<br><br><div class="gmail_quote">2011/6/3 Kathy Dopp <span dir="ltr"><<a href="mailto:kathy.dopp@gmail.com">kathy.dopp@gmail.com</a>></span><br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
Forest,<br>
<br>
While I love the complement of "great idea", I still am not 100% sure<br>
if the method is well-defined. I.e. under what condition do all the<br>
Asset voters (the candidates) get to cast their 2nd choice votes for<br>
the voters? To be fair, wouldn't all candidates who received any<br>
bullet votes have to be allowed to cast 2nd choice votes. But then,<br>
wouldn't that be like voting against themselves? Or doesn't Asset<br>
voting have similar problems to IRV if only the losing candidates get<br>
to reapportion their votes -- i.e. tending to elect extremist<br>
candidates on the right or left and eliminating centrist majority<br>
favorites because the 2nd choices of some voters (that they've<br>
allocated to their 1st choice candidate) would be hidden during the<br>
counting process?<br></blockquote><div><br></div><div>It's important to be concrete when you think about this. Candidates would not be faced with an abstract choice of whether to support each other, in a vacuum. They would know the full first-round results. If you think about specific scenarios, you'll quickly see that a candidate will almost always know whether they can win - and shouldn't approve others - or whether they have no chance. That won't be simply a matter of how many votes they have, but also looking at the preference orders of the other candidates; so "center squeeze" is not a problem.</div>
<div><br></div><div>To put this in the abstract terms that a lot of people here like: insofar as people bullet voted, a Condorcet winner would be knowable, and then it would be a strong Nash equilibrium for that person to win. And insofar as people didn't bullet vote, then the approval winner would be clear, so candidate strategy wouldn't matter.</div>
<div><br></div><div>As for names, I'd prefer ODA (optional delegated approval) over AA, though I'll accept whatever the consensus is.</div><div><br></div><div>Jameson</div></div>