[EM] Unranked-IRV, Cumulative, and Normalized Ratings

Mon Apr 23 15:49:11 PDT 2001

I'll be more specific.  Suppose that there are three candidates A, B, and
C , of which your favorite is A, and that there are five voters. You ask
the other four ahead of time what their utilities for the three candidates
are.  They trust that Cranor's method is strategy proof, so they frankly
tell you exactly how they intend to vote on the CR ballots. 

Before you vote in the real election, you run her algorithm based on your
sincere utilities, as well as the information that you have received from
the other four voters and find that after several iterations of the
algorithm, the strategizers have converged to constant advice on whom each
of the voters should approve. In particular, the final advice you receive
is to bullet vote for A, even though you had a utility of 70% for B.

You decide to take a chance of fooling Cranor, and bullet vote A in the
actual election.

When the election results are in you find out that the other voters voted
exactly as they had foretold.

But you find out that this time Cranor is advising you to approve AB.

That would seem fishy to me.

In other, words if you already know the advice that Cranor is going to
give you, wouldn't you do just as well or better to use that advice on the
initial CR ballot? 

If giving more relative support to A on the initial ballot decreases A's
chances to the point that Cranor's advice changes from "approve only A" 
to "approve AB", then I would have to conclude that the method is
non-monotonic.

Forest

On Sat, 21 Apr 2001, Richard Moore wrote:

> <html><head></head><body>I understand the question and it does seem paradoxical. Since I don't know<br>
> the details about the method I can only speculate.<br>
> <br>
> Imagine that each voter is given a computer which is running a standard piece<br>
> of software that implements a standard strategy algorithm. The software<br>
> takes a voter's ratings input and runs it through the strategizer, which<br>
> calculates the voter's ideal approval ballot based on current statistical<br>
> information (initially a zero-info strategy). The result is forwarded to<br>
> a central computer which counts the votes and then predicts the likelihood<br>
> of various outcomes in the next round. This could be done by treating the<br>
> current round as a statistical sampling, for instance. The information is<br>
> sent back to the voters' computers, which then adjust their approval<br>
> ballots, and the process is repeated until a winner is converged upon.<br>
> <br>
> If I were to give my computer insincere ratings, I am in effect claiming<br>
> to be a better strategizer than the system is. If instead I give it sincere<br>
> ratings, I am trusting the system to be a better strategizer than I am.<br>
> Since the strategizing software is continuously making adaptations to<br>
> the political environment, as reported by the central computer, I suspect<br>
> that I would be wrong to think I could do a better job.<br>
> <br>
> Also, I don't think such a system would perfectly maximize utilities, but<br>
> would do as good a job as Approval would do if all voters voted optimum<br>
> Approval strategy. So the system is more of a "strategy assistant" to the<br>
> voters than it is a "result optimizer". Plus, the system could also collect<br>
> the presumably sincere ratings after the fact to determine how strong the<br>
> winner's mandate really is (in an SU sense, that is).<br>
> <br>
> Richard<br>
> <br>
> <br>
> Forest Simmons wrote:<br>
> <blockquote type="cite" cite="mid:Pine.HPP.3.95.1010419175822.9685B-100000 at orion.cc.pcc.edu"><pre wrap="">This is more of a query about Lori Cranor's method than anything else.<br><br>If it really gives no strategic incentive for distorting ratings, it<br>sounds like the ideal way to use CR ballots.<br><br>Here's what puzzles me. On the one hand, it seems like any method like Ms<br>Cranor's that uses CR ballots to formulate optimal Approval Strategies<br>should be able to do so in a way that would give the win to the candidate<br>with the greatest average rating.<br><br>If that is the case, then it seems like any strategy that would improve<br>the average rating of your favorite on the CR ballot would be tempting. In<br>other words, one would be tempted to distort ratings. <br><br>On the other hand, if the method doesn't give the win to a maximally rated<br>candidate, then it probably isn't much better than plain old Approval in<br>social utility.<br><br>Can you shed any lig!
ht!
>  on this?<br><br>Forest<br></pre>
>   </blockquote>
>   <br>
> </body></html>
> 
> 