Here’s a hypothetical to illustrate my qualms with IRV in the context of the delegate election (it is contrived but all such hypotheticals are):
3 candidates (A, B, C), about half the first round electorate ranks A > B >>> C, remainder ranks C > B >>> A. (That is, half are fine with A or B, prefer A slightly, but would really hate if C won; half are fine with C or B, prefer C slightly, but would really hate if A won).
If they vote by revealing their true preferences under IRV, B will be eliminated first round, and we proceed to the 2nd round with the divisive candidates A and C. On the other hand, under approval voting we probably end up with the more ideal outcome of B getting through to the next round.
My point is that:
- One’s voting strategy is very dependent on beliefs about how others vote - even IRV is not immune to this.
- While you are right that approval also is not immune to “over-strategizing”, I think approval would lead to less divisive, more broadly acceptable candidates being put forward to the wider electorate in the 2nd round, which I hope you would agree is an ideal quality for a delegate in our system of government.