No. The "one man, one vote" ideal really is that no voter should have more power than any other. That is just as true for Range as it is for any other fair system.
And look at it this way. Suppose candidates A and B are tied. Great! You can break that tie and have an impact on the world, with your vote. Well, with range voting, you are exactly as capable of that as any other voter.
And further, Range Voting actually gives more equal power to voters than plurality or IRV, since under plurality you cannot break the tie between A and B if you unfortunately voted for C, making you less powerful than a voter who votes A or B. (Supporters of third parties are "zeroed out" and their votes never have an effect. Fair? I don't think so.) And under IRV voting, your vote, even though it does express a preference A>B (say) still can have zero impact on a crucial A-B tie if your top-rank vote was for somebody else at the time.
So actually, the problem is not that range fails to satisfy one-man-one-vote. In fact, it satisfies it better than plurality and IRV. (Information about the US Constitution vis-a-vis range voting.)
They can if they want. (Just give 99 to one candidate and 0 to all the others.) But apparently most of the time, most voters won't do that. In our 2004 US presidential election range-voting pseudo-election study (#82 here) the number who voted that way was less than 24%.
No!! You can put any numbers between 0 and 99 you want, for every candidate. Or blanks. Please totally disregard what they add up to. Your job is just to rate each candidate. If you wanted to give every single one of them a 99, that would be permitted by the rules (although silly).
No! That non-recommended system is called "cumulative voting" (CV). Unfortunately, in CV, strategic voters who want to avoid "wasting their vote" will always "put all their eggs in one basket," that is, give 100 to the same person they would vote for with the old plurality system, and 0 to everybody else. (Their point is, giving any part of your vote to the third-party candidate, with CV, is "wasting" that part of your vote. You can't afford to be that stupid because the other side won't be – their thinking continues.) So CV with strategic voters would not be any improvement over plurality voting at all! (However, in practice, some voters would be honest rather than strategic, and then CV would probably be an improvement. But even then there would be a lot of voters who added wrong and whose ballots were invalidated as not summing to exactly 100. Why ask for those problems?)
Meanwhile with range voting, giving an honest score to a third party candidate wastes nothing. It does not "use up" any part of your 100 "power" units because there is no requirement the scores add up to 100. You still have full power to give any score you want to any other candidates you want.
Answer: a "vote" is an expression of your opinion. If you express the opinion A=99, B=C=D=0, that is an equally valid and equally powerful vote as A=B=C=99, D=0. The entire set of numbers should be thought of as a single "vote." You have to get away from the bad, plurality-thinking, old-notion of what a "vote" is.
Furthermore, it is silly to think you by "voting for more candidates" have "more power." For example, if you were to vote the maximum for every candidate, that vote would be completely powerless.
You're not splitting, since "splitting" implies diminishing somebody's vote, which does not have to happen with range voting.
You voting for two guys can help the rest of us in picking just one guy. If you think both guys are equally good, why should you be forced to pretend one is better? If you unluckily guess wrong, you might miss your chance to elect somebody good! The bottom line is, with range voting, your option of "voting for more than one guy," as you put it, gives you a better chance of getting somebody you like, elected.
First of all, with range voting, casting an honest vote can never hurt you versus not voting at all. If your vote says A=99 and B=98, that will never cause B to be elected if A would have been elected without you voting. (It could, however, elect B if, without you voting, C would have won.)
However, nevertheless, if it comes down to an A versus B battle. you might end up wishing you'd given B zero! But, in every other circumstance, it helps you. E.g, it might come down to B versus somebody else (C), in which case you'll be happy you gave B a high score with range voting!
You decide how to apportion the points and trade off the risks. You are free to give B zero if you feel that is your best course, but also free to give B 90 if you feel that is the best course. With plurality voting, if you just voted for A, that'd be totally useless for you if it came down to B versus C, so you're tempted to dishonestly vote B – hurting your favorite A. With range, an honest vote can never hurt you versus not voting at all – voting, say, "A=99, B=90, C=0" always helps A against everybody, and always helps B against everybody besides A. It is just a matter of deciding how much you want to help each one versus each other.
Look at it this way. In a 10-candidate race, you with an honest range vote are helping the candidate you like most against the one you like less in each of 45 different pairwise competitions – and you can decide how much oomph to allocate to each one. But with an honest plurality vote you are only helping one candidate, and affecting only 9 pairwise competitions – and quite often your best move is actually a dishonest plurality vote, which helps in some of those 9, but actually hurts in others. For what do you want to weaken yourself that way?
