Range voting
Against my better judgement I checked out the website advocating for range voting for political elections:
http://rangevoting.org/
Supporters of Range voting especially don't like Instant Runoff Voting (IRV) because it is the only politically viable rival to their politically unviable alternative.
http://rangevoting.org/rangeVirv.html Why Range Voting is Better than IRV (Instant Runoff Voting)
I am an IRV support and am NOT a Range Voting supporter, although I acknowledge that IRV is "only better than plurality" and defendably "inferior to a top-two runoff".
A top-two runoff is BETTER primarily because voting is simpler - you pick one candidate to support, and then (if necessary) return for a second vote to confirm a majority winner among the top-two. In contrast IRV requires voters to rank enough candidates so that they can be SURE one of their choices will be among the final two. Ranking may sound easy AND truth-be-told, OUGHT to be done in a top-two runoff as well. Basically in either runoff system, voters must GUESS who could make the final round. I say MUST only in the sense of "best tactic". Honest voting is fine under a runoff, you just have a little less influence than those who think further. How? Basically if there's three strong candidates and the top-two is UNCERTAIN, then a "strategic compromise" might help. Basically you don't want your favorite to BEAT your compromise and then LOSE, if your compromise could have won. Yes, messy, but that's the reality of runoffs.
There's still the alternative - plurality - let the top candidate win no matter how few votes he or she gets. That's what we got for most elections, and plurality and runoffs BOTH rewards parties to organize and unify their support behind one strong candidate.
But why must we do this? I mean WHY can't we all just vote for as many candidates as we want, add all the votes and elect the winner with the most votes. This is called Approval voting and it is JUST like plurality, but there's no restriction of identifying ONLY one best candidate. The winner might have less than a majority, OR there might be multiple candidates above 50%, and still only the top one wins.
I'm all OK with this as it goes, as written above. I mean the strategy is still messy, AND the BEST choice is still to "vote for one" if you think your top choice has a good chance to win. BUT you have the bonus option to hedge your bet for a bunch of decent candidates as you like.
Approval voting doesn't promise a majority winner, BUT worse than that it should NEVER TRY. I mean for instance you might be tempted to say "Well, if there's no majority winner, we'll have a runoff and try again" BUT why?! If you force elimination (like IRV), you open the door to manipulation of false support. If you disallow forced elimination, why would anyone change their vote? And why allow people to change their vote at all? Approvaling voting requires a strategy, like the game Chicken - so taking away consequences takes away the incentive for people to vote honestly and seriously in the first round. Similarly if there was two majority candidates, allowing a revote opens doors that reward false votes in the first round.
Secondly with Range Voting, it is Approval voting, but allowing fractional votes, like percents. I could support Nader 50%, and Kerry 90%, or any combination.
Now there is a "single-vote" intepretation of Approval, has been called "Equal and Even Cumulative voting" (CV), so I could vote for Nader and Kerry and both get 1/2 my vote. And equally I could have fractional Cumulative voting, giving Nader 35%, and Kerry 65%, as long as the sum doesn't exceed 100%. That's merely an extention of plurality, a harmless choice (doesn't given anyone any more power to influence the results), AND it has a property called "Semiproportionality" for multiple winner elections which means in guarantees, if you're electing say 4 seats, any candidate who gets 20% of the vote will be elected, although there's no lower limit on support to be elected, if there's too many candidates running.
And that's the difficulty of CV. Without a runoff, voters have to divide their vote too widely to get a chance to pick a winner and voters with better guesses on where support is needed will have more influence. STV (Single Transferable vote) is a sort of combination of CV (a fractional vote divided among choices automatically optimized where needed) and IRV rank-preference ballots. Not even Range supporters usually put down STV even if they don't like IRV for single winners.
Back to Range voting, there's no great value in supporting someone "half-way" - no reason to not give Nader 100% and Kerry 100%, if you like both. I mean you don't want Kerry losing to Bush because you only gave him 90%. It's always best to maximize the difference between the candidates you don't like and those you do - SO approval voting represents that limit.
