I'd like to run the next monthly competition votings with multi-poll votes, and asking your opinion about this issue. At first, I feared that multi-poll votes would cause a mess, because I can't enforce voters to use only x amount of votes (say three). But as it turned out, it is not even necessary. Read below why.
I had a chat with Simetrical, our Chief Technician (please rep him for his help), and with his permission I post our PM-ing. (You can read the original Q&S thread here, started by Juvenal in the Q&S forum, but I've pretty much covered all issues.)
If you're still not convinced about multi-poll (or approval) voting, consider the following advantages:PW: How can I enforce members to vote for only 3 options (check 3 boxes) in a multi-poll vote?
Simetrical: You can't, it's not enforceable technically. But it's not necessary either, the system levels out itself.
For example, in this poll there are 66 voters, and 173 votes. 66 into 173 is 2.62. If it's using some functionality of the boards that I don't know about and that doesn't show up anywhere as an option when you try to add a poll, evidently that functionality is rather buggy. The CdC polls tell you to vote for the appropriate number of candidates. You can vote for any number you like. I've occasionally even done so, I think, although not recently.
PW: I've though out another solution: I tell the voters to cast 3 votes, and if anyone votes for more, he would get a warning and a temporary ban from the sub-forum. But I'm afraid we could only find the "cheaters" in a public poll. So my question: does it have to be a public poll to see who cheated on a multi-vote poll (who cast more or less votes than allowed?)
Simetrical: If the poll is public, you can tell. If it's not public, you can't tell, unless you're an admin. Personally I would suggest that you just allow everyone to cast any number of votes and ignore the issue. It's not like that will make the poll less fair.
PW: Why are you sure about this?
Simetrical: It really doesn't affect the fairness of the result. If someone votes for more people they improve each one's chance by less, and if they vote for fewer people they improve each one's chance by more. It balances out. Allowing people to vote for however many they want is called approval voting and is the most reasonable way to proceed here. The software doesn't support anything else, and I think you'll find that a lot of voters will accidentally click the wrong number of checkboxes if they aren't prevented from doing so.
PW: After reading thru what you wrote and the wiki article, I think it actually it makes sense. :hmmm:
So I will tell them, they can vote for any number of AARs they liked. The result will show the overall liking of the best AARs anyway.
Example:
Johnny votes for 3 AARs (A, B, C)
Mary votes for 4 AARs (A, B, C, D, E)
Tommy votes for 5 AARs (A, B, C, D, E, F)
Lisa votes for 6 AARs (A, B, C, D, E, F, G)
Result: A, B, C wins.
It doesn't matter that Lisa voted 2x times more than Johnny, because they both liked the "A", "B" and "C" AARs, so these AARs deserved to win. But if Lisa voted for only A or G, it wouldn't matter either... the same AARs would be prevail as the mostly liked ones. I'm not a mathematician, but am I right in this example?
Simetrical: Yes, that's pretty much the logic. You don't get more power with more votes, since each option can only get one vote. You're just dividing them into two groups: those you like, and those you don't. You vote for the ones you like (which increases their chance of winning) and don't vote for the ones you don't like (so you don't increase their chance of winning, and decrease it by comparison).
PW: Many thanks again, Simetrical!![]()
- less AARs would get 0 votes - more encouraging for AARtists
- you can broaden your spectrum of vote - less voting dilemmas (AKA analysis paralysis) and more freedom in your choices
- less chances for tie breaker situations
- a more balanced and refined survey.
If you've read this far, it actually means you're a nice guy/girl, so go and vote now. After that leave your comments, will ya?















