site banner

Culture War Roundup for the week of November 4, 2024

This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.

Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.

We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:

  • Shaming.

  • Attempting to 'build consensus' or enforce ideological conformity.

  • Making sweeping generalizations to vilify a group you dislike.

  • Recruiting for a cause.

  • Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.

In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:

  • Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.

  • Be as precise and charitable as you can. Don't paraphrase unflatteringly.

  • Don't imply that someone said something they did not say, even if you think it follows from what they said.

  • Write like everyone is reading and you want them to be included in the discussion.

On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.

8
Jump in the discussion.

No email address required.

On Prognostication

Over the past several weeks, I've become increasingly irritated by discussions, both here and elsewhere, involving election predictions. While I agree that speculation can be fun, I think too many people try to read too much into the day-to-day ups and downs of the election cycle. While I agree with Nate Silver on a lot of things, there's something I find inherently off-putting about his schtick. I read The Signal and the Noise around the time it hit the bestseller lists and had an addendum about the 2012 election. One of the themes of the book is that the so-called experts who make predictions on television don't base their predictions on rational evidence and don't face any consequences when their predictions fail. No in-studio commentator on The NFL Today is losing his job solely because he picked too many losers.

Around this time, I became interested in probabilities, and I was regularly hitting up a friend who had majored in math and was pursuing a doctorate in economics at Ohio State. At one point he told me "Probability is interesting, but when it comes down to it, the only thing it's good for is gambling. We say there's a 50/50 chance of drawing a black ball from an urn when we know that the urn has 50 black balls and 50 white balls. When we talk about probabilities in the real world, it's like talking about the chances of drawing a black ball from an urn we don't know the size of." When discussing cards or dice, we're discussing random events based on repeatable starting conditions. When discussing elections, we're discussing a non-random event that will only happen once.

Beyond that, though, the broader question is: What's the point of all of this? This isn't a football game where scoring points confers an obvious advantage. If Trump is up by 5 points in June or Harris is up by 5 points in July, it has absolutely no effect on the actual election. My irritation with this started a couple weeks ago when someone posted here about Trump having large odds of winning on some betting site. I mean, okay, but who cares? What am I supposed to do with this information? I guess it's marginally useful if I'm thinking of putting a little money on the line, but I'm not much of a gambler, and the poster wasn't sharing this information to spark discussion on good betting opportunities. I pretty much lost it, though, last weekend, when news of the Selzer poll showing Harris winning Iowa hit and had everyone speculating whether Ms. Selzer was a canary in a coal mine or hopelessly off. Again, who cares? Selzer's prediction may be correct, or it may be incorrect, but it has no bearing on the actual election. Harris doesn't get any extra votes because Selzer shows her doing better than ABC or whoever. Trump doesn't get any extra votes because of his odds on PredictIt.

I will admit that polling is useful to campaigns trying to allocate resources and determine what works and what doesn't. But they have their own internal polling for that. But unless you're actively employed by a campaign, there's nothing you can do with this information. As much as arguing about politics in general may be an exercise in futility, there's at least some chance you can influence someone else's position. Arguing about who's going to win the election doesn't even go this far, since no one is arguing that you should vote based on polling averages. The only utility I see in any of this is entertainment for the small subset of people who find politics entertaining. Which brings me back to my original criticism of Silver: The reason these professional prognosticators don't get called out on their inaccuracies is because their employers understand that their predictions are ultimately meaningless. Terry Bradshaw may predict the Browns to beat the Bengals, but at a certain time we'll know the winner and if the Bengals win the sun will rise the next morning and his being wrong about it will have no effect on anything.

For the record, a think Harris will probably win, but my prediction is low-confidence and isn't based on anything that's happened since campaign season started. In 2016, a lot of people in swing states voted for Trump because he was an unknown quantity and they preferred taking a chance with him rather than Clinton. In 2020, a certain percentage of these people regretted their decision and voted for Biden. I haven't seen anything in the past four years that suggests that any of these people are moving back to Trump. Electorally, the Republicans haven't shown anything, despite the fact that the first half of the Biden presidency wasn't exactly a cakewalk. But that's just my opinion. I don't know what you're supposed to do with it. You can disagree with it, and you may have a point, but after tomorrow what I think and what you think won't matter. The votes will be counted, a winner declared, and Dr. Oz's midterm performance won't matter, and my being wrong about it won't have an effect on anything.

