News for hackers. Stuff that matters.

A comment on 512 Paths to the White House. Parent comment

Re:512 Paths to the White House by baddox on Saturday November 03, 2012 @06:13AM (#4736311)
There are problems with FTPT, but all voting systems are simply compromises between certain mutually exclusive criteria, all of which are generally desirable. You can't have all the desirable characteristics at the same time, as demonstrated by Arrow's impossiblity theorem and the Gibbard–Satterthwaite theorem. Sure, FTPT has some disadvantages, but trading it for something like instant run-off voting introduces some new disadvantages. It's insufficient to argue that FTPT is strictly inferior. You really just have to argue for which criteria you think are more important. The third link is a great comparison of many different voting systems and criteria.

http://en.wikipedia.org/wiki/Arrows_impossibility_theorem

http://en.wikipedia.org/wiki/Gibbard%E2%80%93Satterthwaite_t...

http://en.wikipedia.org/wiki/Comparison_of_instant_runoff_vo...

[ Parent | on HN ]
Re:512 Paths to the White House by harshreality on Saturday November 03, 2012 @11:23AM (#4736784)
FPTP and IRV for single-winner elections are completely compromised, due to tactical voting (and lack of fairness to third parties) in common scenarios. I don't think you can support your claim that other systems are just as bad. In every informed discussion of voting systems I've ever encountered, FPTP and IRV are considered inferior to voting systems like Condorcet variants [1] or Range Voting [2] [3].

Preference between condorcet and range voting is a lot more murky.

http://en.wikipedia.org/wiki/Voting_system#Compliance_of_sel...

http://rangevoting.org/rangeVcond.html

[1] http://en.wikipedia.org/wiki/Condorcet_method

[2] http://en.wikipedia.org/wiki/Range_voting

[3] http://rangevoting.org

[ Parent | on HN ]
Re:512 Paths to the White House by baddox on Sunday November 04, 2012 @03:23AM (#4739272)
Unfortunately, I believe all voting systems are subject to tactical voting, because of Condorcet's paradox. Please correct me if I'm wrong.

edit: I forgot to mention that all voting systems are subject to tactical voting, except for ones which are dictatorial or non-deterministic (and those are generally considered nonviable systems).

[ Parent | on HN ]
Re:512 Paths to the White House by harshreality on Sunday November 04, 2012 @07:42AM (#4739620)
Range voting at its worst turns into approval voting. RV is not an ordered list voting system, so Arrow's theorem, the Gibbard–Satterthwaite thorem, and Condorcet's paradox do not strictly apply.

For ranked ballot voting systems, there's a difference in what kinds of tactical voting are possible, and in how likely it is for tactical voting to improve your perception of the outcome (baysian regret comparison of honest vs tactical voting). Ranked voting systems are not all equivalently bad simply because they're each subject to some kinds of tactical voting.

I think Condorcet is promising (among all the ranked-ballot methods) because the Condorcet principle is easy to agree with. The choice of cycle-breaking method is where the fault lines appear, but new cycle-breaking methods could attract a consensus.

[ Parent | on HN ]
Re:512 Paths to the White House by studentrob on Saturday November 03, 2012 @06:46AM (#4736352)
Most important to me would be a condorcet winner, something we do not have in FPTP (ftfy)
[ Parent | on HN ]
Re:512 Paths to the White House by baddox on Sunday November 04, 2012 @03:21AM (#4739265)
Why do you value the Condorcet criterion so much, especially since a Condorcet candidate might not even exist?
[ Parent | on HN ]