[EM] Randomized MCA, new weird voting method idea

[Abd ul-Rahman Lomax]

At 11:34 PM 2/15/2007, Stephane Rouillon wrote:
>Sorry to target an honest idea,
>but my main critic about random methods is the fact that
>results are not reproductible. Thus, the incentive for fraud
>is huge and the electors confidence within the system should drop
>with time...

It depends. What is interesting about the proposal is that the 
randomness isn't like, for example, Random Ballot. Rather, it moves 
between two alternative methods, each one of which alone is as good 
or better than existing standard methods. By doing this, it 
encourages honest voting. But it isn't going to reverse the decision 
of a majority, for example, as long as both methods obey the Majority 
Criterion.

Fraud becomes possible with this system if it is possible to 
manipulate the selection. But such fraud could be quite risky, and 
would not be as rewarding as ordinary election fraud. And it is 
possible to arrange the determination of a random number in a public 
way, such that it cannot be anticipated or manipulated except through 
a conspiracy of all parties involved.

For example, all candidates (and possibly some others) provide a 
sealed envelope with a random number in it, say a four-digit random 
number, chosen by whatever process the person chooses. The envelopes 
are collected at a single meeting, in public. The numbers are summed 
and the least-significant two digits are the number used to make the 
method selection.

In order to predict the result, you have to be able to control *all* 
contributions. If even one escapes you, you have nothing.

And if all the candidates wanted to control the outcome by agreement, 
well, more power to them!