Sorted by New

Wiki Contributions


Choice Writings of Dominic Cummings

It seesm to me that the only way to make that judgement is to actually read Cummings describe his cause.

What grounds do we have for taking that description at face value? I don't think that even his supporters believe his qualities include scrupulous honesty.

Carmex's Shortform

(Sorry if I'm misreading anything; my excuse is that I'm operating on 3 hours' sleep and am not very familiar with Python syntax.)

I ran your 'regular run' version, modified to keep a count of 1-vote victories, and the results were as I would have predicted: https://imgur.com/Y17ecLq

I'm a bit confused by the 'random voter sample' version -- which scenario is that illustrating, and what's the deal with the 'myvote = random.randrange(-voters, voters)' and ' if votes*myvote > votes*votes:' lines?

Carmex's Shortform

I wrote a long response to a related comment chain here: https://www.lesswrong.com/posts/PcfHSSAMNFMgdqFyB/can-you-control-the-past?commentId=jRo2cGuXBbkz54E4o

My short answer to this question is the same as Dagon's: if we're assuming a negligible probability that the election was close enough for your vote to be decisive, 50% in both cases. 

I tried to explain the conflicting intuitions in that other comment. It turned out to be one of those interesting questions that feels less obvious after thinking about it for a couple of minutes than at first glance, but I think I resolved the apparent contradictions pretty clearly in the end.

Can you control the past?

Why can't I think of myself as a randomly sampled voter? 

Same reason you can't ignore other relevant pieces of information -- doing so makes your probability assignments less accurate. For example, if you know that John is a vocal supporter of the less-popular party, you're not going to ignore that information and assign a high probabiity to the proposition that he votes for the winner.

If you're looking at this ex ante, your probability of voting for the winner is ~50% because your vote is uncorrelated with everyone else's. For every possible arrangement of their votes, there are two equally-likely possible worlds, one where you vote for the winner and one where you don't. (Like tailcalled above, I'm assuming for the sake of simplicity that this is a large, realistic election where the chance of a 1-vote margin is negligible. I'll drop this assumption later on.)

Your use of a coinflip is relevant because it guarantees this absence of correlation between your vote and others'. If you were voting on your preferences, these preferences would provide some evidence about the likely preferences of others; if we didn't have any information to the contrary, our best guess would be that you are a fairly typical person with fairly typical preferences.


If you're looking at it after the result is known, I think it's easiest to examine two cases: one where coinflip voting is rare, and one where it is extremely common. 

In the case where it is rare, things are pretty simple. When guessing at the vote of a random voter after a 60-40 election result, we're basically drawing from a giant urn, in which 60% of the balls are Red, 40% are blue, and a handful have an asterisk painted on them to indicate that they are coinflip voters. If we don't know whether our random voter used a coin, we're simply drawing from the full urn, and we have a 60% chance of picking a red ball. But if we know that they did use a coin, we can eliminate all of the un-asterisked balls. Of the remainder, we expect about half to be red and half to be blue, because most sets of coinflips are about 50:50 heads:tails, and the election result doesn't provide much evidence about this particular set of flips. So the coinflip voter is about 50% likely to have voted for the winner.

In the case where coinflip voting is extremely common, things are more interesting. Here, we can't assume that the coinflippers voted 50:50 Red:Blue, because the election result provides non-negligible evidence that the set of coinflips happened to skew in favour of the Red candidate. (In worlds where this happened, a Red victory was significantly more likely, and vice-versa.) So, once we know the result -- and the fact that coinflip voting was extremely common -- we should indeed update in favour of the proposition that random coinflipper X voted for the winner. 

This might seem paradoxical -- why this conflict between ex ante and ex post probabilities? Why shouldn't I assume, ex ante, that I am more likely to vote for the winner? After all, I'm expecting to update in that direction once I hear the result, so why shouldn't I do it now?

Well, I should -- but only because of the cases where my single vote determines the outcome. For simplicity let's suppose  that everyone, including me, votes by coinflip. Suppose there are three voters, and I'm the one listed first in each possible election:

{H H H, H H T, H T H, H T T, T H H, T H T, T T H, T T T}

I only vote for the loser in two of the eight possible elections, because in a three-voter coinflip election my vote is often decisive. But in the cases where my vote is not decisive (i.e. the other two vote the same way as each other), I'm only on the winning team half the time. This holds for larger elections too.

Still, what about the cases where the coinflips do happen to heavily favour one side -- like the HHH outcome above, or the (extremely unlikely, but possible) case where 60% of voters out of a million happened to flip Heads? A randomly selected voter from those elections would be certain (in the HHH case) or likely (in the 60% case) to have voted for the winner. And, supposing you managed to vote blind, there's no reason to treat yourself differently from the random voter here: you probably voted for the winner too. 

Ex ante, these skewed vote tallies are unlikely, but we know they're possible, so why don't we update at least a little bit in favour of you voting for the winner?

