It’s interesting to me that the thing Trump is highlighting as his concern—AI driven censorship of ‘misinformation’, and perhaps also all the discrimination language—are not from our point of view central to the purpose of the executive order. Given that Trump does seem concerned about AI—he referred to it unprompted as “maybe the most dangerous thing out there of anything”—it would be a shame if the inclusion of more partisan elements in the executive order lead to its repeal. But perhaps the Biden administration wouldn’t have been interested in signing the executive order in the first place without those elements.
In a new interview Trump again brought up the risk of artificial intelligence, discussed the connection to nuclear risks, and mentioned that some people think it will take over the human race.
I’ve never seen Biden directly discuss AI, but during the debate he claimed climate change was the only existential threat facing humanity, implying that he did not consider AI, bio or nuclear war to be existential threats:
If we reach 1.5 degrees Celsius at any one point there’s no way back, the only existential threat to humanity is climate change.
There is still low-hanging fruit. Estimates for how effectively top RCT-tested interventions to generate net swing-state votes this election range from a few hundred to several thousand dollars per vote. Top non-RCT-able interventions are likely even better.Many potentially useful strategies have not been sufficiently explored.
This seems quite surprising given the sums and stakes involved. Do you have sources for this?
Biden won the last election by 42,918 combined votes in three swing states. Trump won the election before that by 77,744 votes. 537 votes in Florida decided the 2000 election.
along with:
generate net swing-state votes this election range from a few hundred to several thousand dollars per vote
You get that the entire election outcome can be purchased for ~$13M - $600M.
But I’m assuming it isn’t actually isn’t that cheap to swing the election?
Well, it implies you could change the election with those amounts if you knew exactly how close the election would be in each state and spent optimally. But If you figure the estimates are off by an OOM, and half of your spending goes to states that turn out not to be useful (which matches a ~30 min analysis I did a few months ago), and you have significant diminishing returns such that $10M-$100M is 3x less impactful than the first $10M and $100M-$1B is another 10x less impactful, you still get:
First $10M is ~$10k per key vote = 1,000 votes (enough to swing 2000)
Next $90M is ~$30k per key vote = 3,000 votes
Next $900M is ~$90k per key vote = 10,000 votes
I think if you think there’s a major difference between the candidates, you might put a value on the election in the billions—let’s say $10B for the sake of calculation; so the first $10M would be worth it if there’s a 0.1% chance the election is decided by <1000 votes (which of course happened 6 elections ago!), the next $90M is worth it if there’s a 0.9% chance the election is decided by >1000 but <4000 votes, and the next $900M is worth it if there’s a 9% chance the election is decided by >4000 but <14000 votes. IMO the first two probably pass and the last one probably doesn’t, but idk.
I think if you think there’s a major difference between the candidates, you might put a value on the election in the billions—let’s say $10B for the sake of calculation.
You don’t need to think there’s a major difference between the candidates to conclude that the election of one candidate adds billions in value. The size of the US discretionary budget over the next four years is roughly three orders of magnitude your $10B figure, and a president can have an impact of the sort EAs care about in ways that go beyond influencing the budget, such as regulating AI, setting immigration policy, eroding government institutions and waging war.
Yes, but it’s kind of incoherent to talk about the dollar value of something without having a budget and an opportunity cost; it has to be your willingness-to-pay, not some dollar value in the abstract. Like, it’s not the case that the EA funding community would pay $500B even for huge wins like malaria eradication, end to factory farming, robust AI alignment solution, etc, because it’s impossible: we don’t have $500B.
And I haven’t thought about this much but it seems like we also wouldn’t pay, say, $500M for a 1-in-1000 chance for a “$500B win” because unless you’re defining “$500B win” with respect to your actual willingness-to-pay, you might wind up with many opportunities to take these kinds of moonshots and quickly run out of money. The dollar size of the win still has to ultimately account for your budget.
I think the core issue is that the lottery wins you government dollars, which you can’t actually spend freely. Government dollars are simply worth less, to Pablo, than Pablo’s personal dollars. One way to see this is that if Pablo could spend the government dollars on the other moonshot opportunities, then it would be fine that he’s losing his own money.
So we should stipulate that after calculating abstract dollar values, you have to convert them, by some exchange rate, to personal dollars. The exchange rate simply depends on how much better the opportunities are for personal spending, versus spending government money.
The fact that opportunities can get larger than your budget size seems not to be the core issue for the reason that you mention—that at realistic sizes of opportunity, it is possible to instead buy a lottery for a chance at the opportunity instead.
