MIKE OSSIPOFF wrote:
> [...]
> The only reason why IRV doesn't have equal ranking problems is
> because it forbids equal ranking of ranked candidates, and forces
> people to order-reverse instead. If you mean that Approval
> requires equal-voting more often than IRV requirees order-reversal,
> my criteria are about absolute guarantees--what will never be
> necessary with a certain method? If you want to write a criterion
> that tries to somehow quantify how much more often Approval needs
> strategy than IRV does, you're welcome to do so. Write one about
> when IRV guarantees against any need for strategy. Under what
> conditions can you make that guarantee? If you're the one saying
> that IRV offers that guarantee under some reasonable conditions,
> then write one or more criteria about that. Instead of just hinting
> that Approval's strategy need is more frequent than that of IRV.
It doesn't matter if actual order-reversal under IRV is less frequent
than equal-voting under AV. When the possibility of needing either
exists in a method, or at least when the frequency is non-negligable,
then you always need to be looking for that situation. In essence, you
are always voting strategically, even if good strategy requires sincere
ranking the majority of the time.
The fact that equal-voting is less severe than reversal, even if you
don't consider it exactly half as severe, is one thing to consider.
There is also the question of how difficult it is to use the strategy.
Best strategy under Approval comes close to following "gut feel", while
other methods' strategies can be counter-intuitive.
Bart
