4 ms·

Well, no voting system is perfect. See http://en.wikipedia.org/wiki/Arrow%27s_impossibility_theorem http://en.wikipedia.org/wiki/Arrow%27s_impossibility_theore

by philipn 12y ago

Well, no voting system is perfect. See http://en.wikipedia.org/wiki/Arrow%27s_impossibility_theorem http://en.wikipedia.org/wiki/Arrow%27s_impossibility_theorem. IRV is pretty good, and is being slowly adopted across the country. In particular, IRV satisfies the important "majority criteria," e.g. if the majority of people vote for a candidate, that candidate should win. Approval / range voting fail this criteria. See, again, Arrow's Theorem. http://archive.fairvote.org/monotonicity/ http://archive.fairvote.org/monotonicity/. Additionally, voting so as to strategically take advantage of non-monotonicity using IRV/STV/Ranked Choice Voting is NP-Complete- see http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.127.97 http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.127....

aninhumer 12y ago
Approval/Range voting fail the majority criteria in that they allow a candidate who is the second preference of 90% of people to beat a candidate who is the first preference of 51% of people. Is that necessarily a bad thing? Your link also discusses the later-no-harm criteria for Approval/Range leading to strategic voting, but all the examples I've seen for this seem to talk about giving a marginally preferred candidate an advantage in the absence of any truly disliked candidates. I think in practice, people will be far more interested in keeping disliked candidates out than furthering their marginal preferences. Meanwhile, IRV suffers from Favourite Betrayal, which is a much more obvious incentive for strategic voting IMO. (Your link about NP-completeness of strategy only talks about STV.)
- ClayShentrup 12y ago
  This issue has been solved by math PhD's and political scientists who've studied the matter for decades. www.electology.org/approval-voting-vs-irv IRV can elect X even if Y was preferred to X by an arbitrarily huge majority and got more first-place votes than X. www.electology.org/core-support Regarding favorite betrayal, you might enjoy this video. https://www.youtube.com/watch?v=JtKAScORevQ https://www.youtube.com/watch?v=JtKAScORevQ
- philipn 12y ago
  In a range voting election, you'd have a lot of people who would bullet-vote for a particular candidate (100/100 points for candidate A). Then possibly lots of other people who would have to think about how people may be bullet-voting for those candidates, and then think about how they might distribute their points so as to ensure those candidates don't win. If I love candidate B and C, kind of like D, and hate candidate A, but all of candidate A's supporters are bullet-voting, how should I vote? It's ridiculous. It doesn't allow voters to express their preferences without strategic voting -- it's totally baked into the process. The range voting folks have come up with various metrics that say their system is the best, but completely ignore the most obvious reasons why its flawed. For the record, though, I'd vastly prefer any system over the current plurality elections (including range voting).
  - aninhumer 12y ago
    >then think about how they might distribute their points Just to be clear, you're not imagining people have to share their vote between parties are you? Range voting means you can give every party 100 if you want to. >If I love candidate B and C, kind of like D, and hate candidate A, but all of candidate A's supporters are bullet-voting, how should I vote? If you really hate A so badly that you think any one of B, C or D would be vastly better, then go ahead and give them all 100. If A not getting in is that important to you then your "strategic" vote is in fact, an accurate representation of your preferences.
  - ClayShentrup 12y ago
    In a range voting election, you'd have a lot of people who would bullet-vote for a particular candidate www.electology.org/bullet-voting > (100/100 points for candidate A). Score Voting (aka Range Voting) does not limit the number of votes. That's WHY it's so resistant to tactical voting. If you give Ralph Nader a "perfect 10", you can still help Gore against Bush as strongly as you want. You could even strategize to give Gore a 10 and Bush a 0.
nitrogen 12y ago
Agreeing with aninhumer, I don't think that the majority criteria is a desirable property of a voting system if it comes at the expense of more desirable properties like monotonicity. Even if nonmonotonicity is not strategically exploitable, accidentally falling into a losing region by gaining votes would be incredibly aggravating. I would expect approval voting to reduce the polarization of politics and allow moderate candidates who are reasonably liked by everyone to enter political office. Rather than see our laws and policies flip-flop every 4-12 years due to one or the other party taking control, maybe we would actually have politicians focus on the mutual good of society rather than the parties' individual power.
- ClayShentrup 12y ago
  It's actually mathematically proven that a group can favor X even if a majority of its members prefer Y.
ClayShentrup 12y ago
Arrow's Theorem doesn't apply to rated voting systems such as Score Voting and Approval Voting. http://ScoreVoting.net/ArrowThm.html http://ScoreVoting.net/ArrowThm.html I've hung out with Arrow in his Palo Alto retirement community condo by the way.
ClayShentrup 12y ago
FairVote lacks expertise and is generally at odds with the election reform community. They literally will just like and make stuff up. http://scorevoting.net/RichieRV.html http://scorevoting.net/RichieRV.html https://sites.google.com/a/electology.org/www/fact-check https://sites.google.com/a/electology.org/www/fact-check That first link is written by Warren Smith, a Princeton math PhD who's the protagonist of William Poundstone's book Gaming the Vote, and arguably the world's foremost expert on voting systems.