I also think it's important to remember the lesson Scott tried to drive home here, that absolutely no one heeded: https://slatestarcodex.com/2016/11/07/tuesday-shouldnt-change-the-narrative/

Back in 2016, I believe Scott was right, and yet, immediately EVERYONE's narrative shifted to "Trump beat Hilary because of <whatever reason: racism, sexism, Trump was better then her, people liked him, people disliked her, etc>". I think Scott's lesson is probably right here as well, but already no one is acting that way. On this very thread, we have many people saying Harris's win is inevitable because leftists control institutions, or Trump's win is inevitable because Harris is less likable or more stupid than Biden or Hilary ever were. Once one of them wins tomorrow, everyone will be frantically updating their priors accordingly, in order to make their world make sense. But should they really, or are they just overfitting to noise?

Oh man, thanks for finding that link! It's what I was thinking about but couldn't find in this comment a couple days ago.

One reason polling and prediction markets pre-election are interesting is that you can see reactions to the politics being preformed before our eyes. When Trump does the McDonald's thing the context of whether this hurt or helped him in the polls is actually very interesting and says something about the American voter. When dems drop Biden and put up Kamala the trajectory of the polls is interesting. Who 'won' the debate does appear to have an answer and the poll swing shows it.

My view on predictions of political races is the opposite--I think they are great and useful--precisely because I don't really care about politics but I do care about polls (and statistics in general). Predictions of political races are a way to test the poll's methodology.

For example, Gallup is but one of many companies whose business is to poll US adults on various questions of interest--say, what percent of US adults identify as LGBT. That's a reasonably interesting question, judging by The Motte's interest in the subject. Also, businesses may want to know how big the group is, if they are considering catering to it.

So the Gallup's poll says that 7.1% of US Adults identify as some flavor of LGBT. But how well does that reflect reality? Gallup provides a snippet of their survey method at the end--surveyed over 12,000 adults by phone (70% cell, 30% landline)--and they give that standard phrase familiar to anyone who took an introductory Statistics course:

For results based on the total sample of national adults, the margin of sampling error is ±1 percentage point at the 95% confidence level.

So they are saying that their result is likely within one percentage point of reality... except that this nice quantitative statement only accounts for sampling variability, and doesn't even try to estimate the systematic bias of their methods.

For example, for many decades now there has been a huge drop in the proportion of people who pick up their phone when a rando calls them. Two decades ago, when I was teaching intro stats and Gallup still published their non-response rate, it was a measly 5%. Now? It's so bad that most respectable polling companies have dropped randomized calling altogether, and they have switched to recruiting people into panels--like, recruit 100,000 US adults who will have your company's phone number in their caller ID, and so would be more likely to pick up the phone. Then the response rate goes up to like 20%-30%.

But how representative are those panels? Why should you trust that they produce polls that are anywhere close to reality? The one great way to test it is if there is a census coming up, and the poll tries to predict the outcome of that census. Well, that's what an election is--a census of the voters.

To be clear, I'm not trying to suggest that polling is useless, or that it shouldn't be done. I just don't think we should put to much stock into polls insofar as they concern the average voter.

Btw: Many countries ban polling before an election:

https://archive.is/9P5VO

Yeah, I think I'll go strongly the other direction on this.

"Terry Bradshaw may predict the Browns to beat the Bengals, but at a certain time we'll know the winner and if the Bengals win the sun will rise the next morning and his being wrong about it will have no effect on anything."

The problem is, this isn't true at all. It's not true in football, it's not true politics, it's not true across many dimensions of life. And the world we live in is worse off for it.

Terry Bradshaw is one guy out of many many people giving opinions, I wouldn't say that his opinion alone is the basis that people's futures ride on, but when Brandon Staley goes for it on 4th down, and Terry Bradshaw says that he's the reason the Chargers lost (everytime that happens, a guys livelihood is on the line), when the Rams lose a game and Terry Bradshaw says the Jared Goff is the reason why they lost (a guys livelihood is on the line). (I don't know if Bradshaw actually had those particular take, I made them up as example, though I do remember various talking heads making them, just not which particular talking head).

Every little statement like this affects public perception, and public perception affects reality. We humans are highly susceptible to group think.

The presumption is that that the guy on TV knows what he's talking about, knows the factors that goes into whether the Browns have what it takes to beat the Bengals [1], I agree that for the most part no one really cares, I'm saying that we should, if in reality, these guys are actually just full of shit constantly. That's actually extremely useful information.

This obviously applies to politics as well, no matter what happens on Tues, the Wednesday morning QBs will come out, it's extremely useful to understand that most of them are full of shit.