Incidentally, there's other compromises that are interesting. You can mix IRV with E&E-CV with an approval ballot rather than a rank preference ballot. So if you support 3 candidates in the first round each gets 33.33% of your vote. If one is eliminated (for being last place), then your new vote is 50% for your two choices. In the final round, you'll ideally have 100% of your vote on one choice. This has an advantage of using simple ballots. It has a "cute factor" for transfering votes more smoothly thank rankings, but it still can have nonmonotonic results like any runoff. (Your favorite can help defeat your compromise and both lose.)
Okay back to Range/Approval, I consider them "mostly harmless" as a single round of voting, no runoff. Voters still have a difficult problem, and different strategies - from "bullet voting" (for one) to "saturation voting" (for ALL acceptable) or something between. How much you want to compromise basically equals how afraid you are of those you don't like.
I accept the conclusion that you should vote in Approval identically as you would in plurality. This is vote for the candidate you WOULD vote for in plurality. THEN add votes for everyone you like better, but didn't vote for because you didn't think they had a chance.
That means if everyone makes rational choices, Approval and Plurality will always agree in winners. So no reform if you're a smart voter already in plurality.
I mean for instance, in the last election for MN Gov, I planned to support Peter Hutchinson from the Independence Party, BUT polls showed he was far behind AND Hatch and Pawlenty were close. I didn't have a great preference between Hatch and Pawlenty, but seeing they were on top, I compromised to Hatch, and abandoned Hutchinson. So in Approval, I could have done my compromise AND supported by true choice. A cute but mostly powerless choice.
That said I believe in Approving voting for something I called "Variable number of winner polls." Specifically asking a question like "Who do you want to hear in the debate?" This question can really have varied answers from EVERYONE to NO ONE and they're all legitimate. So a debate sponsor might offer an Approval (or ratings vote) and allow all candidates above some rating level to participate, like 20% perhaps, or start with a low threshold like 5%, and slowly raise it for later debates. That's a clear example to me because you really do have N independent elections - that's a legitimate use for approval.
Anyway, I don't have too much to disagree with from what I read on the Range Voting website. I won't defend IRV as being "spoiler free", and I accept its weaknesses. I just disagree that approval has much to offer either.
I consider plurality as "the enemy", but I can have sympathy for plurality supporters as well. Majority rule isn't that great of an achievement and not clearly better than "strong plurality". PLUS I know the spoiler factor of plurality is what strengthens parties to unify behind one candidate and what helps leave us with a smaller number of candidates to consider.
Ultimately plurality is also "mostly harmless", and approval as equivalent.
More interesting is STV and multiple winner elections, and that's where FairVote wants to go, and I like experimenting, so happy to support it. PR has its own detractors who say it empowers the wrong candidates - special interest candidates who aren't interested in the bigger picture of the common good. I won't argue except to say no election method promises good candidates will win, or that voters know what they're doing.
But Range voting, oh, intellectually interesting for a while, but I can't support it. Sorry!
http://rangevoting.org/
Supporters of Range voting especially don't like Instant Runoff Voting (IRV) because it is the only politically viable rival to their politically unviable alternative.
http://rangevoting.org/rangeVirv.html Why Range Voting is Better than IRV (Instant Runoff Voting)
I am an IRV support and am NOT a Range Voting supporter, although I acknowledge that IRV is "only better than plurality" and defendably "inferior to a top-two runoff".
A top-two runoff is BETTER primarily because voting is simpler - you pick one candidate to support, and then (if necessary) return for a second vote to confirm a majority winner among the top-two. In contrast IRV requires voters to rank enough candidates so that they can be SURE one of their choices will be among the final two. Ranking may sound easy AND truth-be-told, OUGHT to be done in a top-two runoff as well. Basically in either runoff system, voters must GUESS who could make the final round. I say MUST only in the sense of "best tactic". Honest voting is fine under a runoff, you just have a little less influence than those who think further. How? Basically if there's three strong candidates and the top-two is UNCERTAIN, then a "strategic compromise" might help. Basically you don't want your favorite to BEAT your compromise and then LOSE, if your compromise could have won. Yes, messy, but that's the reality of runoffs.
