Instant Runoff Voting; WHat do you know?

[Matt Collins]

I am looking for opinion and ideas about instant runoff voting. Does anyone know anything about it good or bad?


Thanks!

[mport1]

IRV is much better than our current system - http://en.wikipedia.org/wiki/Instant-runoff_voting, http://www.chrisgates.net/irv/

I'd prefer no voting though.

[RonPaulVolunteer]

Used in Australia and New Zealand to great success.

[OferNave]

I believe the best (mathematically) is the Condorcet Method (http://en.wikipedia.org/wiki/Condorcet_method). It's similar to Instant-Run Off.

The Free State Project used the Condorcet Method to pick the state to move to out of ten candidates when they hit 5,000 pledges.

We also used it one time to vote for cast leader at my Rocky Horror Picture Show cast. That caused a bit of confusion, but it was very educational, I think. :)

[Matt Collins]

Used in Australia and New Zealand to great success.Any links or news articles about this?

[nelsonwinters]

Check out http://instantrunoff.com.

I think that this along with

*Open Debates

*Ballot Access and

*doing away with the current campaign finance restrictions

These are critical for evening out the playing field. I think that having an instant run-off system may be the most important aspect. People figure why even bother learning about a third party candidate if they don't even have a chance to win. Actually, how horrible it would be to learn about about a third party candidate and strongly support that candidate knowing there's no chance that they can win. Ignorance is bliss.

[Mahkato]

Fair Vote Minnesota (http://www.fairvotemn.org/) is a pretty good resource.

IRV would be the single greatest measure to give prominence to the third parties, which is why it has not been implemented.

[fgd]

I cannot say this strongly enough:

IRV is possibly the worst and most problem ridden election system ever devised - even worse than plurality voting. It gives non-monotonic results in a lot of cases, and generally is completely flawed. Using it as a voting method should be a crime. I'm completely serious.

Approval voting and range voting are much much better alternatives.

http://zesty.ca/voting/sim/

[bdmarti]

I don't think most versions of instant run-off voting are that great.

They may be slightly better than what we have, but most variants do tend to encourage a 2 party system. I don't recall the exact numbers, but I believe that around 20% of possible instant run-off vote combinations result in the "wrong" candidate winning. Some variants of instant run-off are even worse.

While you can find these systems used around the world, I don't think they are that successful. Take a look at the countries where these systems are used, and see if competitive races haven't resulted in 2 party dominance.

This isn't a proof, but just an illustration of what happens...you can find proofs and massive simulations on the subject if you really want.
Say we have candidates A, B and C and we have 100 voters (to keep things simple).
Voters vote like this: (A/B/C means a is first preferred, B second...)
A/B/C - 20votes
A/C/B - 11 votes
B/C/A - 11 votes
B/A/C - 18 votes
C/B/A - 35 votes
C/A/B -5 votes

B gets 29 first choice votes and is eliminated and his second choice votes get applied...even though candidate B is the first or second choice of 84% of the voters he loses the election and candidate C, who is the first or second choice of only 62% of the voters wins. This type of result can be expected around 1/5 (or more) of the time...and as people come to understand this, they polarize their votes and we end up with 2 dominant parties just like we have now.
Again...this isn't meant to be a proof...just a sample of the type of thing that happens with instant run-off. (edit for math errors)

A better system of voting is to simply assign numeric values...say a score of 1 to 10 to each candidate. You could give 10s to all candidates you like, or you could give 10 to B, 5 to A, and 1 to C or whatever you like. The candidate with the most points wins. If you are afraid of C winning because you don't give A and B enough points, you can give them both 10. While not perfect, this type of system eliminates some of the problems with 2 party dominance. It also has a couple of major benefits in that it can apply to multiple seat races without problem and it can be used by existing voting machines much easier than instant run off can.
For instance if you had a ballot with space for 5 bubbles next to each candidate, you could have a 5 point scoring scale for an arbitrarily large number of candidates...but you could only use this ballot to instant run off at most 5 candidates...and 4 bubbles means only 4 candidates, but still no limit for the scoring system.

I also think "none of the above" should always be allowed as a vote in any single seat race. Sometimes the candidates running just suck.

[bdmarti]

yeah...what fgd said while I was typing. go to that site.

here's a good site discussing the method I was describing:
http://rangevoting.org/

[Mahkato]

Thanks.

[Matt Collins]

Interesting replies. Any other ideas?

[CUnknown]

IRV is great and much more supportive of 3rd parties than our system, which is heavily biased to 2 parties.

Perhaps IRV is not perfect, and instead a system such as the one used to pick MVPs in major-league baseball should be used. The ranking system is still used, but there aren't any runoff votes. Each 1st place vote is worth a certain # of points (maybe twice that of a 2nd place vote, for example), and the player with the most points wins.

Taking bdmarti's example, and lets say that a 1st place vote is worth 3 points and a 2nd place vote worth 1 point:

A/B/C - 20votes = 60 points for A, 20 for B
A/C/B - 11 votes = 33 for A, 11 for C
B/C/A - 11 votes = 33 for B, 11 for C
B/A/C - 18 votes = 54 for B, 18 for A
C/B/A - 35 votes = 105 for C, 35 for B
C/A/B -5 votes = 15 for C, 5 for A

A gets 60 + 33 + 18 +5 = 116
B gets 20 + 33 + 54 + 35 = 142
C gets 11 + 11 + 105 + 15 = 142

A is the clear loser now, and B and C tie, wow.