[1] Somewhat hilariously, these picks typically aren't even against the spread, to the degree that these guys can't even figure out to just pick the obvious favorites... truly wasting all of our time.

Terry Bradshaw is one guy out of many many people giving opinions, I wouldn't say that his opinion alone is the basis that people's futures ride on, but when Brandon Staley goes for it on 4th down, and Terry Bradshaw says that he's the reason the Chargers lost (everytime that happens, a guys livelihood is on the line), when the Rams lose a game and Terry Bradshaw says the Jared Goff is the reason why they lost (a guys livelihood is on the line).

I don't see what this has to do with anything I said. I'm talking about predictions, not post-mortem analysis.

The ability to make falsifiable predictions is how we know we understand things about the world we live in.

Not understanding the world we live in has real consequences.

I’m arguing for more discourse around predictions, in response to your argument for less discourse around predictions.

I suspect we agree that the discourse could be much better than it actually is.

TLDR, I actually like Nate Silver’s schtick quite a bit, and wish we had more people trying to do it across more areas of interest.

I think it has some use if you’re in a position to be doing long term planning either because the campaign is promising things that will help or harm your group, or because the change in regulations promised will affect a business project. If Trump is going to end a subsidy for solar panels then maybe the fact that he’s way up in the polling would change how you do business. Or maybe you’re trans and worried about your rights in your home state. Or you’re gay and looking to adopt. There are some people who for various reasons need to plan based on who wins in 2024.

For the vast majority of people, the winner of any election will only affect you on the margins. It’s just human history. Unless a regime was specifically looking to harm a group of people, for the most part, life goes on with very little change. If you woke up tomorrow and you were under a Maoist dictatorship, unless you owned a big corporation or challenged the regime in some way, you could live very comfortably. The basics of life don’t change that much. There will still be jobs, schools, sporting events, and people will still get drunk on weekends.

On a related note, but different domain, I saw these two threads this morning (...maybe NOAA's advertising budget has been kicked into gear because they have some bureaucratic fight going on...). To the extent that people have some form of equities at risk, accurate probabilistic models can help people properly allocate their assets/risk... and may even allow some folks to make a boatload of money in trying to allocate appropriately at a high level of abstraction. I recall listening to a podcast with a guy who made literal billions of dollars by making huge bets on insurance markets in one of the hurricane seasons not long after Katrina (I can't remember his name off the top of my head now, but IIRC, it was someone who people would plausibly recognize). He did so by trying to have the most accurate weather prediction possible.

Political outcomes feel a bit less directly-related to outcomes than the much more direct hurricane-to-damage relation, because "probability of president" probably needs to be mixed with "probability of Congress" and "conditional probabilities of those results ending up with the gov't taking Action A". Right now, there is still debate on whether we can do any part of that chain, which makes it difficult to intuitively feel a connection.

What's the point of all of this? This isn't a football game where scoring points confers an obvious advantage. If Trump is up by 5 points in June or Harris is up by 5 points in July, it has absolutely no effect on the actual election.

No, it's a referendum on whose tribe is larger, better organized, and has seized the moral high-ground of the culture. Pumping up your own supporters through exhortation (we're winning - one more great surge of effort, comrades, and we shall triumph!) or demoralizing the enemy through propaganda (your puny candidate is weak and is failing! Abandon hope and don't waste your efforts and resources in what is obviously a doomed cause!) are valid tactics.

It's an election, though, not a bloody melee. In a melee you want your side to believe you've got the enemy grossly outnumbered and so there's definitely no need for any of Us to break and rout before we force Them to. In an election you want your side to believe you're tied with the enemy and so there's definitely a need for Us to get our lazy butts off the couches rather than either conceding to Them or letting our overconfidence cause an upset.

Hopefully we're just doing the exhortation thing because our psychology evolved through far less history with elections than with bloody melees (100s of years vs 100s of millennia?), not because we've had a few slightly-bloody melees recently and we want to be prepared for when they get much bloodier.

Well, you want your side to believe it is close and while you are winning if you don’t do something we’ll lose. People like a winner so gives them good feelings there but you don’t want complacency.

It's an election, though, not a bloody melee.

Elections are, in no small part, symbolic melees. We do things memetically and through balloting as a slightly-more-symbolic version of the "stand on opposing hill-tops outside of spear range and ineffectually chuck things at each other while yelling to establish bravery and dominance" paradigm that prevailed throughout most of human history, and a much-more-symbolic version of the "line up in two opposing phalanxes and grind against each other until someone breaks and runs away" that got more common in Europe during recorded history.