There's still the alternative - plurality - let the top candidate win no matter how few votes he or she gets. That's what we got for most elections, and plurality and runoffs BOTH rewards parties to organize and unify their support behind one strong candidate.
But why must we do this? I mean WHY can't we all just vote for as many candidates as we want, add all the votes and elect the winner with the most votes. This is called Approval voting and it is JUST like plurality, but there's no restriction of identifying ONLY one best candidate. The winner might have less than a majority, OR there might be multiple candidates above 50%, and still only the top one wins.
I'm all OK with this as it goes, as written above. I mean the strategy is still messy, AND the BEST choice is still to "vote for one" if you think your top choice has a good chance to win. BUT you have the bonus option to hedge your bet for a bunch of decent candidates as you like.
Approval voting doesn't promise a majority winner, BUT worse than that it should NEVER TRY. I mean for instance you might be tempted to say "Well, if there's no majority winner, we'll have a runoff and try again" BUT why?! If you force elimination (like IRV), you open the door to manipulation of false support. If you disallow forced elimination, why would anyone change their vote? And why allow people to change their vote at all? Approvaling voting requires a strategy, like the game Chicken - so taking away consequences takes away the incentive for people to vote honestly and seriously in the first round. Similarly if there was two majority candidates, allowing a revote opens doors that reward false votes in the first round.
Secondly with Range Voting, it is Approval voting, but allowing fractional votes, like percents. I could support Nader 50%, and Kerry 90%, or any combination.
Now there is a "single-vote" intepretation of Approval, has been called "Equal and Even Cumulative voting" (CV), so I could vote for Nader and Kerry and both get 1/2 my vote. And equally I could have fractional Cumulative voting, giving Nader 35%, and Kerry 65%, as long as the sum doesn't exceed 100%. That's merely an extention of plurality, a harmless choice (doesn't given anyone any more power to influence the results), AND it has a property called "Semiproportionality" for multiple winner elections which means in guarantees, if you're electing say 4 seats, any candidate who gets 20% of the vote will be elected, although there's no lower limit on support to be elected, if there's too many candidates running.
And that's the difficulty of CV. Without a runoff, voters have to divide their vote too widely to get a chance to pick a winner and voters with better guesses on where support is needed will have more influence. STV (Single Transferable vote) is a sort of combination of CV (a fractional vote divided among choices automatically optimized where needed) and IRV rank-preference ballots. Not even Range supporters usually put down STV even if they don't like IRV for single winners.
Back to Range voting, there's no great value in supporting someone "half-way" - no reason to not give Nader 100% and Kerry 100%, if you like both. I mean you don't want Kerry losing to Bush because you only gave him 90%. It's always best to maximize the difference between the candidates you don't like and those you do - SO approval voting represents that limit.
Incidentally, there's other compromises that are interesting. You can mix IRV with E&E-CV with an approval ballot rather than a rank preference ballot. So if you support 3 candidates in the first round each gets 33.33% of your vote. If one is eliminated (for being last place), then your new vote is 50% for your two choices. In the final round, you'll ideally have 100% of your vote on one choice. This has an advantage of using simple ballots. It has a "cute factor" for transfering votes more smoothly thank rankings, but it still can have nonmonotonic results like any runoff. (Your favorite can help defeat your compromise and both lose.)
Okay back to Range/Approval, I consider them "mostly harmless" as a single round of voting, no runoff. Voters still have a difficult problem, and different strategies - from "bullet voting" (for one) to "saturation voting" (for ALL acceptable) or something between. How much you want to compromise basically equals how afraid you are of those you don't like.
I accept the conclusion that you should vote in Approval identically as you would in plurality. This is vote for the candidate you WOULD vote for in plurality. THEN add votes for everyone you like better, but didn't vote for because you didn't think they had a chance.
That means if everyone makes rational choices, Approval and Plurality will always agree in winners. So no reform if you're a smart voter already in plurality.