The most obvious ways are on our ways to help page.
A famous theorem of Gibbard and Satterthwaite states, roughly, that every "deterministic voting system" has the property that, at least sometimes, "honest votes" and "optimally strategic votes" are not the same thing. In other words, yes, "strategic voting" does pay and it can still pay with range voting, as with any other voting system.
Range voting, however, reacts comparatively mildly to "Bulgarian judge" strategic voting. In fact if everybody strategically exaggerates, i.e. rates every candidate as a full 10 or 0 on (say) an 0-to-10 scale, then range voting simply turns into "approval voting" which is where you vote for the set of candidates you "approve of" and not for the rest. (I.e., approval voting is like the present plurality system but with "overvoting" now permitted.) That is not a bad reaction.
Strategic voting is not a problem per se. The problem is if voters who try to strategize tend therefore to cast dishonest votes which misorder candidates. That basically does not happen with range voting.
For example, suppose you honestly prefer Nader top (and Gore second), in the Nader vs Gore vs Bush 2000 US presidential election.
Incidentally, skating can solve the Bulgarian judges problem by discarding the max and min votes, then using range voting. However, that is not a good solution for use in elections, which have some important differences from the Olympics. In the Olympics, encouraging judge-conformity and having all judgements=votes equal is good. In elections, voters differ a lot, and we want that diversity – i.e. forcing voters all to be the same is bad. Discarding (say) the 10% "most extreme" voters would be an outrage, which might in fact cause many little-known third parties to get zero votes!
The reason range voting here reacted well is that our Range Voters, although exaggerating, were not very dishonest – they did not actually change their ordering of their candidates e.g. to rank Nader below somebody. Indeed it may be shown that it is never strategically wise for a range voter to rank his true favorite below anybody else, and if there are 3 candidates (here "Bush," "Nader," and "Gore") it never is strategically wise for a range voter to misorder X>Y when he truly feels Y>X. Plurality, Condorcet, and Instant Runoff Voting cannot make either claim.
Also, note, our computer simulations of elections involved honest and strategic and "ignorant" voters in a wide range of scenarios, and in all such scenarios (over 100 tried) range voting was the best voting system measured by "Bayesian regret" (versus about 30 other systems). In other words, strategic voting was not an "oversight" that we "forgot" to consider. Instead it was a "central consideration all along in both our computer studies, and that underlies the whole idea of range voting."
There are two main schools of thought about this. The first school is that "votes are just moves in a game, and they have no intrinsic meaning aside from those that logically follow from the rules of the game. Just try to play the game by making the moves (vote) most likely to achieve the goals you want." Here the "game rules" are just the method of determining the "winner" from the "votes."
So a range voter from this school might think something like this: "My top goal is to elect Bush. So first of all, I'll give Bush a 99, which according to the game rules, gives Bush the maximum chance." Then the voter might continue "My second goal is: if Bush cannot be elected, then I want to elect Kerry." So then you might also give Kerry a 99 to maximize his chances. However, if you do that, it could come into conflict with your main goal of electing Bush. So there is a tradeoff you have to reckon between your two top goals. So you might want to give Kerry some score below 99. (Which actually would be more honest of you.) So then you say "And my third goal is..." and so on.
If you are then wondering "what is the difference between a score of 60 and a score of 50?" – this school's answer would be (ignoring "blanks") that five 60-scores will have the same effect as six 50-scores toward making that candidate win, so that is the meaning.
The second school is that there is a genuine range vote meaning we would like to recommend. That is this. Give the best candidate 99 and the worst 0. Now, for the remaining candidates, put them on the right positions on the 0-99 scale based on your happiness. For example, if your estimated lifetime happiness after Gore were elected would be 60 on a scale where 0 is the worst candidate and 99 is the best candidate, then give Gore a 60. Notice: if
Also: if you are ignorant about a candidate, you could leave that candidate's score blank, causing no effect on his election chances. On the other hand, a lot of people feel that if they have not heard anything about a candidate, that is a demerit and in fact they would prefer to give him a zero, not a blank. (Or perhaps a low score, like 25, but not zero.) If you feel that way, go right ahead – we leave the decision up to you.
