I agree with Marla that our presidential system is quite good; much better than the US one IMHO (although I like the US system of primaries). However, Concordet would be a lot better. IRV would be a disater though.
Hitro, I believe that your centrist exemple is somewhat flawed. First, because if a good part of the left-wing and right-wing voters believe that only their own candidate will win and that the centrist candidate is either an exploiter-in-disguise (to left wing voters) or a socialist-in-sheep-clothing (to right-wing voters) and thus only vote for their own candidate, the centrist no longer win. Consider:
20%: left
26%: left > centrist
5%: centrist > left
5%: centrist > right
24%: right > centrist
20%: right
The left-wing candidate now wins. In other words, for the centrist candidate to win, he must truly be acceptable to both side. And if that is the case, then he can beat both of the other candidates in a "one on one" election, and it is normal for him to win.Plus, Condorcet is the only method I know of which would properly handle the rise of a strongly left or right party in a two-party system. Consider a US state with 45% of republican voters and 55% of democrat voters (replace democrat with republican and green with libertarian if you prefer). Such a state would elect a democrat, 55-45. Normal. Now consider that there is a move to the left of some democrat voters, cauisng them to switch to the green party. Normaly, such a move to the left should certainly not cause the state to elect a republican instead of a democrat. However, in the current US plurality system, as soon as the support for the greens rise above 10%, the state elects a republican governor - as a direct consequence of the move to the left of some voters! Clearly, this is nonsense.
A two-round system or a an instant runoff in this case both produces the same result (assuming that all the new green voters still prefer a democrat governor to a republican one, and that democratic voters are equaly split between centrists prefering a republican to a green and left-wing prefering the reverse): as long as the green candidate get a lower score than the democratic one, the democrat continues to win. However, as soon as the green candidate get over 27.5% of the vote and thus beat the democrat in the first round, the republican wins, since he get half of the democratic vote in the second round. In other word, the rise of the green party has once again caused the election of a republican candidate. The only change is that it happens latter.
In a condorcet system however, the rise of the greens will NEVER caused the election of a republican. The democrat candidate will continue to be elected, until the support for the green is strong enough to allow him to be elected.