Brian, can you just occasionally shut Matt up?
Most of us pay to hear Matt talk.
Nate Silver has real ontological confusions around his models, and has for years. I work with statistical models all the time, and he is particularly bad about this—perhaps because he does not hang out in academic circles where people are so often corrected on such confusions.
1) When the medical professional talks about the chances of survival, they have a LOT of data and rather less theory. The question of ontology (i.e., is the die already cast) vs. epistemology (i.e., are our tools and data just not good enough yet) is VERY hard to disentangle. Obviously, it's some of each, but they don't know how much is ontological uncertainty vs. epistemological uncertainty.
2) That is NOT the case with most models, and certainly not formal models designed for prediction. There's this idea of "parsimony," in modeling. Make it as simple as you can and still get good results. That means making a lot of simplifying assumptions. That makes the model more flexible, avoids overfitting to old data, but DEFINITELY gets in the way of precisely matching future outcomes. It aims to avoid really catastrophic modeling mistakes, but those simplifying assumptions mean that the model is NOT doing what reality is doing. It's predictive model, but not an ambitious/tight simulation.
3) This kind modeling really gives up on leaning into reducing ontological uncertainty. It doesn't try to match reality.
4) But then Nate makes this really unusual ontological mistake. He refers to "The Model," all the time, rather than "My model," or "this version of my model." He talks as though the current version of his model for election prediction as though it has high epistemological certainty, as though the reality of who will win is really so unfixed as to be THAT malleable to each news cycle. But, as Matt suggestion, no the race is not shifting that much that often. It's just that Nate's model is that sensitive to minor polling shifts.
5) Diving into the substance, I suspect that a HUGE driver of election uncertainty is whom can be convinced to turn out and vote. I STRONGLY suspect this. I don't just mean GOTV efforts. I mean the slightly leaning folks who don't necessarily care THAT much about voting....people who are. NOT good GOTV targets because of their weaker candidate preferences. Can the annoyed just enough or inspired just enough to actually go vote? That's a crazy uncertain thing, and very delicate. But I don't think it shows up in polling or voter models.
So, prediction models look in the other places because that's where the light (or data) is better, not because that's where the keys (or real variance) are. In fact, there is a lot of epistemological uncertainty there, in addition to real ontological uncertainty.
6) Nate likes to brag about how well calibrated his models are. What I would like to see—and I don't that he has ever published this—is week-by-week or month-by-month examinations of calibration. I would expect that the calibration gets better as we approach the predicted event. I just don't believe that his models are as good as he want to believe they are. I don't even think we should be trying to run them before Labor Day, and I certainly do not think that they are precise enough to give meaning information when run every day a month or two out.
If i was a republican, I would simply replace my unpopular candidate with a more popular one :^)
Well they could've done that, _either_ by voting for somebody else in the primary, or by having him step aside before the convention, and picking somebody else there. At this point, it's too late to swap him out on the ballot, the party has made its choice and there wouldn't be any legal mechanism to make a new one and ask the states to list somebody else.
I suppose in some fantasy world where Trump wasn't a malignant narcissist, he could at least say that if elected, he will resign immediately, and that a vote for him is a vote for JD Vance. Not that Vance is exactly "a more popular one".
I want to stipulate 2 things:
1. Harris crushed Trump in the debate.
2. It likely makes zero difference (brian you should read this and really take it to heart because it fundamentally interferes with your basic thesis analysis about media…crushing Trump when their is like 60m people watching and it makes zero difference means maybe media isn’t the major issue)
Matt, dude, you got to stop being contrarian about everything a dem does and bitching about activists. Dude, all you do is criticize dems and dem voters. Which is fine because a lot of them are annoying and a lot of their policies are stupid…but talking about throwing rocks in glass houses. You do realize you act exactly like an activist right?
Brian I am shocked you haven’t realized that it is impossible for dems to work the refs like republicans. Until you own half the media ecosystem that is basically a propaganda network, it work work. Why? You have (not you in particular but you as in dems) no power to actually work the refs. Where the fuck else are these people going to go (dem pundits and politicians)? Fox News? That is the difference. Oh and dems can’t do propaganda for all the other reasons that matt actually points out…dems can’t agree about anything. To do propaganda you have to actually pretend things are great. Do you see that happening?
My god man, stop talking about Arlington. No one cares. You do realize Trump got more popular when Dems did what you wanted (charging him with OBVIOUS crimes). I am just lost how you think your strategy works when you have real world examples at the highest level (Trump trials) and all that did is make Trump MORE popular than before 2016. At some point, you have to show your work. You can’t just say “see it works for republicans!!” No shit. They have a homogenous voting block..dems don’t.