I mean for instance, in the last election for MN Gov, I planned to support Peter Hutchinson from the Independence Party, BUT polls showed he was far behind AND Hatch and Pawlenty were close. I didn't have a great preference between Hatch and Pawlenty, but seeing they were on top, I compromised to Hatch, and abandoned Hutchinson. So in Approval, I could have done my compromise AND supported by true choice. A cute but mostly powerless choice.
That said I believe in Approving voting for something I called "Variable number of winner polls." Specifically asking a question like "Who do you want to hear in the debate?" This question can really have varied answers from EVERYONE to NO ONE and they're all legitimate. So a debate sponsor might offer an Approval (or ratings vote) and allow all candidates above some rating level to participate, like 20% perhaps, or start with a low threshold like 5%, and slowly raise it for later debates. That's a clear example to me because you really do have N independent elections - that's a legitimate use for approval.
Anyway, I don't have too much to disagree with from what I read on the Range Voting website. I won't defend IRV as being "spoiler free", and I accept its weaknesses. I just disagree that approval has much to offer either.
I consider plurality as "the enemy", but I can have sympathy for plurality supporters as well. Majority rule isn't that great of an achievement and not clearly better than "strong plurality". PLUS I know the spoiler factor of plurality is what strengthens parties to unify behind one candidate and what helps leave us with a smaller number of candidates to consider.
Ultimately plurality is also "mostly harmless", and approval as equivalent.
More interesting is STV and multiple winner elections, and that's where FairVote wants to go, and I like experimenting, so happy to support it. PR has its own detractors who say it empowers the wrong candidates - special interest candidates who aren't interested in the bigger picture of the common good. I won't argue except to say no election method promises good candidates will win, or that voters know what they're doing.
But Range voting, oh, intellectually interesting for a while, but I can't support it. Sorry!
posted by Ares Olympus at 5:10 PM
1 Comments:
Instant Runoff Voting is one of the worst voting methods, according to the world's most extensive Bayesian regret calculations, by Princeton math Ph.D. Warren D. Smith. Whereas Range Voting surpasses all other methods that have ever been seriously proposed (and when he discovered this, he formed the Center for Range Voting).
The results show that Range Voting surpasses plurality voting by as much as plurality surpasses non-democratic random selection; so we can say that Range Voting literally doubles the effect of democracy. IRV, while slightly better than plurality, comes nowhere near that. This means that if you had a choice to live in a society with Range Voting, or one with IRV (or Approval Voting), you'd be a great deal more satisfied with election results, on average, by choosing Range Voting. IRV yields a substantially less representative outcome. Range Voting is substantially better.
When it comes to the political viability of Range Voting, you may be right. But Approval Voting is a good start, since it is the simplest form of Range Voting, and could open the door to further improvement (and even if it doesn't, it's almost as good as Range Voting).
Regarding IRV: Range Voting is simpler to implement and tally than IRV; and can be sub-totalled at the precinct level, unlike IRV; and can be performed on all standard plurality voting machines, unlike IRV. Many election integrity experts, such as Rebecca Mercuri, dislike IRV because it incentives the adoption of fraud-conducive electronic voting machines. Range Voting doesn't have that problem; Range Voting is better on the back end of the election. But how about the front end? Well, IRV causes spoiled ballots to become 7 times as numerous (true here in my home of San Francisco), but Range Voting experimentally reduces the number of spoiled ballots (and Approval Voting is even better). Voters evidently agree that something is simpler about Range Voting, since they mess up less often with it.
A top-two runoff is BETTER primarily because voting is simpler - you pick one candidate to support, and then (if necessary) return for a second vote to confirm a majority winner among the top-two.
While simplicity certainly is desirable, there's nothing about simplicity that definitively makes a voting method better. For example, Range Voting is more complex than plurality voting, but is still enormously better, and even causes fewer spoiled ballots. Range Voting makes up for its added complexity by being a superior algorithm for determining the winner.
In contrast IRV requires voters to rank enough candidates so that they can be SURE one of their choices will be among the final two.
