tbf with an electorate of just 110 people casually acquainted with each other, it might not be too difficult for some to guess who cast the Fringe Party vote. But yes, you're not likely to deanonymise them all.

You can guess, but you can guess before you know how they voted - you can narrow your error bars, but it seems very unlikely you can out someone that you couldn't already out. Say you see all the votes, one is for an extreme party, there's going to be plenty of people whose politics you don't know in depth - people (IME) will either keep secret their allegiances or be very vocal. Those who are vocal won't care that you "know", those who're secretive ... did it help?

Knowing that somebody in your village definitely voted for the Fringe Party is a piece of information you wouldn't otherwise have. There's a big difference between having your suspicion about someone's voting intent supported by data and not having your suspicion supported by data.

This matters if said community is liable to alienate people based on their assumed view; the fact it isn't 100% accurate doesn't make things better...

I don't think there's a big difference between: 'there was one person out of 20,000 [smallest electoral ward in UK] who voted this way' and 'I think this person will vote this way'.

If you were going to beat them up you'd do it either way, I'd expect?

I'm not sure why you've changed the subject from "an electorate of just 110 people casually acquainted with each other" in the Australian local example I addressed in my OP to UK electoral wards which are 200-1000x the size, and obviously far beyond the scope of people being acquainted with everyone eligible to vote. I believe that's the reason the Electoral Commission doesn't collect results at polling station level and distribute that data despite the theoretical capability to do so.

And no, I don't think witch hunts are as likely to take place if people aren't informed of the presence of a witch.

Point of order: the electorate is 114,000 people who can vote at any one of the 86 booths. Or by mail. Or absentee somewhere else. The booth itself had 110 people vote there. I’ve since found another with only 98!

I personally think your fears are misplaced. If a small community is going to go after people for their votes they know each other well enough to go after each other with proxies for their vote instead.

This isn’t really an electoral problem at this point, it’s a regular small town rumour and innuendo problem.

Sure, if there are other voting options available to voters in that locality that don't involve using that station and the records of which station eligible voters voted at aren't widely accessible, that eliminates the ballot secrecy issue - thanks for the clarification.

The problem with people going after each other based on perceived ethnic/religious/education/wealth divides or things they've actually said may exist in politically fractious small towns regardless of electoral design, but very small voter pools means your identifiable fringe minority's chance of being on the end of a reprisal is linked to how/if you actually vote. And that really is a ballot secrecy issue.

Ok, suppose the ward is 3 people (you, Alice, Bob) you're still operating on a hunch. Or Alice is overtly fascist, so you assume she voted that way .. but it's her overt political stance that you're directly operating on. If you're going to coerce/vilify her do you care that you could be wrong? Mathematically it's different, socially/operatively I'm not convinced it would change your actions in a significant way.

A lot of people coercing and vilifying don't care that they might possibly be wrong, and a lot of retrospective vote analysis can be made based on demographics or things other than overt stances. Alice might stand out in sufficiently small groups simply by not being an overt liberal, whilst Bob might stand out in a much larger group not for his quiet avoidance of discussion about politics but because it's pretty damn unlikely that the white church attendees in the village voted for the Nation of Islam

It's pretty obvious that the degree of comfort someone in Bob's situation has in voting for something associated with his highly visible minority status is closely linked to the size of the electorate relative to the size of his minority. The secret of his radicalism would be entirely safe if votes were tabulated at state level, and certainly exposed if it's tablulated at polling station level and he's the only black guy there. And the world has no shortage are stark divides and identifiable groups who are tolerated a lot more when they're not perceived as politically active or at least not in that way. Some of the political causes that can be problematic for some voters when ballots aren't secret enough aren't even that ugly or that fringe...

The same goes for fairly quite and much less visible Alices. The three people who voted against continued British sovereignty over the Falklands Islands [turnout: 1518] successfully kept their secret from a population which had its fair share of outrage. Wouldn't have been very likely they would have succeeded, or necessarily felt safe to vote in the first place, in the hypothetical event it would be reported that all three came from the Port Howard [Population: 20] polling station.

The one person voting for the fringe candidate isn't the real worry. In a close election the 10 who you will shoot if they don't vote for you is the real worry. Those 10 can turn the election, the fringe voting guy doesn't, and you probably want him because it shows the election isn't rigged in some minds.