IRV better than old runoffs
Marilyn Marks claims an “analysis” of the recent Aspen election discovered a “bizarre problem” with instant runoff voting, known as non-monotonicity. Her claim is quite humorous because she is apparently unaware of the fact that the identical non-monotonic possibility has always been a characteristic of the former two-election runoff system Aspen used prior to IRV.
My own city of Burlington, Vt., has used IRV with great success since 2006. As a political scientist I am quite familiar with the characteristics of various election methods. In a nutshell, in any runoff system (whether instant with ranked ballots, or Aspen's former two-election runoff system) there is a small possibility that in some situations voters with prior knowledge of how all other voters are likely to vote, could be able to help their favorite candidate by strategically voting for a weak opponent in the first round, instead of their true favorite, in hopes of eliminating the stronger opponent for the runoff round.
A recent example of an attempt at this was Rush Limbaugh in 2008 urging Republicans to vote in the Democratic primaries for the candidate he thought McCain would have an easier time defeating in the general election, rather than “wasting” their vote on McCain in the primary. The analysis of the Aspen IRV election showed such non-monotonic strategy games did NOT occur.
The irony is that IRV actually reduces the likelihood of such strategic manipulation because, unlike in a two-election runoff, with IRV a voter can't change his/her first choice between rounds of the vote count, so there is a greater chance of the strategy backfiring, if too many supporters pursue the strategy.
So if this “bizarre problem” is a concern, Marks would do better to support IRV than restore the former runoff system, which has the same dynamic, only worse.