slashdot should totally have a post-by-phone option

Of course, there are some issues with instant-runoff voting.

For example, there's a (fictional) example used on Wikipedia to compare various voting methods: choosing a Tennessee state capital from among the four cities of Memphis, Nashville, Chattanooga and Knoxville. It's not unreasonable to think that most voters would rank these in increasing order of distance from their towns, resulting in the following:

Those near Memphis (42%) voting: Memphis, Nashville, Chattanooga, Knoxville.

Those near Nashville (26%) voting: Nashville, Chattanooga, Knoxville, Memphis.

Those near Chattanooga (15%) voting: Chattanooga, Knoxville, Nashville, Memphis.

Those near Knoxville (17%) voting: Knoxville, Chattanooga, Nashville, Memphis.

In instant-runoff voting, Chattanooga would lose in the first round, transferring its votes to Knoxville (which would then have 32% of the vote). In the second round, Nashville loses and Knoxville gets its votes (for a total of 58%) and wins the election.

However, notice that the majority of voters (58%) prefer Nashville to Knoxville. Indeed, Nashville beats out Chattanooga and Memphis when compared one-on-one as well. That is, Nashville is preferred to each other city by a majority of the people. Yet instant-runoff fails to elect it.

Another problem is that it doesn't really eliminate the spoiler effect that is present in a plurality election (i.e., what we have now), just postpones it.

Consider major party candidates A and B with third-party contender X. Assume that X is closer to B in ideology and thus most voters will rank them adjacent to each other.

A's supporters (percentages are in terms of the entire electorate):
30% A,B,X
15% A,X,B

B's supporters:
15% B,X,A
10% B,A,X

X's supporters:
30% X,B,A

Round 1: B is eliminated (only 25% have it as first choice). In the second round, A has 55% and X has 45% so A wins. Notice that B is preferred to A by 55% of the populace and to X by 55%.

By listing X over B, X's supporters caused their last choice (A) to win. That is, reducing the ranking of their favorite candidate and putting it in second place would create a more favorable result (from their perspective) since B would win.

So, the bottom line is that once a third-party becomes strong enough to survive the first round but not yet strong enough to win, voting for it in top position can contribute to the election of one's least favorite candidate. Once this fact becomes well-known, many voters will learn (or be told) to avoid putting their preference of a third party candidate as first choice unless they can be reasonably certain that it will win (i.e., the same thing that happens with many third party voters with the current system of plurality).