The answer is that, excluding the cases where your vote is decisive, these big margins are balanced out by the cases where your candidate loses narrowly. This is because, in a coinflip election, close outcomes are much more likely than less close ones -- and when we exclude the cases where your vote is decisive, a significant number of cases with a 1-vote margin remain, in all of which you vote for the loser. 

You can see this in the three-vote case above, where there are two 3-0 outcomes and six 2-1 outcomes; excluding the four cases where your vote is decisive, we're left with two cases where you win 3-0 and two cases where you lose 2-1. Overall your probability of voting for the winner is 6/8, but that's only because in 4/8 cases your vote is decisive.

It's the same in a five-vote case, where there are 32 equiprobable outcomes: in two cases you win 5-0; in 8 cases you win 4-1 and in 2 cases you lose 4-1; and in twelve cases you win 3-2, while in 8 cases you lose 3-2. In all twelve 3-2 wins your vote is decisive, so excluding them, your win-loss record is 10-10. Including them, it is 22-10, because of the 12 cases where your vote is decisive.

This generalises: you always have a >50% probability of voting for the winner, but this is always fully accounted for by the chance that your vote causally affects the outcome in the usual way.

Common knowledge about Leverage Research 1.0

I want to draw attention to the fact that "Kerry Vaughan" is a brand new account that has made exactly three comments, all of them on this thread. "Kerry Vaughan" is associated with Leverage. "Kerry Vaughan"'s use of "they" to describe Leverage is deliberately misleading.

To be fair, KV was open about that association in both previous comments, using 'we' in the first and including this disclaimer in the second --

(I currently work at Leverage research but did not work at Leverage during Leverage 1.0 (although I interacted with Leverage 1.0 and know many of the people involved). Before working at Leverage I did EA community building at CEA between Summer 2014 and early 2019.)

-- which also seems to explain the use of 'they' in KV's third comment, which referred specifically to "Leverage 1.0".

(I hope this goes without saying on LW, but I don't mean this as a general defense of Leverage or of KV's opinions. I know nothing about either beyond what I've read here, and I haven't even read all the relevant comments. Personally I wouldn't get involved with an organisation like Leverage.)

Petrov Day 2021: Mutually Assured Destruction?

I don't know if this would defeat part of the purpose, but what about making it opt-in over a long time period, e.g. giving people all year to put themselves on the list of people who might be chosen to receive codes? 

Other than that, I think it's mostly a question of (to the extent possible without undermining what you're trying to do) making it pretty clear to the recipients that people take this seriously and would genuinely like them to refrain from using the codes. As far as I can tell, that has already improved from last year. (It seems like there might have been some tonal ambiguity last year, with phrasing intended to be heightened but mostly serious coming across to some readers as playful and mostly joking.)

Petrov Day 2021: Mutually Assured Destruction?

I understand why this was downvoted and I think it is harsh, but I also think it might be good if people take the sentiment seriously rather than bury+ignore it.

If I received a code, I would do nothing, because it's clear by now that pressing the button would seriously upset some people. (And the consequences seem potentially more significant this year than last.) And I think the parent commenter undervalues the efforts the pro-taking-it-seriously people made to keep their emotions in check and explain why they take the ritual seriously and would like others to do so too.

But I share the instinctive reaction that the whole thing is a bit overblown and pompous, and even on reflection I think it's at least reasonable to hold that it was obnoxious to throw unconsenting people into a situation that looked like a game, where the stakes appeared (and IMO were) very low, only to reveal after the fact that playing the game -- by taking an action explicitly enabled by the people who run and probably care most about the site -- had apparently caused non-trivial distress to others and significant reputational harm to the player.

Insights from Modern Principles of Economics

This also applies to the sweatshop example. If everything else is fixed, then yeah, probably those poor families are better off being allowed to work in awful conditions for low wages. But everything else isn't fixed. 

When a bunch of relatively wealthy and powerful people are benefiting financially from the existence of an extremely poor underclass (who, due to their poverty, are willing to do hard unpleasant work for little money), this creates resistance to reforms that would improve the situation of the poor and thereby give them greater bargaining power. 

And, slightly more optimistically, the alternative to sweatshop employment might not be as dire as it seems. When people are literally at risk of starving for lack of work, there tends to be greater political and charitable will to help them, compared to when they're out-of-sight out-of-mind doing something unpleasant but useful.

LessWrong is paying $500 for Book Reviews

Your book review must be published after the posting of this announcement, i.e., no submitting book reviews you wrote a month ago and already published elsewhere on the Internet.

What's your policy on previously-partially-published reviews? The specific case I have in mind is a rough review I put up on Goodreads, which would need major reworking to be suitable here. (It's currently more of a notes-dump than a proper review.)

How much COVID protection is gained from protecting your eyes?

glasses evidence better explained by the fact that fogging glasses are a good feedback loop/incentive to wear your mask properly

This could be an interesting thing to study. Anecdotally, I think I've seen more people give 'fogged glasses' as a reason/excuse not to wear a mask, or to pull it right down, than as a reason to fit the mask properly. For some people and some mask types, there seems to be an assumption that air escaping out the top is inevitable.

Load More