I presume you’re saying that utility-based simulations are not credible. I don’t think you’re actually trying to say that they’re not numerical estimates. So let me explain what I’m talking about, then say what parts I’m claiming are “credible”.
I’m talking about monte-carlo simulations of voter satisfaction efficiency. You use some statistical model to generate thousands of electorates (that is, voters with numeric utilities for candidates); a media model to give the voters information about each other; and a strategy model to turn information, utilities, and choice of voting system into valid ballots for that voting system. Then, you see who wins each time, and calculate the average overall utility of that winners. Clearly, there are a lot of questionable assumptions in terms of the statistical, media, and strategy models, but the interesting thing is that exploring various assumptions in all of those cases shows that the (plurality-dictatorship)≈(good system-plurality) equation is pretty robust, with various systems such as approval, condorcet, majority judgment, score, or SODA in place of “good system”.
There are certainly various ways to criticize the above.
“Don’t believe it”: If you think that I’ve messed up my math or not done a good job with the sensitivity analysis, of course you’d question my conclusions. But if you want to play with my code to check it, it’s here.
“Utilitarianism is a bad metric”: It may not be perfect, but as far as I can tell it’s the only rational way to put numbers on things.
“Democracy is a bad idea”: In other words, if you think that the average voter’s estimate of their utility for a candidate has 0 or negative correlation with their true utility of that candidate winning, then this simulation is garbage. I’d respond with the old saying about democracy being the worst system except all the others.
“The advantages of democracy over dictatorship aren’t in terms of who’s in charge”: if you think that democracy’s clear superiority to dictatorship in terms of human welfare comes from something other than choosing better leaders (such as, for instance, reducing the prevalence of civil wars), then improving the voting system might not be able to have comparable payoff as instituting a voting system to begin with. I’d respond that this critique is probably partially right, but on the other hand, better leadership could credibly have better responses to crises (financial, environmental, and/or existential-risk) which could indeed be on the same order as the democracy dividend.
All in all, taking a more outside view, I see how the combination of the above objections would reduce your estimate of the expected “voting system dividend”. Still, when I “shut up and multiply” I get: $80 trillion world GDP plausible (conservative) effect size in a good year of 2% .1 plausible portion of good years over time .5 plausible portion of good years over space (some country’s economies might already be immune to the kind of harm this could prevent) .5 chance you trust my simulations .1 correlation of voter preference with utility .5 probability leadership makes any difference = about $2 billion/year potential payoff in expected value, even without compounding. That seems to me like (a) quite a conservative choice of factors, (b) not a totally implausible end result, and (c) still big enough to care about. Of course, it’s incredibly back-of-the-envelope, but I invite you to try doing the estimation yourself.
I presume you’re saying that utility-based simulations are not credible, because they’re clearly numerical estimates.
Actually, no, that’s not what I mean. I have no problems with numerical estimates in general.
What I mean by “credible”, in this context, is “shown to be relevant to real-life situations” and “supported by empirical data”.
You’ve constructed a model. You’ve played with this model and have an idea of how it behaves in different regimes. That’s all fine. But then you imply that this model reflects the real world and it’s at this point that I start to get sceptical and ask for evidence. Not evidence of how your model works, but evidence that the map matches the territory.
The model is not easy to subject to full, end-to-end testing. It seems reasonable to test it one part at a time. I’m doing the best I can to do so:
I’ve run an experiment on Amazon Mechanical Turk involving hundreds of experimental subjects voting in dozens of simulated elections to probe my strategy model.
I’m working on getting survey data and developing statistical tools to refine my statistical model (mostly, posterior predictive checks; but it’s not easy, given that this is a deeper hierarchical model than most).
In terms of the utilitarian assumptions of my model, I’m not sure how those are testable rather than just philosophical / assumed axioms. Not that I regard these assumptions as truly axiomatic, but that I think they’re pretty necessary to get anywhere at all, and in practice unlikely to be violated severely enough to invalidate the work.
I haven’t started work on testing / refining my media model (other than some head-scratching), but I can imagine how to do at least a few spot checks with posterior predictive checks too.
The assumptions that preference and utility correlate positively, even in an environment where candidates are strategic about exploiting voter irrationality, are certainly questionable. But insofar as these are violated, it would just make democracy a bad idea in general, not invalidate the fact that plurality is still a worse idea than other voting systems such as approval. Also, I think it would be basically impossible to test these assumptions without implausibly accurate and unbiased measurements of true utility. Finally, call me a hopeless optimist, but I do actually have faith that democracy is a good idea because “you can’t fool all the people all the time”.
I do actually have faith that democracy is a good idea
Democracy is complicated. For a simple example, consider full direct democracy: instant whole-population referendums on every issue. I am not sure anyone considers this a good idea—successful real-life democratic systems (e.g. the US) are built on limited amounts of democracy which is constrained in many ways. Given this, democracy looks to be a Goldilocks-type phenomenon where you don’t want too little, but you don’t want too much either.
