Some people get entirely confused over preferences, to the degree they think people's second preferences are counting more than other people's first preferences. Please think about how this system really works.
AV isn't about totalling up everyone's opinion of preference cumulatively, it's about totalling up peoples absolute votes...their expressions of endorsement of a candidate.
When we go into each round of voting we are essentially re-running the election, asking everyone to keep their results the same where their candidate is still running, and for everyone else to re-allocate their vote to someone still in the running where their candidate has been eliminated. Their preference order doesn't really come in to it other than as a factor of allowing the system to run automatically round by round.
If the exact same system was used but non-instantaneously (the system,Run Off or Exhaustive voting, used by the Tories to elect their leader, and the House of Commons to elect their Speaker), where there is a gap and a re-vote between each round, no-one would call the vote cast a preference...it'd simply be a vote, counted equally with all the others cast alongside it.
The question asked, round after round (though instantaneously through the use of a preferential order), is "Which of these candidates is best for the job", where one candidate is removed each time to narrow the field. It's not "Please rate this candidate out of 10, we'll add all the scores up at the end".
You can *try* to weight those votes, the second preferences and beyond, but you'd be wrong, as this person has only one vote...an endorsement to carry someone to office...and where it ends up is a statement of only one thing...
"I want this candidate more than I want any of the others still standing".
So when, under AV, someone wins on 55% when they would have come "second" under FPTP with 30%...it's not that 25% of people voting for them care any less, all 55% will have said they want that person more than anyone else still in the running, beating the other 45% minority. Ask those people in the street and they may explain their choices, but will ultimately reinforce that if the options came down to it again, they would still vote, and endorse, that same candidate over the other available.
To claim that any of that re-allocated vote means less than someone else's is to call in to question the whole validity of FPTP as a system as well. After all, can we ever be sure that a winner on 40% has as much "belief" or "support" behind them as the second place person on 30%?
Or, really, do we understand that a voting system (at least a single member constituency based one) is a binary process...and that it's easy to get confused by the use of "preferences" meaning people care progressively less, when the reality is that the voting system cares not how much you support each preference...only that you do support that preference enough to put a mark down, or not.