<HTML><BODY>
<div> From: wds@math.temple.edu<br>

> 3. Random tie-breaking is essential so all <span class="correction" id="">candiate</span> <span class="correction" id="">winnign</span> chances are always<br>

> 100% independent of the candidate-ordering.<br>

                                                                                                                                                           
 <br>

<span class="correction" id="">Definately</span>.<br>

                                                                                                                                                           
 <br>

> 4. I find <span class="correction" id="">Venzke's</span> discovery with a 10-candidate set that "<span class="correction" id="">IRV</span> tends to favor<br>

> outsiders"<br>

> whereas "Approval(mean-based cutoff) tends to favor <span class="correction" id="">centrists</span>" very interesting.<br>

> But it needs more investigation with<br>

                                                                                                                                                           
 <br>

This <span class="correction" id="">sorta</span> makes sense as <span class="correction" id="">centre</span> squeeze is a known issue of <span class="correction" id="">IRV</span><br>

                                                                                                                                                           
 <br>

> 5. Approval(mean based cutoff)  looks pretty bad in these <span class="correction" id="">sims</span>, although so far<br>

> the<br>

> <span class="correction" id="">sims</span> have not employed correct random <span class="correction" id="">tiebreaking</span> so I don't know how much of<br>

> them to believe.   But anyway, it would be interesting when that issue is<br>

> repaired.   This seems to be the possible basis for a good attack against<br>

> Approval Voting.<br>

                                                                                                                                                           
 <br>

The new results (on Brian <span class="correction" id="">Olson's</span> site) do in fact have the random tie<br>

break rule.<br>

                                                                                                                                                           
 <br>

I think a zero info strategy is always going to have problems for approval.<br>

                                                                                                                                                           
 <br>

><br>

> 6. However, I have proved the following theorems in the large#voters limit:<br>

> (a) approval with randomized-oblivious thresholds chosen by voters yields<br>

> <span class="correction" id="">Voronoi</span> diagram.<br>

> (b) approval with the following kind of strategic voters, also yield <span class="correction" id="">Voronoi</span><br>

> diagrams:<br>

>        1. run approval election.  (Say X wins.)<br>

>        2. cast votes using <span class="correction" id="">X's</span> utility as cutoff where<br>

>               Y>X  ==>  approve Y.<br>

>               Y<X  ==>  disapprove Y.<br>

>              
Y=X  ==>  toss a fair coin to decide to approve or
disapprove Y.<br>

>        3. go back to (1) until stabilizes on a single winner who keeps winning.<br>

> which two theorems, I suppose, form some sort of defense for approval voting.<br>

                                                                                                                                                           
 <br>

This means that the winner of the previous election has a 50% chance of being<br>

approved by each voter, so only the <span class="correction" id="">condorcet</span> winner can be stable ?<br>

                                                                                                                                                           
 <br>

><br>

> 7. Why the heck are you simulators not trying RANGE VOTING?  (With voters<br>

> who "normalize" their range scores x  via   x -->  (<span class="correction" id="">x-worstScore</span>)/(<span class="correction" id="">bestScore-worstScore</span>)<br>

> so that the best candidate gets range vote 1, the worst 0, and the rest are<br>

> <span class="correction" id="">reals</span><br>

> somewhere in between?  [<span class="correction" id="">Bolson</span> actually had "range voting" = "social utility<br>

> winner"<br>

> computing twice the same thing with different names, which was both false and<br>

> silly.]<br>

                                                                                                                                                           
 <br>

Yeah, I said the same thing.<br>

                                                                                                                                                           
 <br>

The source is available to add new methods.<br>

<br>

</div>

<div> </div>

<div style="clear: both;"><span class="correction" id="">Raphfrk</span><br>
--------------------<br>
Interesting site<br>
"what if anyone could modify the laws"<br>
<br>
<span class="correction" id="">www</span>.<span class="correction" id="">wikocracy</span>.<span class="correction" id="">com</span></div>

<div> </div>
 <br>
<br>

<!-- end of AOLMsgPart_0_3fb1ec37-f49a-449a-a89b-15622f8a2195 -->


<div class="AOLPromoFooter">
<hr style="margin-top:10px;" />
<a href="http://pr.atwola.com/promoclk/100122638x1081283466x1074645346/aol?redir=http%3A%2F%2Fwww%2Eaim%2Ecom%2Ffun%2Fmail%2F" target="_blank"><b>Check Out the new free AIM(R) Mail</b></a> -- 2 GB of storage and industry-leading spam and email virus protection.<br />
</div>

</BODY></HTML>