Yeah I roll to disbelieve too. One of my quantitative takeaways from Andrew Gelman’s modelling of the 2020 elections was that very few states (4/50, in particular New Hampshire, Pennsylvania, Wisconsin, and Michigan) were modelled as close enough that p(one vote changes outcome) > 1 in 10 million; New Hampshire tops the list at 1 in 8M. Optimistically assuming $100 per voter that’s still nearly a billion dollars at the very low end; a more realistic estimate would probably be ~1 OOM higher. Probably some sort of nonlinearity kicks in at this scale, or the most cost-effective tactics to sway voters cap out at relatively low levels for whatever reason?
On the flip side, I’m reminded of Scott’s essay Too much dark money in almonds? which provides an intuition pump for why it might be the case that it’s not as expensive as you may expect to swing the election:
Everyone always talks about how much money there is in politics. This is the wrong framing. The right framing is Ansolabehere et al’s: why is there so little money in politics? But Ansolabehere focuses on elections, and the mystery is wider than that. …
(in case you’re keeping track: all donations to all candidates, all lobbying, all think tanks, all advocacy organizations, the Washington Post, Vox, Mic, Mashable, Gawker, and Tumblr, combined, are still worth a little bit less than the almond industry. And Musk could buy them all.) …
In this model, the difference between politics and almonds is that if you spend $2 on almonds, you get $2 worth of almonds. In politics, if you spend $2 on Bernie Sanders, you get nothing, unless millions of other people also spend their $2 on him. People are great at spending money on direct consumption goods, and terrible at spending money on coordination problems.
(I don’t really have an opinion either way on whether more or less money should be spent on this)
I don’t think Gelman’s models are correct here. But even if they were, the numbers definitely don’t scale linearly. The vote margin ex post in the last 6 elections were all under 1M, the most recent 2 elections were under 100k. You’ll need pretty implausible assumptions (or really bad targetting) before you can fit an expectation of 10M ex ante votes to the observed ex post distribution.
[Caveat that I don’t know anything else about this]
I recall Rob Wiblin’s 80K article on voting referencing this summary table from the 2015 edition of Get Out The Vote claiming “$30-100 or a few hours of work as a volunteer” to “persuade one stranger to vote for your preferred candidate”, a lot lower than the OP’s claimed figures, and that even adjusting upwards for various factors doesn’t worsen this by more than an OOM.
(That said, I regard these figures basically the same way GW treats the best cost-effectiveness estimates in the DCP2/3)
This number is crazy low. It seems bad to make a Cause Area post on the forum that entirely rests on implausibly low numbers taken from some proprietary data that can’t be shared. You should at least share where you got this data and why we should believe it.
Many arguments about the election’s tractability don’t hinge on the impact of donations.
Donating is not the only way to contribute to the election. Here is a public page showing the results of a meta-analysis on the effectiveness of different uses of time to increase turnout (though the number used to estimate the cost-effectiveness of fundraising is not sourced here). The analysis itself is restricted, but people can apply to request access.
Polling and historical data suggest this election has a good chance of being won by thousands to hundreds of thousands of swing-state votes. That means any intervention that can swing thousands of votes (or maybe hundreds) has a meaningful chance of swinging the election. I discussed some potential interventions in the post.
There is recent evidence that suggests there are events that quickly cause a large portion of voters to change their mind about who to vote for. Nate Silver wrote “The impact of Comey’s letter is comparatively easy to quantify, by contrast. At a maximum, it might have shifted the race by 3 or 4 percentage points toward Donald Trump, swinging Michigan, Pennsylvania, Wisconsin and Florida to him, perhaps along with North Carolina and Arizona. At a minimum, its impact might have been only a percentage point or so. Still, because Clinton lost Michigan, Pennsylvania and Wisconsin by less than 1 point, the letter was probably enough to change the outcome of the Electoral College.” (See full article for more details/analysis). That said, others disagree about the effect of Comey’s email + media reporting.
Other heuristics mentioned in the post (like the reasons for campaign-related work being undesirable).
The above numbers are based on RCTs and information shared with me by multiple organizations. I’m sorry I’m unable to share more details publicly, I’m respecting the preferences of these organizations.
Ok. Sorry about the tone of the last response, that came off more rude than I would have liked. I do find it unsettling or norm-breaking to withhold information like this, but I guess you have to do what they allow you to do. I remain skeptical.
This seems reasonable to be, though we should factor in the risks that come with being seen to influence politics. I think it makes sense for individual EAs to get involved as opposed to EA orgs getting involved.
