It seems pretty clear that there are lots of people out there who, even if they can grasp the mechanics of voting in AV, nonetheless have issues with how the counting system works.
AV Voting is pretty simple, of course: instead of the current British system where you put an X for the only person you want to represent you, and let's use the word 'want' loosely, leaving aside issues of second-guessing and tactical voting, in AV you put a '1' next to the candidate you'd most like to see represent you, and if you so wish, a '2' next to the one you'd next most like, and so on. It's the same process that says 'I'd like a ginger beer, please, but if that's not possible I'd like an orange juice.'
What of counting? People seem to think it's horribly complicated, even though the Irish and the Australians find it fairly straightforward. Well, let's take a simple real-world example of AV in action, and look at the 1990 Irish Presidential election*, as that's about as simple an instance of AV as you're ever likely to see. I was in school at the time, and we did a mock class election to learn how this worked. I've understood the system since.
There were three candidates. Fianna Fáil's Brian Lenihan -- the deputy prime minister -- and Fine Gael's Austin Currie both ran as centrist figures, put forward by the two traditional centrist parties. Mary Robinson, an independent, ran with the support of the Labour Party; anything but a bland centrist, Robinson was an eminent liberal, who had campaigned back in the seventies for the legal availability of artificial contraception and who worked as a legal advisor for the Campaign for Homosexual Law Reform. She was about as polarising a character as one could get in Irish politics back in 1990.**
When the election was held, Lenihan received the highest share of first preferences, with 44.1%, Robinson coming second with 38.8%, and Currie coming third with 17%.
Now, the problem with this is that all three candidates received a minority of votes. Not one of them had received a direct and explicit mandate from the majority of voters.
Currie, as the recipient of the fewest votes, was eliminated so that his votes could be redistributed. It turned out that while a few Currie voters had expressed a secondary preference for Lenihan, and about half that number had taken the view that it was Currie or nothing, the vast majority of people who had voted for Currie had expressed a secondary preference for Robinson.
The end result of this was that Robinson won in the decisive final count, with 51.93% of the poll, Lenihan coming second with 46.44%.
Like I've said, this is basically how I learned how transferable voting works. We held a mock election in English class, of all places, with us being asked who we wanted to win, and with the class roughly breaking down along what would be the national line: three minorities, with Lenihan supporters being the largest minority and Currie ones being the smallest, and then almost all of the Currie voters switching their allegiance to Robinson.
Before you ask, yes, I was a schoolboy would-be Currie voter. And I was one of the ones who switched to Robinson. Was she my first choice? No. Was I happy with her? Yes.
If you're good, I'll talk about the 1997 Presidential election soon. That's a little bit more complex, with more candidates, and probably a better guide for how AV would be likely to play in a typical British parliamentary constituency. But still, the principle's the same.
* Yes, Ireland. Because despite the lies that even the Guardian seems to be swallowing, AV isn't something that's only used in Australia, Fiji, and Papua-New Guinea.
** Because preferential voting systems don't inevitably lead to bland centrists getting elected, despite what you'll read in the Spectator, the Mail, the Telegraph, and on ConHome.