Poll

I love approval and range voting, think they're generally the best option for choosing a single winner, and am delighted to see pipedot using one of them.

But , in a poll on "which features are the most important", I'm not sure approval's the right answer.

Of course, it depends on what exactly you plan to use the results for -- if the idea is to only implement features that a certain fraction of voters approve, approval's good. But if the idea is that you aim for feature-parity with /, so all these features are to be implemented eventually, but you're seeking to prioritize the order of feature implementation according to community feedback, I think using a rank ballot with Kemeny-Young counting would be better. To me, the poll title suggests the latter, but I could be misinterpreting it...

The thing about approval voting is that (provided the voter is smart about their own threshold of approval) it's good at reflecting the strength of preferences (even though it quantizes them to 1 bit!) -- if you can't stand any but your two favorites, you won't accidentally contribute to the election of the guy you put as number 3. But the flip side is, it loses the order of preference between any two candidates you do approve, and between any two candidates you don't approve. So in a case where all candidates are winning, we just want to sort them, the information of preference order is IMO more important than preference strength.

For example, suppose 60% of voters approve both items A and B (may also approve others, but at least those two), 30% approve neither -- with approval voting, you're letting the remaining 10% decide which order to tackle these two features in (by the difference of (A, !B) vs. (!A,B) votes). But using a rank ballot, you get the priorities of all voters. Of course the 10% with mixed votes will have a known preference, but the 60% block gets to specify which of these two beloved features they find most important, and the 30% block gets to specify which of these useless-to-them features they care least about... end result, Kemeny-Young spits out an order of tackling the features that makes the most people happiest.