Either way they're still contests with dominance and submission on the line; we've just decided to accept proxy battles for the real thing.

Hopefully we're just doing the exhortation thing because our psychology evolved through far less history with elections than with bloody melees (100s of years vs 100s of millennia?)

Exactly.

The way I've heard it is that an election is a simulated civil war. Both teams show up, you count up the members, assume the larger side would win, then everybody accepts the results and goes home because actually fighting the war would involve lots of unnecessary death and destruction.

Problem #1: Because of median voter theorem, both sides converge to political positions that are almost equally appealing to the center, meaning each side is only winning elections by a few percentage points of the population. This makes it much more likely that the losing side could win the war than e.g. a 30%-70% split, and makes the incentive for the loser to flip the table much larger.

Problem #2: When democracy was first implemented, the franchise was limited to white landowning men. Each expansion of the franchise to people who make for worse soldiers and workers (women, blacks, etc.) has represented a decoupling between the number of voters on each side and that side's actual military and productive capacity, meaning it is no longer at all obvious that the smaller side would lose a civil war. As the sides specialize to appeal to different demographics, the incentive for the party of white men to realize that they are getting screwed for no good reason and start shooting is, again, greatly increased.

(The elites know this, which is why they spend so much time and effort preventing white men from developing class consciousness.)

Probability and stats have always been my favourite parts of mathematics and I have zero interest in gambling (I went to Vegas once, lost $50, won $30, and felt vaguely annoyed and bored and went to drink cocktails in the hotel pool and haven’t gambled since).

The reason is simply that I think intelligence basically just is predictive ability. This operates at different levels, of course. A smart goalkeeper will be able to guess which way a penalty-taker will shoot, and a smart driver will predict that the car ahead will change lanes suddenly. By contrast, a smart lawyer might give you advice on how to avoid getting sued and a smart accountant on how to avoid paying as much tax. A smart scientist may identify a novel experimental technique for testing a theorem. A smart philosopher might give advice which if followed will reliably allow you to live a more virtuous or contented life. In all these cases, we’re identifying the threads of cause and effect and making claims about how one informs the other.

This is why when I hear people say LLMs are “just next token predictors” it doesn’t tell us much about their psychological capacities, because we’re next token predictors too (or more exactly: prediction error minimisers). That’s not an uncontroversial claim, of course, but it’s also not an outrageous one in the context of contemporary cognitive science; the predictive coding/free energy frameworks of people like Andy Clark and Karl Friston are rapidly becoming the consensus or at least plurality view on questions of ur-principles of cognition.

But why should we care about predicting the US election specifically? I think because it’s a big meaty complex problem that is amenable to insights from a variety of different methods and life experiences. Sure we can take the Nate Silver approach and read daily polls and build models, but you can also glean insights from just talking to people at the bar or from sampling the vibe in your niche industry or from developing your world-historical theories or a million other methods. Nate Silver will probably beat you on average, but it’s not crazy to think that for a general election these kinds of interactions and experiences might give you a useful insight. That makes election punditry an unusually inclusive form of prognostication — you can debate it with your mom or your Uber driver or your babysitter. In that sense, it’s a very American sideshow to the very American event.

"Probability is interesting, but when it comes down to it, the only thing it's good for is gambling"

I mean, yeah. But on a fundamental level, just about every decision we make is a gamble of some sort. Any significant choice in life is boils down to a decision to expend some amount of time/money/energy with the hope of reaping some reward. Any information that helps you make these decisions more clearly is therefore extremely useful.

I didn't mean to suggest I was doubting that. But I can't conceive of how any election prediction would have any effect on anything I do, in any circumstance.

While I agree with Nate Silver on a lot of things, there's something I find inherently off-putting about his schtick. I read The Signal and the Noise around the time it hit the bestseller lists and had an addendum about the 2012 election. One of the themes of the book is that the so-called experts who make predictions on television don't base their predictions on rational evidence and don't face any consequences when their predictions fail. No in-studio commentator on The NFL Today is losing his job solely because he picked too many losers.

And that's why you listen to Nate Silver and super-forecasters who have a public track record of being better than the average when it comes to their predictions, often markedly so.

I'd say the state of affairs is better than it could be, given that Nate has modest internet celebrity status. Making good models and publicizing them has some payoff.

