Why you need to unify behind range voting, even if you prefer other voting systems

(Executive summary)

Depend upon it, sir: when a man knows he is to be hanged in a fortnight, it concentrates his mind wonderfully. —Doctor S. Johnson.

IN CHESS, when you are about to get checkmated unless you play RxB, you play RxB. Minor stuff like losing a pawn, or worsening your positional structure, doesn't matter. Only checkmate matters. Chessplayers call this a "forced move." ♕ If you play any other move, you are what chessplayers contemptuously call a "fish."

Well, I've got news for all you voting reform advocates out there. Unifying behind range voting is a forced move.

Minor stuff like the details of whether one of the latest Condorcet variants maybe is the tippy top best voting system and maybe slightly better than range, or whether approval voting maybe has some arguable advantages over range, doesn't matter. The only thing that matters is: can we get a system a lot better than plurality, or not. Keep your eye on the ball. Checkmate matters. ♚ The rest doesn't.

Now there is only one hope for avoiding checkmate in our lifetimes, and that is to unify behind range voting. Let me try to analyse the game situation to explain why that is clear. (Chessplayers call this "God is in the details.")

In order to get a system other than plurality, we need troops. A lot of support worth of troops. I am not speaking here of the military. I am speaking of political activists, people who gather signatures for ballot initiatives, write letters, talk to legislators, know the right people in power and can make them listen. Lawyers. Computer programmers. Pollsters. Journalists. CEOs. People who care, really are boiling mad about what plurality has done to them, because it is one of the most important things in their lives.

Where do you think those hardcore support people are going to come from? Thin air? No. They are going to come from "third parties." Because third parties are the ones being checkmated by the plurality system and Duverger's law. Because if third parties do not get a better voting system that does not systematically prevent third parties from getting off the ground, then they are going to die. And third parties, weak though they may be, still are far vaster in numbers and resources than voting system reform activists. Especially when you put them all together. Unified. All going for the same voting system. Not confused, some thinking system A, some thinking system B, some so confused and deluded they actually think IRV is good for third parties (?!), etc. No. That won't do.

Good. So we have realized that unified third party support is key. Without them, you lose, game over.

Put another way: If you cannot get the third parties to line up behind a voting system, then who the heck do you think you can get?

(I am not saying that the major parties are unimportant. They of course are super important. That is where the Iowa 08 effort comes in. I am just saying that if you do not first have enough raw power in the form of a dedicated core group of supporters behind you, then you cannot get the major parties to do anything and game over. OK? Agree?)

Now. What amazing flag-carrying voting system is going to make the third parties unify behind it? The right answer is range voting.

To confirm that, let us check out the wrong answers (the "fish moves," in chessplayer lingo):

  1. Should the flag-carrier be "approval voting"? No. That is because experimentally, our 2004 poll (#82 here, or see the much quicker data summary) showed, range voting (RV) generates more votes for every third party than either approval (AV) or plurality. Lots more votes. Badnarik(Libertarian) would get 15 times as many votes with RV than with AV. Cobb(Green) would get 2.5 times as many votes with RV than AV. Peroutka(Constitutional) would get 6 times as many votes with Range than AV. Nader(Indep. & Reform) would get 20% more votes with Range than AV. [Calero(Socialist) also looks like lots more RV than AV votes, but not enough data collected to be certain.] This is the only poll addressing approval and range voting vis-a-vis USA third parties, ever conducted. (Understand why this happens.) Do you think third parties are going to want to ignore the only study ever conducted, when making this decision? Do you think, after examining this data, they are going to unify behind Approval Voting? Do you think approval voting is going to be the miraculous flag-carrying system they all want to unify behind? Hello? Have you lost your mind??
  2. Should the flag-carrier be IRV (Instant Runoff Voting)? (Or "BTR-IRV"?) No. That is because IRV leads to 2-party domination, sure as death and taxes, just like the plurality system that has been killing USA third parties for 200 years. How do we know that? Both theoretical arguments 1, 2, 3, 4, 5, and the plain fact that the only three IRV countries in the world, Australia, Malta, and Ireland, each with 40-to-100 years of IRV experience, are 2-party dominated. Do you think third parties, once acquainted with this data, are going to unify behind IRV because they really like living in a 2-party dominance country? Hello?
  3. Should it be the "Borda Count" system? No, this also (and this should be far more obvious and clear than with IRV!) will lead to 2-party domination. Remember: the key is unifying US third parties. They do not want 2-party domination. They want to survive. Remember. Checkmate is the only thing that matters. ♚ Stay focused.
  4. Should the flag carrier be some Condorcet method? (There are many many proposed flavors of Condorcet methods; I happen personally to be acquainted with 50 different proposed such methods and it is unlikely that is all of them and new ones are being invented each year or so.) No. Why not? Numerous reasons. First of all, it is not clear which of the 50-or-more systems is best. I happen to think a few of them feel better than the rest, but this is certainly not "clear" by any means. That makes it hard to unify behind one. Second, even the simplest Condorcet methods are more complicated to explain than IRV, which in turn is more complicated to explain than Range. The more complicated and better Condorcet methods are seriously complicated to explain. That matters because our 2004 poll found that a clearly statistically significant majority of US voters, if given 1 minute to think about it, did not want to switch from plurality to range voting, with the primary reason being that the latter was "too complicated." Hello. If US voters won't go for range voting at least without serious education or exceptionally favorable circumstances, do you think they are going to go for a way-more-complicated-to-explain Condorcet method? I don't think so. I think the chances of pushing that through, are a lot less than the chances of pushing range through. I think the US third parties have enough brains to realize that.