And, of course, democracy involves much more than just voting—there are heavily… entangled concepts like the rule of law, human rights, civil society, etc.
Full direct democracy is a bad idea because it’s incredibly inefficient (and thus also boring/annoying, and also subject to manipulation by people willing to exploit others’ boredom/annoyance). This has little or nothing to do with whether people’s preferences correlate with their utilities, which is the question I was focused on. In essence, this isn’t a true Goldilocks situation (“you want just the right amount of heat”) but rather a simple tradeoff (“you want good decisions, but don’t want to spend all your time making them”).
As to the other related concepts… I think this is getting a bit off-topic. The question is, is energy (money) spent on pursuing better voting systems more of a valid “saving throw” than when spent on pursuing better individual rationality. That’s connected to the question of the preference/utility correlation of current-day, imperfectly-rational voters. I’m not seeing the connection to rule of law &c.
Full direct democracy is a bad idea because it’s incredibly inefficient
No, I don’t think so. It is a bad idea even in a society technologically advanced to make it efficient and even if it’s invoked not frequently enough to make it annoying.
whether people’s preferences correlate with their utilities
People’s preferences are many, multidimensional, internally inconsistent, and dynamic. I am not quite sure what do you want to correlate to a single numerical value of “utility”.
The question is, is energy (money) spent on pursuing better voting systems more of a valid “saving throw” than when spent on pursuing better individual rationality.
Why are you considering only these two options?
I’m not seeing the connection to rule of law &c.
The connection is that what is a “better” voting system depends on the context, context that includes things like rule of law, etc.
You’re raising some valid questions, but I can’t respond to all of them. Or rather, I could respond (granting some of your arguments, refining some, and disputing some), but I don’t know if it’s worth it. Do you have an underlying point to make, or are you just looking for quibbles? If it’s the latter, I still thank you for responding (it’s always gratifying to see people care about issues that I think are important, even if they disagree); but I think I’ll disengage, because I expect that whatever response I give would have its own blemishes for you to find.
(small note: the sentence you quote from me was unclear. “because” related to “presume”, not “saying”. But your response to what I accidentally said is still largely cogent in relation to what I meant to say, so the miscommunication isn’t important. Still, I’ve corrected the original. Future readers: lumifer quoted me correctly.)
The first three words here are in contradiction with the last three words… :-/
I presume you’re saying that utility-based simulations are not credible. I don’t think you’re actually trying to say that they’re not numerical estimates. So let me explain what I’m talking about, then say what parts I’m claiming are “credible”.
I’m talking about monte-carlo simulations of voter satisfaction efficiency. You use some statistical model to generate thousands of electorates (that is, voters with numeric utilities for candidates); a media model to give the voters information about each other; and a strategy model to turn information, utilities, and choice of voting system into valid ballots for that voting system. Then, you see who wins each time, and calculate the average overall utility of that winners. Clearly, there are a lot of questionable assumptions in terms of the statistical, media, and strategy models, but the interesting thing is that exploring various assumptions in all of those cases shows that the (plurality-dictatorship)≈(good system-plurality) equation is pretty robust, with various systems such as approval, condorcet, majority judgment, score, or SODA in place of “good system”.
There are certainly various ways to criticize the above.
“Don’t believe it”: If you think that I’ve messed up my math or not done a good job with the sensitivity analysis, of course you’d question my conclusions. But if you want to play with my code to check it, it’s here.
“Utilitarianism is a bad metric”: It may not be perfect, but as far as I can tell it’s the only rational way to put numbers on things.
“Democracy is a bad idea”: In other words, if you think that the average voter’s estimate of their utility for a candidate has 0 or negative correlation with their true utility of that candidate winning, then this simulation is garbage. I’d respond with the old saying about democracy being the worst system except all the others.
“The advantages of democracy over dictatorship aren’t in terms of who’s in charge”: if you think that democracy’s clear superiority to dictatorship in terms of human welfare comes from something other than choosing better leaders (such as, for instance, reducing the prevalence of civil wars), then improving the voting system might not be able to have comparable payoff as instituting a voting system to begin with. I’d respond that this critique is probably partially right, but on the other hand, better leadership could credibly have better responses to crises (financial, environmental, and/or existential-risk) which could indeed be on the same order as the democracy dividend.