At one point he told me "Probability is interesting, but when it comes down to it, the only thing it's good for is gambling. We say there's a 50/50 chance of drawing a black ball from an urn when we know that the urn has 50 black balls and 50 white balls. When we talk about probabilities in the real world, it's like talking about the chances of drawing a black ball from an urn we don't know the size of." When discussing cards or dice, we're discussing random events based on repeatable starting conditions. When discussing elections, we're discussing a non-random event that will only happen once.

Downstream, I was mustering a defense of Bayesian reasoning, but you don't have to be anywhere near as formal to address this. We are lucky enough to be in a universe that doesn't exist in perfect uncertainty, things that have happened inform what will happen, and some humans are somehow able to do better than random at answering unresolved questions, with better modeling, be it explicit or intuition based. You can notice them, or pay attention to the gestalt impression of a crowd squeezing wisdom from attempts to win either Reputation Points or actual money out of their competition.

Knowing the odds has some utility, and should be of some use. You wouldn't throw up your hands if some rando used the argument your friend used to claim that Randomly Picked American X had 50% odds of winning the presidential election. If you were sensible and he was willing to put money where his mouth is, you'd take said bet and make your money back expecting to win that bet several million times per loss.

Odds convey information, or at least expectation. The sheer complexity of the universe and the vast sampling space is, thankfully, not as much of an impediment as it seems. I'd expect so, after all we evolved in it and had to understand our world and other actors better than a rock.

I was considering writing a post on this exact topic, but I refrained, because I wasn't sure how to press my vague discontent into a substantive thesis.

I'll be happy when we can go back to weekly threads that don't contain the term "prediction market". I've always found the LW/SSC fascination with monetary betting to be somewhat crass. (I'd be inclined to speculate that "make all your beliefs pay rent" is only a few steps away from "make all people pay rent".)

It's sometimes claimed that intelligence just is pattern matching. So, trying to successfully predict a major one-off event like a presidential election could then be seen as the ultimate IQ test, the ultimate test of one's ability to reason in an uncontrolled and data-scarce environment. And I can see why that would be an intriguing challenge to take on. But pattern matching has its limits, and ultimately it just feels obvious to me that there's not much that can be said in terms of election forecasting besides "well, we'll see when we see".

The point of estimating the election winner is that the election can be quite impactful (for regulator policies, culture war stuff, gambling like you mentioned, etc). If e.g. you work in a financial institution and you prefer one candidate over the other, knowing whether they have a 90% chance of winning or a 50-50 chance can make a huge difference. It's also a good chance to test Bayesian reasoning capabilities, as there's a correct answer at the end that you can check your work against (and the work of those you follow).

You're correct that most prognosticating won't have a material impact on the result, but that's a non-sequitur since that's not what people are trying to do when they're predicting who will win.

No one is changing internal policies based on polling predictions on who is going to win the next election. Even if Candidate X is posing sweeping regulatory changes to your particular industry, you're not going to start changing your policies because Candidate X is favored to win. You're not even going to start changing your policies after Candidate X wins. You're going to start changing your policies when the new regulations are actually enacted, because a lot can happen between proposals and final legislation.

If new regulations are coming down the pipe, they can make certain products or even entire sectors less desirable. The issue is that this can be priced-in by competitors long before the regulations actually hit, which can hurt the ability of slow-movers to reposition. If you think e.g. oil companies were going to get hit by a bunch of new environmental regulations, institutions might want to offload their stakes before their competitors to get the best price. A good real-world example of this is what happened to ESG funds. They used to be a hot thing during peak woke, but the industry really started backing off once Republicans started making a stink. I'm not sure how much actual regulation happened, as just the threat of it was enough to strangle the baby in the cradle.

Knowing if there's a 50% chance or a 90% chance that these things will happen absolutely makes a big difference. I was at a gathering of many of the chief risk officers of big financial institutions, and the election was easily the most-discussed topic in terms of worrying potential outcomes.

I mean, one-off near 50-50 gambles are just objectively quite difficult to evaluate properly. So unless you screw up in an obvious fashion, there are very few good ways to call someone out to begin with.

Ironically, Nate had a good article very recently on at least one way: a particular pollster, Redfield & Wilston, had released a poll about the seven "battleground states", and ALL results are within a 1% margin of each other. It's easy to show that this is statistically speaking pretty much impossible to happen given the base variance from the low-ish number of people they've polled. So they have to have fucked up somehow, and Nate can even explain the most likely technique, called herding, and how it is working. Also ironically, it's mostly used to avoid reputational damage, since it works by pooling results from different sources to avoid extreme results that naturally arise from base variance, but which are then called out as wrong afterwards.