by philipn 12y ago

aninhumer 12y ago

Approval/Range voting fail the majority criteria in that they allow a candidate who is the second preference of 90% of people to beat a candidate who is the first preference of 51% of people. Is that necessarily a bad thing? Your link also discusses the later-no-harm criteria for Approval/Range leading to strategic voting, but all the examples I've seen for this seem to talk about giving a marginally preferred candidate an advantage in the absence of any truly disliked candidates. I think in practice, people will be far more interested in keeping disliked candidates out than furthering their marginal preferences. Meanwhile, IRV suffers from Favourite Betrayal, which is a much more obvious incentive for strategic voting IMO. (Your link about NP-completeness of strategy only talks about STV.)

ClayShentrup 12y ago
This issue has been solved by math PhD's and political scientists who've studied the matter for decades. www.electology.org/approval-voting-vs-irv IRV can elect X even if Y was preferred to X by an arbitrarily huge majority and got more first-place votes than X. www.electology.org/core-support Regarding favorite betrayal, you might enjoy this video. https://www.youtube.com/watch?v=JtKAScORevQ https://www.youtube.com/watch?v=JtKAScORevQ
philipn 12y ago
In a range voting election, you'd have a lot of people who would bullet-vote for a particular candidate (100/100 points for candidate A). Then possibly lots of other people who would have to think about how people may be bullet-voting for those candidates, and then think about how they might distribute their points so as to ensure those candidates don't win. If I love candidate B and C, kind of like D, and hate candidate A, but all of candidate A's supporters are bullet-voting, how should I vote? It's ridiculous. It doesn't allow voters to express their preferences without strategic voting -- it's totally baked into the process. The range voting folks have come up with various metrics that say their system is the best, but completely ignore the most obvious reasons why its flawed. For the record, though, I'd vastly prefer any system over the current plurality elections (including range voting).
- aninhumer 12y ago
  >then think about how they might distribute their points Just to be clear, you're not imagining people have to share their vote between parties are you? Range voting means you can give every party 100 if you want to. >If I love candidate B and C, kind of like D, and hate candidate A, but all of candidate A's supporters are bullet-voting, how should I vote? If you really hate A so badly that you think any one of B, C or D would be vastly better, then go ahead and give them all 100. If A not getting in is that important to you then your "strategic" vote is in fact, an accurate representation of your preferences.
- ClayShentrup 12y ago
  In a range voting election, you'd have a lot of people who would bullet-vote for a particular candidate www.electology.org/bullet-voting > (100/100 points for candidate A). Score Voting (aka Range Voting) does not limit the number of votes. That's WHY it's so resistant to tactical voting. If you give Ralph Nader a "perfect 10", you can still help Gore against Bush as strongly as you want. You could even strategize to give Gore a 10 and Bush a 0.

nitrogen 12y ago

Agreeing with aninhumer, I don't think that the majority criteria is a desirable property of a voting system if it comes at the expense of more desirable properties like monotonicity. Even if nonmonotonicity is not strategically exploitable, accidentally falling into a losing region by gaining votes would be incredibly aggravating. I would expect approval voting to reduce the polarization of politics and allow moderate candidates who are reasonably liked by everyone to enter political office. Rather than see our laws and policies flip-flop every 4-12 years due to one or the other party taking control, maybe we would actually have politicians focus on the mutual good of society rather than the parties' individual power.

ClayShentrup 12y ago
It's actually mathematically proven that a group can favor X even if a majority of its members prefer Y.

ClayShentrup 12y ago

Arrow's Theorem doesn't apply to rated voting systems such as Score Voting and Approval Voting. http://ScoreVoting.net/ArrowThm.html http://ScoreVoting.net/ArrowThm.html I've hung out with Arrow in his Palo Alto retirement community condo by the way.

ClayShentrup 12y ago

FairVote lacks expertise and is generally at odds with the election reform community. They literally will just like and make stuff up. http://scorevoting.net/RichieRV.html http://scorevoting.net/RichieRV.html https://sites.google.com/a/electology.org/www/fact-check https://sites.google.com/a/electology.org/www/fact-check That first link is written by Warren Smith, a Princeton math PhD who's the protagonist of William Poundstone's book Gaming the Vote, and arguably the world's foremost expert on voting systems.