On the contrary, IRV can easily give a voter an incentive to rank none (i.e. stay at home), rather than cast an honest vote.
Ranking may sound easy AND truth-be-told, OUGHT to be done in a top-two runoff as well.
Actually, Approval Voting is much better than either method, and requires no ranking at all. It sounds like you may be acknowledging that.
Basically in either runoff system, voters must GUESS who could make the final round. I say MUST only in the sense of "best tactic". Honest voting is fine under a runoff, you just have a little less influence than those who think further. How? Basically if there's three strong candidates and the top-two is UNCERTAIN, then a "strategic compromise" might help. Basically you don't want your favorite to BEAT your compromise and then LOSE, if your compromise could have won. Yes, messy, but that's the reality of runoffs.
Ah, but what you probably don't know is that this scenario happens in about 20% of IRV elections. So what a voter has to do is ask himself, "Does my favorite candidate have at least a 20% chance of winning?" If not, then the voter is strategically forced to top-rank his favorite of the front-runners, regardless of who his actual favorite is. In fact that could be said more generally as, "Only place a more preferred candidate ahead of your favorite front-runner if that more preferred candidate has at least a 20% chance of winning."
There's still the alternative - plurality - let the top candidate win no matter how few votes he or she gets. That's what we got for most elections, and plurality and runoffs BOTH rewards parties to organize and unify their support behind one strong candidate.
Actually that's another myth. Parties can easily nominate, and then "unify their support behind" the weakest candidate, because of the horrible plurality voting method. This is why parties should use Range Voting for their nomination process (e.g. Iowa caucuses), to gain a competitive edge.
But why must we do this? I mean WHY can't we all just vote for as many candidates as we want, add all the votes and elect the winner with the most votes. This is called Approval voting and it is JUST like plurality, but there's no restriction of identifying ONLY one best candidate. The winner might have less than a majority, OR there might be multiple candidates above 50%, and still only the top one wins.
As you basically point out, Approval Voting is just Range Voting on a 0-1 scale, as opposed to e.g. a 0-10 scale. (Thanks for supporting Range Voting.) While Approval Voting is simpler, and probably more politically viable, using a larger range (e.g. 0-10) is much better.
I'm all OK with this as it goes, as written above. I mean the strategy is still messy, AND the BEST choice is still to "vote for one" if you think your top choice has a good chance to win. BUT you have the bonus option to hedge your bet for a bunch of decent candidates as you like.
Actually no, the strategy isn't messy at all. You start by voting for the candidate you'd vote for if you could only vote for one (like we all do already with plurality voting). Then you vote for every other candidate you like better than that one. For example, a Nader supporter in 2000 who strategically voted for Gore, would start by voting for Gore - but then he would be free to vote for Nader, and any other candidates he liked better than Gore. So the strategy with Approval Voting is really no messier than with plurality voting. It just takes another minute or so to cast those additional votes, if your favorite isn't the first one you voted for. Having much better election outcomes is certainly worth that.
Approval voting doesn't promise a majority winner
"Majority winner" is an antiquated concept. People need to understand the majority myth. The quality of a voting method is a matter of how utilitarian it is.
BUT worse than that it should NEVER TRY. I mean for instance you might be tempted to say "Well, if there's no majority winner, we'll have a runoff and try again" BUT why?!
Well, some would argue in favor of Condorcet voting here (elect a beats-all winner when one exists). That would then require you to point out the bankrupty of the majoritarian concept, as well as the susceptibility of Condorcet methods to strategic voting, as well as the fact that Range Voting plausibly is a better Condorcet method than real Condorcet methods.
Now there is a "single-vote" intepretation of Approval, has been called "Equal and Even Cumulative voting" (CV)
Cumulative voting is a total farce. It is strategically identical to plurality voting, since the goal is to give 100% of your vote to the candidate you'd vote for in a plurality election.
