> In fact other systems actually encourage cooperation and favor candidates that are more in the center.
Actually, single-round winner-take-all democracy induces a race toward the middle, not polarization. (If everyone is aligned on a 1D spectrum and votes for the candidate closest to their position, then both candidates are incentivized move toward the median position.)
What you're really getting at isn't the idea of polarizations versus moderation, it's voting coordination (i.e. political parties) aka collusion. Unfortunately, alternate voting systems don't eliminate collusion, they just push it to a different level (e.g. proportional representative systems still have 51% coalitions at the level of parliament). You might reasonably argue that that's a better level to put that collusion, but it's definitely not eliminated through some simple choice of voting system.
If you really want to eliminate collusion and induce consensus making, you can consider "randomized parliaments". They operate in a way very analogous to the mechanism that makes poker showdowns rare. Although he chooses an unfortunate name of "quantum parliment" (since there's nothing quantum about the randomness), Aerts explains the idea well here: http://arxiv.org/abs/physics/0503078
Arguably, what we've seen happen in practice is both a race toward the middle and polarization - the two parties and their supporters become more and more hostile, aggressive and unwilling to compromise over an increasingly small set of differences between the two.
Yes, I think that's right. I was really criticizing the idea that winner-take-all deomocracy is not a system that "favor[s] candidates that are more in the center". "Polarization" was a poor choice of words; I think "extremization" would have been better.
It's only a race to the middle in the short term. In the long term, you tend to get two parties as a defense against vote splitting.....that's where the polarization comes from. The collusion causes it. That's not to say that collusion always causes polarization, but this particular type does for very direct mathematical reasons. (see Duverger's Law)
Proportional representation is not a great solution, in my opinion, since it actually assumes parties rather than making them unneeded.
That's not to say there won't be collusion under a different system -- there will and that's ok...but there won't be a particular type of collusion whereby the Nash Equilibrium is two diametrically opposed camps.
The main point, though, is that the rules are important. People cooperate better if the rules reward cooperative behavior.
Actually, single-round winner-take-all democracy induces a race toward the middle, not polarization. (If everyone is aligned on a 1D spectrum and votes for the candidate closest to their position, then both candidates are incentivized move toward the median position.)
What you're really getting at isn't the idea of polarizations versus moderation, it's voting coordination (i.e. political parties) aka collusion. Unfortunately, alternate voting systems don't eliminate collusion, they just push it to a different level (e.g. proportional representative systems still have 51% coalitions at the level of parliament). You might reasonably argue that that's a better level to put that collusion, but it's definitely not eliminated through some simple choice of voting system.
If you really want to eliminate collusion and induce consensus making, you can consider "randomized parliaments". They operate in a way very analogous to the mechanism that makes poker showdowns rare. Although he chooses an unfortunate name of "quantum parliment" (since there's nothing quantum about the randomness), Aerts explains the idea well here: http://arxiv.org/abs/physics/0503078