Hacker News new | ask | show | jobs
by brianolson 3266 days ago
Instant Runoff Voting is severely flawed. It can solve basic problems like the Bush-Gore-Nader election where Nader voters probably would have preferred Gore, but it can also get the wrong answer and has in a real world election in 2009 in Burlington, VT: http://bolson.org/~bolson/2009/20090303_burlington_vt_mayor....
1 comments

I find this example kinda 'eh'. FPTP also elects a 'wrong' winner, and it's not clear to me that the IRV is substantially wronger.

I agree that there's probably a better way, but I fear that it could be at the cost of either ease of use or methodological clarity. Ranking candidates is relatively intuitive, and so is sequential elimination with votes reassigned to the next preference.

I recognize that you could transform the rankings people provide into head to head match up, but if the methodology is too opaque to the layperson, it's going to breed distrust in the system, and I fear that that would be overall worse (if it discourages participation) than having a minority of elections be won by the 'wrong' candidate versus better approaches.

I recognize that this could also be an argument in favor of FPTP. I do think there's a continuum here of conceptual difficulty, and I think finding the sweet spot of correctness vs intuitiveness is tricky.

IRV is less wrong than pick-one; but more wrong than a dozen other election algorithms. If you want to read about one, read about Condorcet.