I think there’s some sense to this, but at the same time the two biggest public EA donors were also major public Democratic donors (and uh, one of them is in jail), so I somewhat wonder if this ship has sailed by now.
If Trump still thinks AI is “maybe the most dangerous thing” I would be wary of giving up on chances to leverage his support on AI safety.
In 2022, individual EAs stood for elected positions within each major party. I understand there are Horizon fellows with both Democrat and Republican affiliations.
If EAs can engage with both parties in those ways, added to the fact the presumptive Republican nominee may be sympathetic, I wouldn’t give up on Republican support for AI safety yet.
100%, I’d like to see the stats on what Politicians say they will repeal pre-elections, and what they actually end up repealing once they are in power. In New Zealand here at least I can think of multiple anecdotal examples where there is a lot of bluster pre election but then the law either doesn’t get changed, or only modified in a minor way;
Perhaps Obamacare might be one example of this in America? I think Trump had a decent amount of rhetoric saying he would repeal it, then didn’t do anything repeal it when he reached power.
Perhaps Obamacare might be one example of this in America? I think Trump had a decent amount of rhetoric saying he would repeal it, then didn’t do anything when he reached power.
My recollection was that Trump spent quite a lot of effort trying to repeal Obamacare, but in the end didn’t get the votes he needed in the Senate. Still, I think your point that actual legislation often looks different from campaign promises is a good one.
I love how clear and concise this argument is nice job! I don’t know enough about elections (and especially the American one) to judge the quality of the argument.
One interesting thing I learnt was that Trump plans to repeal the AI order—it surprises me a little as I would have thought his protectionist mindset might have been inclined towards protecting people and jobs from future AI threats—the free speech argument feels weak to me.
I find myself quite skeptical of this analysis following the dramatically failed predictions (and more direct calls to action) regarding the tractability of the Carrick Flynn campaign in 2022, which now seems like a major blunder. If anything I think there’s a stronger case for that sort of thing than there is for national presidential elections...
I think the 2nd place result for Carrick is quite good for a 1st-time candidate with 1st-time political action team behind. There were many mistakes obviously, but deciding to run was not one of them IMO. No political action will result in certainty, the goal is ~always to move the needle or take a bunch of swings.
Yes, to be clear I’m not criticizing the initial decision to run but rather the dubious impact estimates and calls to action towards the end of that campaign.
I think it’s good to critically interrogate this kind of analysis. I don’t want to discourage that. But as someone who publicly expressed skepticism about Flynn’s chances, I think there are several differences that mean it warrants closer consideration. The polls are much closer for this race, Biden is well known and experienced at winning campaigns, and the differences between the candidates in this race seem much larger. Based on that it at least seems a lot more reasonable to think Biden could win and that it will be a close race worth spending some effort on.
It’s interesting to me that the thing Trump is highlighting as his concern—AI driven censorship of ‘misinformation’, and perhaps also all the discrimination language—are not from our point of view central to the purpose of the executive order. Given that Trump does seem concerned about AI—he referred to it unprompted as “maybe the most dangerous thing out there of anything”—it would be a shame if the inclusion of more partisan elements in the executive order lead to its repeal. But perhaps the Biden administration wouldn’t have been interested in signing the executive order in the first place without those elements.
In a new interview Trump again brought up the risk of artificial intelligence, discussed the connection to nuclear risks, and mentioned that some people think it will take over the human race.
I’ve never seen Biden directly discuss AI, but during the debate he claimed climate change was the only existential threat facing humanity, implying that he did not consider AI, bio or nuclear war to be existential threats:
This seems quite surprising given the sums and stakes involved.
Do you have sources for this?
This is my biggest confusion as well.
If you take:
along with:
You get that the entire election outcome can be purchased for ~$13M - $600M.
But I’m assuming it isn’t actually isn’t that cheap to swing the election?
Well, it implies you could change the election with those amounts if you knew exactly how close the election would be in each state and spent optimally. But If you figure the estimates are off by an OOM, and half of your spending goes to states that turn out not to be useful (which matches a ~30 min analysis I did a few months ago), and you have significant diminishing returns such that $10M-$100M is 3x less impactful than the first $10M and $100M-$1B is another 10x less impactful, you still get:
First $10M is ~$10k per key vote = 1,000 votes (enough to swing 2000)
Next $90M is ~$30k per key vote = 3,000 votes
Next $900M is ~$90k per key vote = 10,000 votes
I think if you think there’s a major difference between the candidates, you might put a value on the election in the billions—let’s say $10B for the sake of calculation; so the first $10M would be worth it if there’s a 0.1% chance the election is decided by <1000 votes (which of course happened 6 elections ago!), the next $90M is worth it if there’s a 0.9% chance the election is decided by >1000 but <4000 votes, and the next $900M is worth it if there’s a 9% chance the election is decided by >4000 but <14000 votes. IMO the first two probably pass and the last one probably doesn’t, but idk.