Matt can be insufferable like that. I read him to know how DC Dems increasingly think. He won't be satisfied until they're all in lockstep with him. And then he'll find something else to be contrarian about. It's the way of the pundit. Get clicks and listens and reads.
Your last part is what is so frustrating to me. On the one hand he will do what he always does, bitch and moan about dems, and then say “dems need to do x.” Then dems for some unknown reason do x. Next day, he complains about dems doing c
X
I also find the Arlington stuff exhausting -- I'm not sure why anyone thinks that the 18th controversy about Trump "Disrespecting The Sainted Troops" would be any more politically efficacious than the first 17. But I also understand why pro-Dem pundits want to push whatever buttons they have in front of them.
Yep. Look I hate Trump. It would make me feel GREAT. However there is no evidence this works. In fact acting like a victim seems to be trump’s mutant power. If anything the fact he is being tried for openly committing crimes and it’s made him more popular kind of disproves Brian’s basic premise
Agree completely. But what I don’t understand about Brian’s opinion is that it just won’t happen. Although it works for republicans it won’t work with dems. Republicans are homogeneous group: white men. Dems are just too diverse to do this effectively
All of this.
The most important part that you highlight is that the very left is very much like maga. Not in policy or in temperament but in the need to own/control the party. This is fundamentally what the freedom caucus is doing. They couldn’t give a shit about policy. They just want to control the Republican Party. I’m not even sure they want their insane policies to pass.
On media, it’s definitely about the voters and integrity but it’s also that Dems, in general, want to bitch about Dems. This doesn’t mean they like maga but they definitely don’t care as much about just hating on maga. If you look at the most leftist media (the nation or pod save America…maybe not leftist but partisan), they complain all the time about Dems.
Matt fails in my mind because he HAS to be be contrarian. He physically can’t agree with a policy of either party, even if he likes it. He will always complain about something (specifics, tradeoffs, or process). That’s why it’s hard to take him seriously on “what to do.” He is smart and thoughtful but just awful politically
Isn't the most likely explanation for the tightening of the polls the fallout and coalescing of voters after RFK Jr. dropped out?
There are two-way polls that only ask about about Harris and Trump. I've not seen that these polls tightened less since RFK Jr.'s dropout/endorsement than more open polls.
Have you observed that?
I'm not that sophisticated--I really meant it as a question, not a statement of fact. If you've looked at it and seen the two-way polls and open polls track in lockstep, that likely suggests the tightening is more reversion to the mean than event-driven. Though I'd still think the presence of an active 3rd party candidate in the race influences the two-way poll results, as does RFK Jr.'s endorsement of Trump on his way out the door?
Looking at Nate Silver's polling averages (I'm eyeballing this, not doing any detailed analysis): On the day RFK Jr. dropped out, the polls were Harris 48.0% (in the midst of her convention bump), Trump 43.7%, Kennedy 3.9%. Today, they show Harris 48.7% (+0.7%, though down from her peak by ~0.5%), Trump 46.8% (+3.1%). That looks to my untrained eye like it could plausibly be driven largely by RFK Jr. voters coming off the two-way sidelines and skewing Trump.
Body language said it all. She walked across the stage to shake his hand; he did everything but hide behind his podium. She spoke with passion; he was pinch-faced and angry. All the white eye makeup in the world couldn't make his eyes seem open.
I think it’s inaccurate and shortsighted to say that all of those 55% that say they want major change from the Biden policy agenda are inherently saying they won’t vote for her, or even wouldn’t have voted for Biden. I would love to see major changes from many of his policies AND i would’ve voted for him and will vote for her because they are the lesser of two evils.
It was bizarre that during the question on Israel, Trump pivoted to Ukraine and Russia. Israel is a good issue for Trump as Democrats are deeply divided on the issue, while Republicans are not. Ukraine is the opposite, where Democrats are united but Republicans are divided. So why take precious time on the Israel question to talk about Ukraine?
Per the discussion of Nate's forecast, y'all forgot to factor in "response bias" into consideration. After the "trauma" of the "will he / wont he?" July, there was no doubt a "response bias" in exactly who was answering the phone and responding to the polls themselves such that Democrats were more likely to answer the phone and take a call and Republicans were less likely due to the dismay of Trump's struggles post Biden leaving the race.
I would suggest some of the "slip" in polls picked up in the forecast was associated with a mean reversion in this bias as excitement tempered a bit post Convention... (also probably why Harris's bump was so small historically speaking).
It's debatable whether a response bias correlates with any real shift in actual turnout or not...
ah well it was basically brought up in the wrap up as "factored in" by Brian but i am not sure how it could be factored in if we spent the first 90% of the podcast looking for a reason that the mean reverted without bringing up response effects post swap / convention...