Election mechanics (not about the US)

A few days ago I was musing elsewhere about some online elections. Specifically, Stack Exchange has been running elections to replace all the moderators who have quit, and it's highlighting some weaknesses in their election scheme. Ranked voting is much better than "first past the post" but you still have to put the right checks in place.

If your election system uses ranked voting, think about how voters can reject candidates. The Hugo awards have "no award" as an automatic candidate in each category and you rank all candidates. My local SCA group lets you mark candidates as not acceptable and any who get 35% NA are removed, which gives the voters a veto when needed. Systems in which you pick N candidates lack this safety check.

"Cast N votes" doesn't let you distinguish between "this candidate is ok but not in my top N" and "I oppose this candidate". And even if you allow "not acceptable" marks on candidates (like my SCA group), you still need to allow ranking those candidates so voters can express "the clueless candidate before the evil one". If I recall correctly, my SCA group gets that part wrong; if you vote "not acceptable" you can't also rank the candidate, so you can't express degrees of unacceptability. If your goal is to deter NA votes that's a positive; if your goal is to elect people who are broadly acceptable then it's a negative.

Stack Exchange uses "cast three ranked votes" and now allows uncontested elections, so the only way for a community to reject a candidate is to round up more candidates. Because Stack Exchange royally screwed some things up with its communities, recently there have been newly-elected moderators who'd only been users for a few months. A candidate in one election is largely inactive (and said so).

The new and mostly-inactive users might be fine people, but in the past the bar was higher -- moderators were expected to have been regular, positive contributors for a while. Desperate times call for desperate measures, I guess; SE has lost a ton of mods in the last year for good reasons that still apply, but they don't want to admit there's anything wrong. So it's important to them to have bodies in seats.

Every voting system has flaws. When choosing, you need to decide which flaws are ok, which you actually prefer, and which must be prevented. Ranking all candidates, allowing an NA mark or "no award", and applying a threshold is more expressive than "rank N" but also carries more voter burden. Too complex? Depends on the characteristics of the electorate and the importance of the results, I guess.

Codidact isn't going to mandate a particular election scheme for its communities. Nothing is baked into the software, and on the network we host ourselves, our policy is that our communities can choose their moderators in any way they choose so long as the method produces unambiguous results that can be audited. (That's because any disputes are going to be escalated to us, so we'd better know how to fairly adjudicate them.) But even though our communities can choose how to choose, we should probably plan on offering some sort of facilitated options -- we can run election type X or Y for you, or y'all can do something else. Not every community wants to build its own system, after all; we shouldn't make them. I think we're a ways away from moderator elections yet (our communities are in start-up mode), so there's time to talk with our participants about what makes sense.