You don’t need to think there’s a major difference between the candidates to conclude that the election of one candidate adds billions in value. The size of the US discretionary budget over the next four years is roughly three orders of magnitude your $10B figure, and a president can have an impact of the sort EAs care about in ways that go beyond influencing the budget, such as regulating AI, setting immigration policy, eroding government institutions and waging war.
Yes, but it’s kind of incoherent to talk about the dollar value of something without having a budget and an opportunity cost; it has to be your willingness-to-pay, not some dollar value in the abstract. Like, it’s not the case that the EA funding community would pay $500B even for huge wins like malaria eradication, end to factory farming, robust AI alignment solution, etc, because it’s impossible: we don’t have $500B.
And I haven’t thought about this much but it seems like we also wouldn’t pay, say, $500M for a 1-in-1000 chance for a “$500B win” because unless you’re defining “$500B win” with respect to your actual willingness-to-pay, you might wind up with many opportunities to take these kinds of moonshots and quickly run out of money. The dollar size of the win still has to ultimately account for your budget.
I think the core issue is that the lottery wins you government dollars, which you can’t actually spend freely. Government dollars are simply worth less, to Pablo, than Pablo’s personal dollars. One way to see this is that if Pablo could spend the government dollars on the other moonshot opportunities, then it would be fine that he’s losing his own money.
So we should stipulate that after calculating abstract dollar values, you have to convert them, by some exchange rate, to personal dollars. The exchange rate simply depends on how much better the opportunities are for personal spending, versus spending government money.
The fact that opportunities can get larger than your budget size seems not to be the core issue for the reason that you mention—that at realistic sizes of opportunity, it is possible to instead buy a lottery for a chance at the opportunity instead.
Yeah I roll to disbelieve too. One of my quantitative takeaways from Andrew Gelman’s modelling of the 2020 elections was that very few states (4/50, in particular New Hampshire, Pennsylvania, Wisconsin, and Michigan) were modelled as close enough that p(one vote changes outcome) > 1 in 10 million; New Hampshire tops the list at 1 in 8M. Optimistically assuming $100 per voter that’s still nearly a billion dollars at the very low end; a more realistic estimate would probably be ~1 OOM higher. Probably some sort of nonlinearity kicks in at this scale, or the most cost-effective tactics to sway voters cap out at relatively low levels for whatever reason?
On the flip side, I’m reminded of Scott’s essay Too much dark money in almonds? which provides an intuition pump for why it might be the case that it’s not as expensive as you may expect to swing the election:
(I don’t really have an opinion either way on whether more or less money should be spent on this)
I don’t think Gelman’s models are correct here. But even if they were, the numbers definitely don’t scale linearly. The vote margin ex post in the last 6 elections were all under 1M, the most recent 2 elections were under 100k. You’ll need pretty implausible assumptions (or really bad targetting) before you can fit an expectation of 10M ex ante votes to the observed ex post distribution.
[Caveat that I don’t know anything else about this]
I recall Rob Wiblin’s 80K article on voting referencing this summary table from the 2015 edition of Get Out The Vote claiming “$30-100 or a few hours of work as a volunteer” to “persuade one stranger to vote for your preferred candidate”, a lot lower than the OP’s claimed figures, and that even adjusting upwards for various factors doesn’t worsen this by more than an OOM.
(That said, I regard these figures basically the same way GW treats the best cost-effectiveness estimates in the DCP2/3)
These can’t be shared publicly, but I’ll DM you.
This number is crazy low. It seems bad to make a Cause Area post on the forum that entirely rests on implausibly low numbers taken from some proprietary data that can’t be shared. You should at least share where you got this data and why we should believe it.
A few quick thoughts:
Many arguments about the election’s tractability don’t hinge on the impact of donations.
Donating is not the only way to contribute to the election. Here is a public page showing the results of a meta-analysis on the effectiveness of different uses of time to increase turnout (though the number used to estimate the cost-effectiveness of fundraising is not sourced here). The analysis itself is restricted, but people can apply to request access.
