[EM] Re: wiki

[Abd ulRahman Lomax]

At 02:01 AM 9/1/2005, Jobst Heitzig wrote:
>Dear Abd ulRahman!
>
>You wrote:
> > I'll disagree that "only randomized methods can do so," since there are
> > other alternatives that are neither deterministic or randomized,
> > beginning with the simple one of holding some kind of runoff.
>
>Assume there is no sincere CW. Then each candidate is defeated by some
>other. You suggest runoffs which count as deterministic in the sense
>that the *whole* procedure will produce a single winner which depends
>only on the ballots cast.

This is a stretch of the term "deterministic." If, for example, there 
is a tie in an election (according to the rules, i.e., no definite 
winner), and the resolution is to undergo some process which does not 
involve ballots (there would be many such methods), but instead 
involves either chance or the exercise of judgement, it is not common 
usage to call this "deterministic." I'll point out that resolving the 
election by random choice would be, by the apparent definition you 
use, "deterministic." It produces a single winner.

Deliberative methods may involve at some point the casting of 
ballots, but they are not, as a whole, deterministic. Neither with 
random choice nor with a runoff or asset-type procedure is the 
outcome predictable from the ballots cast *in the original election."

>  I claimed there cannot be a group strategy
>equilibrium in that case, that is, a way of voting such that after the
>election there is no group of voters who can say 'If only we had voted
>like so-and-so, we had the result so-and-so which we all prefer to the
>result at hand'.

Neither does random choice procedure produce this result. I have not 
examined DFC in intimate detail, but based on what I have seen, 
random choice is only used if the election result is not clear based 
on the primary criteria. If the election proceeds to the point of 
random choice, one of the possible outcomes will very likely involve 
a set of voters who could have altered the outcome in the direction 
they favor, had they voted differently, because had they voted 
differently, the election would not have proceeded to the random choice phase.

I agree that random choice has an element of appeal, and *especially* 
when used for representative elections, because it would give a 
possibility of representation to parties which might otherwise not 
quite make it. Representative elections almost inherently fail to 
represent fully the public. Some PR methods can approach success in 
this, however. Delegable Proxy, practically by definition, produces 
full representation. (Presumably in an assembly with a large 
jurisdiction, there would be full participation limits -- but not 
voting limits in DP --; however, voters may either find someone who 
has already qualified, thus optimizing their representation, and give 
their proxies to that person, or such voters may combine with other 
such voters, if they can find enough of them. With a wide public FA, 
it would be quite easy to find them. This is one reason why Free 
Associations and Delegable Proxy are linked in my concept, though 
they really are severable. (The largest known FA is not DP, and the 
only known political DP application is not an FA.)

>  It's quite easy to see that such an equilibrium does
>not exist: Assume the whole election process elected X, and that X is
>defeated by Y. Then the majority which prefers Y to X can say 'If only
>we had voted Y>rest, we had the result Y which we all prefer to the
>result at hand'. QED.

I may be arguing over a non-issue. I reread the DFC page and found 
the method sufficiently unintelligible there that I could not say 
that I understand it well enough.

> > Perhaps someone could take Mr. Heitzig's mail, with its concise
> > definition, and use it to clean up the wiki page. I'd do it if I had time.
>
>Please not "clean up" the page "imagine democratic fair choice" but add
>a new page "democratic fair choice". That was the whole point of my
>naming of the page. The idea is to have different descriptions for
>different tastes.

Some wikis, depending on admin setup, allow creators of a page to 
lock it. Presumably the attached discussion page would not be locked....

I read the Imagine Democratic Fair Choice page months ago and had the 
same reaction. That is, I'm exactly sure, from it, what the method 
is, and I see aspects that I don't like. It is entirely possible that 
I have misunderstood the method and that what I don't like is only my 
own imagination of the method, not the method itself. This is why a 
good simple description of the method, which would include the ballot 
form (as is on the Imagine DFC page), plus, without extra verbiage, 
the procedure for determining the winner.

By the way, the use of a ranking submitted by candidates bears a 
resemblance to one possible implementation of Asset Voting, in which 
the revotes of candidates would be expressed in advance. Or of some 
candidates, not necessarily all. This, again, resembles delegable 
proxy for a newcomer who might consider the set of available proxies 
(in DP implementations that I support, the existing proxy assignments 
of those available to serve as proxies would be public record); by 
selecting the proxy, the voter has selected the whole chain of 
proxies, which then could determine, in advance, how votes would be reassigned.

In this system used as an election device (which is not its primary 
intention), there would be private, secret proxies, assigned by 
secret ballot, to a set of public proxies, people who have declared 
their willingness, essentially, to vote in public.... Analysis of the 
published proxy assignments would reveal loops, which could then be 
broken by any member of the loop changing his or her assignment. With 
loops broken (or, more accurately, becoming large enough that loops 
do not exclude anyone from representation, since the whole loop 
membership is larger than the assembly quota), there is then a 
specific procedure available for resolving any election difficulty, 
based on proxy assignments in effect as of a certain deadline.