This one? Any other voting experiments and papers? I want to read them. But the experiments are about comparing "different voting method" instead of in the evolution system, what's the membership and policies changes toward 2 parties. I know plurality favor extreme policies, but it doesn't explain why there is only be 2 polarities, instead of more. Unless policies themselves are naturally positive correlation in 2 polarities already.
I'll see if I can't dig them up, it's been a while though. Basically though, plurality voting doesn't necessarily lead to polar opposites, it simply leads to two candidates in each individual election, because it encourages strategic voting:
If you have a candidate you like most, and one you like least, plurality voting means that voting for the one you like most can result in a victory for the one you like least (where it otherwise wouldn't matter). For an example, look at how many people blame Nader for causing Gore to lose the election.
It becomes now a question of "who do I most want to win", but "who do I most want to win out of the two most likely to win".
Now, this in and of itself quite obviously leads to a "two party system" for simple strategic stability reasons. It is not always the case, there are periods of transition (the US doesn't have the same two parties it started with, for example), but it means third parties are quickly relegated to the back row if it doesn't seem like they can win here and now - this makes them LESS likely to win in the future (or at least appear to, since they get less publicity, which is the same thing), which further diminishes their power. No one thinks they can win, because no one is willing to take the risk to vote for them and cause their LEAST favorite candidate to win.
Now, there are times when on a national level you can get a different situation because local favorites don't match national favorites, and have a bunch of two party elections making it look like a multi-party "system", and regional differences have long been the cause of new parties rising to and falling from dominance. However, a national election (like president), often gives enough of a push to override that. Without the presidential election working the way it does, congressional elections would likely shake out differently and there might be a number of regional parties as well, but the forced "national party" association involved in voting for president does a lot to discourage that.