That is pretty good. For example, under plurality voting, even if all voters were fully honest, fully informed, and exactly correct in their happiness-assessments for every candidate, and the worst-to-best happiness difference was the same for every voter, and if every voter voted for the exactly best candidate in his view, then there unfortunately still would be absolutely no reason the best candidate for all of society (maximizing total happiness) would be elected. That is because plurality is a bad voting system. In fact it is entirely mathematically possible that plurality voting would elect the worst candidate minimizing total happiness!! So: range voting is clearly going in the right direction.
But unfortunately, neither (1), (2), (3) or (4) is going to be exactly true in reality and therefore range voting (and every other voting system) will sometimes elect winners not best for all of society. We could make (4) more true by increasing the range to, say, 0-99999. We could evade the problem of (3) if voters who cared 3 times less than the voters who care the most, voluntarily rated candidates on a 33-wide subscale of the full 0-99 scale. Many voters will strategically exaggerate (disobeying assumption 1) but even in the presence of that kind of voters, computer simulations indicate that range voting still seems to do as well or better than all other commonly-mentioned voting systems.
However, you are free to choose your own answer to the question of what your vote means.
The questioner continued by noting that range voting was basically the method used by schools grading students 0-100 on tests. The student with the highest average becomes valedictorian. But: maybe Professor Smith has a different view of what grades mean, than Professor Jones? Isn't it unfair to make Sally the Valedictorian just because her professors were easy graders?
Response: That analogy was incorrect. The correct analogy would be if Professor Smith and Professor Jones both graded everybody's homework independently, and then the student received as her grade, the average of the two grades awarded by both professors. Of course, because schools do not have enough money to pay professors to do that much work, that does not happen. But if they did have that much money, then that wouldn't be "unfair." It would actually be "better and more fair."
With range voting, every voter grades every candidate. If some voters (professors) grade differently than others, that is fine. It does not hurt. It is not unfair. In fact, such diversity is good.
One reason is that all three (with the possible exception of some types of Condorcet system) lead to self-reinforcing 2-party domination, which was one of the problems with the present plurality system. I.e. none of them fix that problem. (For "strategic" reasons, almost nobody votes for third parties in the plurality system, causing them to die off, causing us voters to be stuck with at most 2 viable choices! Bad scene! Not very "democratic"!)
Another reason is that IRV and Condorcet simply will not work with a lot of today's voting machines. So if you want an easy switchover, forget them.
Another reason is all three are more complicated than range voting. Finally: Borda reacts very badly to "candidate cloning" election-manipulation attempts, and Condorcet reacts very badly to "strategic voting". Click the links on the main page for these rival voting systems to find out more about them. IRV, because it has many "rounds," each one of which could be a tie or near-tie like the Gore vs. Bush Florida 2000 nightmare, has vastly larger chance of incurring such a nightmare than does either plurality or range. (Range, on the other hand, because each vote consists of numbers like 99 larger than just 0 and 1, incurs vastly smaller chance of a tied-election nightmare.) As an election supervisor or voting reformer, which way would you prefer to go: much more chance of lawsuit-slinging, chad-examining nightmare (IRV), or much less chance of nightmare (Range)?
Experimentally, range voting has a much gentler effect on small "third parties" – it "coddles" them protectively in a "nursery" which allows them to grow in size. The nursery effect is absent in approval voting, i.e. AV gives much smaller vote counts to small third parties. It is possible this effect is so huge that third parties will simply be unable to grow out of infancy under approval voting, in which case AV would still leave us in a state of 2-party domination. One clear implication of this is that voting reformers will be unable to get unified third-party support if they push AV, preventing voting reform from happening because there will be no core group of do-or-die supporters of the reform. But with range voting, reform is not prevented.
Approval voting tends to produce worse-quality winners than range voting especially when there are a large number of candidates and the voters' considerations are comparatively highly dominated by "honesty" and less by "strategy." That most-best-for-range situation is precisely what we expect to happen in Iowa caucuses.
It seems to us that the Iowa 2008 caucuses are a perfect starting place to replace plurality voting with the superior range voting system. That's because every one of the following groups is motivated to want range voting:
plurality voting got me elected. Why should I want to change to range voting? Why should I want to let "third parties" have more of a chance at my seat? See Answer!
Return to main page