All in all, taking a more outside view, I see how the combination of the above objections would reduce your estimate of the expected “voting system dividend”. Still, when I “shut up and multiply” I get: $80 trillion world GDP plausible (conservative) effect size in a good year of 2% .1 plausible portion of good years over time .5 plausible portion of good years over space (some country’s economies might already be immune to the kind of harm this could prevent) .5 chance you trust my simulations .1 correlation of voter preference with utility .5 probability leadership makes any difference = about $2 billion/year potential payoff in expected value, even without compounding. That seems to me like (a) quite a conservative choice of factors, (b) not a totally implausible end result, and (c) still big enough to care about. Of course, it’s incredibly back-of-the-envelope, but I invite you to try doing the estimation yourself.
Actually, no, that’s not what I mean. I have no problems with numerical estimates in general.
What I mean by “credible”, in this context, is “shown to be relevant to real-life situations” and “supported by empirical data”.
You’ve constructed a model. You’ve played with this model and have an idea of how it behaves in different regimes. That’s all fine. But then you imply that this model reflects the real world and it’s at this point that I start to get sceptical and ask for evidence. Not evidence of how your model works, but evidence that the map matches the territory.
The model is not easy to subject to full, end-to-end testing. It seems reasonable to test it one part at a time. I’m doing the best I can to do so:
I’ve run an experiment on Amazon Mechanical Turk involving hundreds of experimental subjects voting in dozens of simulated elections to probe my strategy model.
I’m working on getting survey data and developing statistical tools to refine my statistical model (mostly, posterior predictive checks; but it’s not easy, given that this is a deeper hierarchical model than most).
In terms of the utilitarian assumptions of my model, I’m not sure how those are testable rather than just philosophical / assumed axioms. Not that I regard these assumptions as truly axiomatic, but that I think they’re pretty necessary to get anywhere at all, and in practice unlikely to be violated severely enough to invalidate the work.
I haven’t started work on testing / refining my media model (other than some head-scratching), but I can imagine how to do at least a few spot checks with posterior predictive checks too.
The assumptions that preference and utility correlate positively, even in an environment where candidates are strategic about exploiting voter irrationality, are certainly questionable. But insofar as these are violated, it would just make democracy a bad idea in general, not invalidate the fact that plurality is still a worse idea than other voting systems such as approval. Also, I think it would be basically impossible to test these assumptions without implausibly accurate and unbiased measurements of true utility. Finally, call me a hopeless optimist, but I do actually have faith that democracy is a good idea because “you can’t fool all the people all the time”.
tl;dr: I’m working on this.
Democracy is complicated. For a simple example, consider full direct democracy: instant whole-population referendums on every issue. I am not sure anyone considers this a good idea—successful real-life democratic systems (e.g. the US) are built on limited amounts of democracy which is constrained in many ways. Given this, democracy looks to be a Goldilocks-type phenomenon where you don’t want too little, but you don’t want too much either.
And, of course, democracy involves much more than just voting—there are heavily… entangled concepts like the rule of law, human rights, civil society, etc.
Full direct democracy is a bad idea because it’s incredibly inefficient (and thus also boring/annoying, and also subject to manipulation by people willing to exploit others’ boredom/annoyance). This has little or nothing to do with whether people’s preferences correlate with their utilities, which is the question I was focused on. In essence, this isn’t a true Goldilocks situation (“you want just the right amount of heat”) but rather a simple tradeoff (“you want good decisions, but don’t want to spend all your time making them”).
As to the other related concepts… I think this is getting a bit off-topic. The question is, is energy (money) spent on pursuing better voting systems more of a valid “saving throw” than when spent on pursuing better individual rationality. That’s connected to the question of the preference/utility correlation of current-day, imperfectly-rational voters. I’m not seeing the connection to rule of law &c.
No, I don’t think so. It is a bad idea even in a society technologically advanced to make it efficient and even if it’s invoked not frequently enough to make it annoying.
People’s preferences are many, multidimensional, internally inconsistent, and dynamic. I am not quite sure what do you want to correlate to a single numerical value of “utility”.
Why are you considering only these two options?
The connection is that what is a “better” voting system depends on the context, context that includes things like rule of law, etc.
You’re raising some valid questions, but I can’t respond to all of them. Or rather, I could respond (granting some of your arguments, refining some, and disputing some), but I don’t know if it’s worth it. Do you have an underlying point to make, or are you just looking for quibbles? If it’s the latter, I still thank you for responding (it’s always gratifying to see people care about issues that I think are important, even if they disagree); but I think I’ll disengage, because I expect that whatever response I give would have its own blemishes for you to find.
In other words: OK, so what?
Some people find blemish-finding services valuable, some don’t :-)
Fair enough. Thanks. Again, I agree with some of your points. I like blemish-picking as long as it doesn’t require open-ended back-and-forth.
(small note: the sentence you quote from me was unclear. “because” related to “presume”, not “saying”. But your response to what I accidentally said is still largely cogent in relation to what I meant to say, so the miscommunication isn’t important. Still, I’ve corrected the original. Future readers: lumifer quoted me correctly.)