Polling and historical data suggest this election has a good chance of being won by thousands to hundreds of thousands of swing-state votes. That means any intervention that can swing thousands of votes (or maybe hundreds) has a meaningful chance of swinging the election. I discussed some potential interventions in the post.
There is recent evidence that suggests there are events that quickly cause a large portion of voters to change their mind about who to vote for. Nate Silver wrote “The impact of Comey’s letter is comparatively easy to quantify, by contrast. At a maximum, it might have shifted the race by 3 or 4 percentage points toward Donald Trump, swinging Michigan, Pennsylvania, Wisconsin and Florida to him, perhaps along with North Carolina and Arizona. At a minimum, its impact might have been only a percentage point or so. Still, because Clinton lost Michigan, Pennsylvania and Wisconsin by less than 1 point, the letter was probably enough to change the outcome of the Electoral College.” (See full article for more details/analysis). That said, others disagree about the effect of Comey’s email + media reporting.
Other heuristics mentioned in the post (like the reasons for campaign-related work being undesirable).
The above numbers are based on RCTs and information shared with me by multiple organizations. I’m sorry I’m unable to share more details publicly, I’m respecting the preferences of these organizations.
Ok. Sorry about the tone of the last response, that came off more rude than I would have liked. I do find it unsettling or norm-breaking to withhold information like this, but I guess you have to do what they allow you to do. I remain skeptical.
I don’t think this is norm-breaking for the EA forum or general discourse (though I might still prefer people act differently).
This seems reasonable to be, though we should factor in the risks that come with being seen to influence politics. I think it makes sense for individual EAs to get involved as opposed to EA orgs getting involved.
I think there’s some sense to this, but at the same time the two biggest public EA donors were also major public Democratic donors (and uh, one of them is in jail), so I somewhat wonder if this ship has sailed by now.
… and then I read Advice for Activists from the History of Environmentalism
If Trump still thinks AI is “maybe the most dangerous thing” I would be wary of giving up on chances to leverage his support on AI safety.
In 2022, individual EAs stood for elected positions within each major party. I understand there are Horizon fellows with both Democrat and Republican affiliations.
If EAs can engage with both parties in those ways, added to the fact the presumptive Republican nominee may be sympathetic, I wouldn’t give up on Republican support for AI safety yet.
100%, I’d like to see the stats on what Politicians say they will repeal pre-elections, and what they actually end up repealing once they are in power. In New Zealand here at least I can think of multiple anecdotal examples where there is a lot of bluster pre election but then the law either doesn’t get changed, or only modified in a minor way;
Perhaps Obamacare might be one example of this in America? I think Trump had a decent amount of rhetoric saying he would repeal it, then didn’t
do anythingrepeal it when he reached power.My recollection was that Trump spent quite a lot of effort trying to repeal Obamacare, but in the end didn’t get the votes he needed in the Senate. Still, I think your point that actual legislation often looks different from campaign promises is a good one.
Fair point you may well be right there!
I love how clear and concise this argument is nice job! I don’t know enough about elections (and especially the American one) to judge the quality of the argument.
One interesting thing I learnt was that Trump plans to repeal the AI order—it surprises me a little as I would have thought his protectionist mindset might have been inclined towards protecting people and jobs from future AI threats—the free speech argument feels weak to me.
I find myself quite skeptical of this analysis following the dramatically failed predictions (and more direct calls to action) regarding the tractability of the Carrick Flynn campaign in 2022, which now seems like a major blunder. If anything I think there’s a stronger case for that sort of thing than there is for national presidential elections...
I think the 2nd place result for Carrick is quite good for a 1st-time candidate with 1st-time political action team behind. There were many mistakes obviously, but deciding to run was not one of them IMO. No political action will result in certainty, the goal is ~always to move the needle or take a bunch of swings.
Yes, to be clear I’m not criticizing the initial decision to run but rather the dubious impact estimates and calls to action towards the end of that campaign.
I think it’s good to critically interrogate this kind of analysis. I don’t want to discourage that. But as someone who publicly expressed skepticism about Flynn’s chances, I think there are several differences that mean it warrants closer consideration. The polls are much closer for this race, Biden is well known and experienced at winning campaigns, and the differences between the candidates in this race seem much larger. Based on that it at least seems a lot more reasonable to think Biden could win and that it will be a close race worth spending some effort on.
I appreciate that this post was short and concise enough that I could read the whole thing during my 10 minute lunch break ❤️🙏
Do you have to live in the U.S. (or even in a swing state) to do something useful?