7 ms·
This sort of scheme can cause problems with vote splitting and 'spoilers'; see https://en.wikipedia.org/wiki/Spoiler_effect#Bush.2C_Gore.2C_and_Nader_.282000_U.
by bshanks 13y ago
This sort of scheme can cause problems with vote splitting and 'spoilers'; see https://en.wikipedia.org/wiki/Spoiler_effect#Bush.2C_Gore.2C_and_Nader_.282000_U.S._presidential_election.29 https://en.wikipedia.org/wiki/Spoiler_effect#Bush.2C_Gore.2C... , and also
https://en.wikipedia.org/wiki/Independence_of_clones_criterion https://en.wikipedia.org/wiki/Independence_of_clones_criteri... .
For example, if 50% of voters are "for" gun control, and 50% are "against" gun control, but there are 2 very popular, well-written posts supporting gun control, and only one very popular, well-written post opposing it, then the gun control supporters will "split the vote" and their best comments will only be ranked about half as highly as the opposition.
Which may or may not matter depending on how people interpret comment scores.
One alternative that i like is reweighted score voting: http://rangevoting.org/RRV.html http://rangevoting.org/RRV.html
- gregw134 13y agoInteresting post on reweighted score voting. How would that work in practice for sorting comments? Would you have people rank the top-level comments in the order they agree with? The strategy I had in mind for comments was to create a column of arguments for and against, and to only allow users to vote on comments in the column where they've placed their opinion. That way the strongest arguments from both sides would be shown.