    Keep your eye on the ball. We are trying to unify third parties behind something that is going to save their skins. We do not want to present them with something unattainable as the "cure" for their ills. Systems that the US public will reject are unattainable. If you really want those systems, then get range first. That will prepare (via debate, media coverage, study, good experience with range) the public mind for them to later adopt system X. You have no chance of getting X directly in one stage. Remember, niceties of pawn structure do not matter. Checkmate matters. ♚

    Third, Condorcet and IRV are not doable on a large class of USA voting machines such as mechanical lever machines (used in New York State) and non-reprogrammable machines generally. Meanwhile range is performable right now on every voting machine in the USA, without modification. (And check the link if you are under foolish delusions like thinking most machines are reprogrammable.) The worst you can say is that on some voting machine types range voting is inconvenient. But it is doable. On many machine types such as levers, optical scan, computer-display types, and many punch card types, it is convenient. That again makes range voting more adoptable, and hence better suited as a flag carrier for the third parties to unify behind. Do you really think third parties are going to unify behind a flagship consisting of saying: "oh, USA, we need you to change half your voting machines just for us?" No. I don't think that is a suitable flagship for their purposes. It will sink.

    Although USA third parties have a lot of money, their total amount of money is far less than the amount of money already invested in USA's voting machine inventory. [In 2006, CA's Alameda County decided to spend $13.25 million on Sequoia voting machines. Compare that expenditure in one county alone with the $7 million expenditures on the 2004 presidential campaign activities of all US third party candidates, including Nader, combined.] Think about that when considering the relative power of different groups pushing different aims. Keep your eye on the ball. The third parties do not want to have to push against too many opponents stronger than them. Every single one of those opponents is like a death threat for them. They want as few and weak opponents as possible. They want a system adoptable on all the USA's voting machines right now without modification.

    Fourth, there is good reason to suspect (or see this mathematical proof) that many Condorcet methods will merely lead back again to 2-party domination. Will every Condorcet method do that? We don't know. It has been suggested that maybe methods with both the "winning votes" (not "margins") and "equality-rankings permitted" enhancements, perhaps can avoid 2-party dominance. Maybe so. But more likely not. But the trouble is that the better enhanced Condorcet methods are so complicated they defy mathematical analysis and easy understanding. (For example, despite over 200 years of research, nobody has managed to characterize "optimum voting strategy" in any Condorcet method whatsoever. Since 2-party dominance is a side-effect of strategic voting, how then can we understand it? And human voters will not necessarily use the optimum strategy. They will behave in ways mathematics alone cannot predict and which will change as they use the system more. What will be the result? Nobody knows. Maybe after 200 years more research, we can understand that. But then you will be dead. Keep your eye on the ball. The goal is to do it within our lifetimes.) In the over 200 years since Condorcet methods were invented (by Condorcet), no government anywhere has ever adopted a single one. That makes it hard to tell if they will lead to 2-party dominance.

  5. What about just throwing out single winner elections entirely and going to proportional representation (multiwinner) elections? Yes, that clearly leads to multi-partyism. However, it is not achievable in the USA without a complete rewrite of the constitutional setup. Amending the constitution requires a 2/3 supermajority vote by both houses of the US congress. (Also, within individual states, one could do constitutional amendments. That also requires similar hurdles to be overcome.) Either major party could unilaterally block that forever. At present the total number of third-party congressmen is zero. So evidently the chances of this happening, are zero. Trying to unify the third parties behind a completely unachievable course of action – like a mouse trying to eat a whale – is just not going to work.

So: if you want to get third parties in the USA to unify behind a flagship, are those third parties going to go for something complicated with unclear life-or-death consequences (2-party dominance="death") for them, and which nobody has ever managed to enact anywhere in 200+ years of trying? (I.e. Condorcet?) Or are they going to prefer to unify behind a simple-to-understand method which is clearly known to lead to a tremendous improvement in vote-counts for them and which they can easily see (without need of ultra-fancy math or analysis) does not force 2-party dominance: range voting? Hello? Which is more likely to cause a successful unification? Assess the relative likelihoods, please.

Ok. All clear now? We have, as the chessplayers say, "completed the tactical analysis." We have found that there is a unique forced move if you want voting reform. It is range voting.

More precisely, it is range voting with Xs as intentional blanks.

So even if you prefer other voting systems a little, still your best move is to unify behind range, because that is the only move that can get you there. And you can console yourself with the thought that range voting is a pretty darn good system. I can't prove to you it is the tippy top best system that ever was or ever will be proposed. But it is clearly within the top few percent.

Range voting is very robustly the best among about 30 systems tried in my big comparative Bayesian regret study in 1999-2000 (#56 here). OK, maybe you can attack that. Maybe you can say I did not put in your favorite system or favorite voting strategy. (Some of the systems I am being attacked for, were not even invented at the time I did the study.) Those attacks don't matter: The bottom line is, you should now be convinced range is pretty darn good as far as quality is concerned. It is also among the simplest systems. And it does not seem to exhibit any major exploitable weaknesses since it scored top in every one of 720 different parameter setting-combinations tried in the study, which covers a heck of a lot of behavioral ground.

So in chessplayer lingo, making the move to unify behind range is actually quite good (perhaps not tippy top best, but quite good) as far as the niceties of pawn structure are concerned. But what looms over and trumps everything else is checkmate. Keep remembering that. The only move to avoid checkmate is to unify behind range voting. ♚

Return to main page