I find that IRV is a shell game. It basically creates the illusion that your third-party vote counts. If you choose a Green, say, for your first choice, IRV will increase your candidates odds. But, not by much. The Dems and Reps will still dominate the outcome.

The system makes you 'feel' like your votes count because your second or third choices will most likely be Dems or Reps. One of those two will win. Hence, you 'feel' as if your vote mattered when in fact, your true voting affection (The Green) only winds up with a slightly better chance of actually winning.

A better way would be to use Commulative Voting in multi-member districts like Lani Guinier proposed. In that, 'super districts' composed of four or five representatives each replace the single member districts now in place. If there are five seats, then the voter gets five votes to allocated amongst the candidates s/he wants. The top five vote winners get seats.

So, if you want a Green to actually win, you'd pour all five of your votes into the Grean candidate. Or, if you want to hedge, you can give 3 votes to the Green and 2 for the Dem.

For absolutely single-winner races (most executive branch offices like Governor or President fit this bill), then IRV is prefereable to the current system in that it will increase voter turnout. But the smarter voters will do the math and recognize it for the con game that it is.

As I understand it, Arrow's theorm shows that there is no good voting system. http://www.electionmethods.org/Arrow.htm so I am willing to accept the current system. On the other hand I would suggest two other modifications:

1) I'd like to see some races, say the Senate, where instead of voting on geographical lines we voted for the top 100. That way the "greens" would be able to get one or two voices perhaps.

2) I'd like a MANDITORY "None of the Above" on all votes and if None of the Above won a plurity, then the top 2 other candidates would not be able to run in the followup.

It's a great idea. It's hard to disagree that Nader "stole" the election from Gore in 2000 by getting 1% or 2% in a few close states. Now that he's running again, he may do the same in 2004.

Who cares if people are under the false impression that their third party vote will count? At least their vote will be counted somewhere in determining the winner when it comes down to the top two candidates.

Unfortunately, we won't be able to implement this until we get rid of those terrible punch cards.

I don't think that Arrow's theorem is really relevant here. We are not interested in technical issues in social choice, we just want something that will work better than the curent plurality system most of the time, given all the problems with the way people actually vote.

Since most voters do not seem to understand strategic voting (e.g. that they shouldn't vote for Nader if they would rather have Gore than Bush), then IRV will better represent the majority view.

IRV will also reveal a lot more about people's actual preferences, which is very good if we care about social change. For example, if it turns out that 25% of Americans would vote Green but are now "afraid" to, we might learn this under IRV, as voting Green first is (almost) costless.

SmartBoy's comment doesn't demonstrate that IRV wouldn't be desirable. It doesn't falsely claim that "your vote counts." Rather, it accurately claims that "your vote won't be wasted." And it prevents minor parties from functioning as "spoilers."

PS: What I mean is that under IRV your vote would count in determining who is the winner between the last two men standing.