[EM] Intro to list (etc)

[Rob Brown]

At 01:52 PM 10/28/2003, Rob LeGrand wrote:
>Rob Brown wrote:
> > Here is a UI I am working on for doing for ranking
> > candidates:    http://weblogz.com/voting/2000pres.html
>
>It's a good-looking, easy-to-use interface.  The only improvement I'd
>suggest would be to allow tied ranks anywhere in the ballot, so that a
>Browne>Keyes>Buchanan=McCain=Bush>Bradley=Gore>Nader vote would be allowed.
>  That way you could get rid of the notion of picking and unpicking since
>every option would be somewhere on the ballot.

I am not sure how to get rid of the notion of picking and unpicking while 
maintaining an intuitive interface.  Note that picking all candidates, in 
order from first choice to last choice, allows you to automatically set 
their order without having to keep pressing the move up and move down 
buttons.  I think having to use only the move up and move down buttons 
would be much more tedious, and more influenced by the ordering of 
candidates on the ballot.

You are correct that there is no way to set two candidates as equal (aside 
from all the non-picked candidates which of course are treated as being 
equally disliked).  I considered it, but the reasons I chose not to allow 
this option are:  1) it seemed to complicate things, and I want to keep the 
interface as simple as possible, and 2) the closer the election method 
comes to being strategy free, the less value it seems to have to the voter 
to set two candidates as equal.  If things really behave as we'd like them 
to, there should be no advantage to the voter of setting two candidates 
equal vs. just flipping a coin and putting one ahead of another.  So why 
give them an option that doesn't really benefit them?  I think it would 
only serve to confuse, and make them ask "why would I want to do 
that?"....which I doubt we could provide a simple answer to.

>I generally prefer Schulze's beatpath to Ranked Pairs, but the ranking
>returned by Ranked Pairs has an important advantage:  Every candidate beats
>pairwise the one just below him in the ranking.

Really?  That is not at all my understanding.  If it were true, things 
would be a lot easier for me.

<snip>
>My personal opinion is that directly using Approval Voting would be ideal
>for online polls when voters are allowed to change their vote in light of
>more recent results.  The running poll results would be very easy to
>display and understand and the voting interface could be extremely simple.

Well, I can see from your sig you are a fan of approval voting, and there 
is no question that I could make a much simpler interface for an approval 
voting ballot.  (thereby reducing the need for my ui/dhtml stuff....hmmm, 
maybe not a wise move on my part! :) ).  I prefer condorcet and ranking 
because to me it seems to have less strategy involved, and better allows 
people to fully state their true preferences, but I understand that these 
issues can be debated forever.

-rob