..it has a property called "Semiproportionality" for multiple winner elections which means in guarantees, if you're electing say 4 seats, any candidate who gets 20% of the vote will be elected
I don't know how relevant the concept of "semiproportionality" is. Say you have an election where the voters are 21% Reds, and 79% Blues, and there are two candidates being elected out of a field of 2 Reds and 8 Blues. You could conceivably elect both Red candidates, because of the vote splitting problem. We might suppose that, on average, there is some kind of rough proportionality here, but this is a terrible P.R. method overall.
Two superior methods are Reweighted Range Voting and Asset Voting. These methods are simpler and better than STV. And Reweighted Range Voting defaults to standard Range Voting in the single-winner case, the same way STV becomes IRV. (Only RRV/RV are massively better than STV/IRV.)
Not even Range supporters usually put down STV even if they don't like IRV for single winners.
STV such a bad voting method, but it has been surpassed by more recent methods (like the two above, invented by a Princeton math Ph.D.) which take advantage of the 100+ years of advances in election science that have occurred since STV was invented.
Back to Range voting, there's no great value in supporting someone "half-way" - no reason to not give Nader 100% and Kerry 100%, if you like both.
Strategically, you are correct. And by that same logic you could also say that there's no good reason for voters to register outside of one of the major two parties, because it diminishes their power, since it prevents them from voting in most states' primaries. Yet nearly 1/3 of voters do it anyway. There's strong evidence that a substantial fraction of voters will use those intermediate scores with Range Voting. The result is a greater net voter satisfaction, which is the same as saying "more representative elections". Range Voting also helps with problems like the Burr dilemma.
I mean you don't want Kerry losing to Bush because you only gave him 90%. It's always best to maximize the difference between the candidates you don't like and those you do - SO approval voting represents that limit.
So take your Range Voting ballot and only cast 0's and 10's, making it identical to Approval Voting. A lot of other people won't do that, and the result is substantially better election outcomes for both you and society as a whole. Range Voting helps strategic voters more than it hurts those who use intermediate values, and so it causes a net increase in happiness/representativeness relative to Approval Voting. And if more than 60% of voters are sincere, then even the sincere voters are better off than they would have been under Approval.
Now maybe you feel sympathy for those sincere voters in the event that less than 60% of voters are honest. But then I'd have to ask you, why would you feel any less sympathy for random voters? Because if a society switches from Range to Approval, the net happiness goes down. So the only thing you've done is to cause more unhappiness, but distribute the unhappiness more randomly. Then a rationalist might say, "Hey! What have you got against random voters?"
Incidentally, there's other compromises that are interesting. You can mix IRV with E&E-CV with an approval ballot rather than a rank preference ballot..
This is more complicated than Range Voting (and way more complicated than Approval Voting), and almost assuredly has a higher Bayesian regret.
Okay back to Range/Approval, I consider them "mostly harmless" as a single round of voting, no runoff. Voters still have a difficult problem, and different strategies - from "bullet voting" (for one) to "saturation voting" (for ALL acceptable) or something between.
Here is a precise economic measure of the efficacy of those strategies. I note that bullet voting is a horrible strategy (worse than normalized sincerity), and that "saturation voting" is effectively identical to not voting at all.
I accept the conclusion that you should vote in Approval identically as you would in plurality. This is vote for the candidate you WOULD vote for in plurality. THEN add votes for everyone you like better, but didn't vote for because you didn't think they had a chance.
That's technically correct, but practically false, since imperfect knowledge prevents voters from always knowing who the best single candidate would be to vote for. Making that choice is generally easy in plurality voting, since the vote splitting issue leads to two-party domination, and so it's generally strategically best to just vote for your favorite of the two major party candidates (and it's generally just a choice between two options). But with Approval Voting, the field of viable candidates could get rather crowded. So how does one know where to draw an "approval threshold"? A generally good solution is to approve every candidate you like better than the average of how much you like all viable candidates (those who could potentially win).
That means if everyone makes rational choices, Approval and Plurality will always agree in winners. So no reform if you're a smart voter already in plurality.
Well, no. This would only be true if voters were "perfect" at picking the best single candidate to vote for. For instance, a huge group of voters might vote for candidate X, thinking he was the most viable candidate to defeat a dislike candidate, Y, even though they preferred Z over X. So in a plurality election, they might all vote for X. But in an Approval Voting election, they would vote for X and also vote for Z. It could then turn out that Z wins, because Z was actually a lot more supported than many of them realized. Z could have also won in plurality, if they had known about Z's true level of support, and therefore felt safe in voting for him. But say Z was a candidate liked by many, but of uncertain "electability". In a plurality election, he could easily lose just because voters who actually supported him were scared he couldn't win, and therefore scared to vote for him. Range Voting makes it safe to always support one's favorite candidate.
A cute but mostly powerless choice.
Maybe powerless in that case. But think about all the candidates in recent history who have seemed well-supported but not "electable". Approval Voting could have easily changed things for them. Bayesian regret calculations show that it makes a phenomenal difference.
That said I believe in Approving voting for something I called "Variable number of winner polls." Specifically asking a question like "Who do you want to hear in the debate?"... - that's a legitimate use for approval.
An even more legitimate use of Approval is for things like Presidential elections.
Anyway, I don't have too much to disagree with from what I read on the Range Voting website. I won't defend IRV as being "spoiler free", and I accept its weaknesses. I just disagree that approval has much to offer either.
If you think that substantially more satisfied voters, and a substantially more representative government isn't "much to offer", then I'm not sure what you're looking for.
Majority rule isn't that great of an achievement and not clearly better than "strong plurality".
The measure of a voting method is Bayesian regret, or social utility efficiency (two ways of expressing the same thing).
PLUS I know the spoiler factor of plurality is what strengthens parties to unify behind one candidate and what helps leave us with a smaller number of candidates to consider.
Well, parties would have to unify behind one candidate with any voting method. Plurality unfortunately can cause them pick the wrong one. This is a bad thing, not a good thing.
Ultimately plurality is also "mostly harmless", and approval as equivalent.
This is wrong. The logic you have used to make that judgement is flawed, as I explained above. Approval Voting is much better than plurality voting.
More interesting is STV and multiple winner elections, and that's where FairVote wants to go
1) While more "interesting", there is no clear case that proportional representation is better. It's a complex issue.
2) FairVote is making a mistake by supporting IRV, since IRV has historically led to two-party duopoly everywhere that it has seen long-term widespread use (e.g. Australia, Ireland, Malta, Fiji); and since a lot of our races (e.g. mayor, senator, governor) will always be single-winner, and we cannot afford an incredibly poor voting method like IRV just for the sake of getting STV in e.g. Congress. Range Voting is a better stepping stone to better P.R. methods.
3) The U.S. arguably has insurmountable legal hurdles to wide-spread adoption of STV, such as proportional representation being federally illegal for Congress. This means that if FairVote wants to implement proportional representation, they had first better implement a single-winner voting method that disrupts that strangehold.
4) It is important to note that FairVote (namely its head, Rob Richie) has consistently engaged in widespread dissemination of misinformation about IRV. They are NOT a reputable scientifically rigorous organization.
PR has its own detractors who say it empowers the wrong candidates - special interest candidates who aren't interested in the bigger picture of the common good. I won't argue except to say no election method promises good candidates will win, or that voters know what they're doing.
No election method can promise that a good candidate will always win, especially since a good candidate may not always even run. But Bayesian regret measures the statistical goodness of the candidates picked by a voting method, and shows marked differences in quality amongst numerous voting methods.
As far as whether voters know what they're doing, Smith's Bayesian regret figures incorporated ignorance factors to account for just that very problem. As I said, Range Voting handily dominated amongst all commonly proposed voting methods. (There are some theoretical, but generally infeasible, methods that are slightly better, in certain circumstances.)
But Range voting, oh, intellectually interesting for a while, but I can't support it. Sorry!
But you've yet to offer any valid reason for not supporting it. Once you better understand utility theory and Bayesian regret, I believe you will likely come to agree with the Range Voting proponents. The evidence is simply overwhelming.
Clay Shentrup
San Francisco, CA
clay@electopia.org
415.240.1973
Post a Comment
<< Home