[Start] Voting systems in the public sphere

[Ed Pastore]

On Jan 5, 2009, at 9:49 PM, Michael Allan wrote:

> What if sock puppets were to cross-rate each other, and so generate
> their own reputations?  Then you might have an army of them, run by
> bots, all controlled by a single person or organization?

Not sure why I failed to see that obvious criticism. Nevermind  
then. :) As for reputations, see below.

>    Both DemocracyLab and Metascore rely on reputation systems.  The
>    details are undocumented, but it would appear they are liable to
>    systematic discrimination.

We have not settled on whether or not we will use a reputation system.  
Our plan is to enable a basic one, but not to enforce it. We'll then  
allow small communities to play with the system and see what they do  
with the reputation aspect.

This is our approach to many such problems, as we figure real-world  
use is going to come up with better solutions than a small group  
hypothesizing.

>    Metascore has a synthetic consensus facility in which "a
>    synthesizing resolution effectively 'steals' points from the
>    resolutions that it synthesizes".  The intent is to encourage
>    consensus.  But the component's formal loss of points has no basis
>    in an actual loss of support; its supporters may continue to think
>    that the component is the better proposal.  In that case, actual
>    support for the component is suppressed, and a false support is
>    generated for the synthesis.  In this way, an artificial consensus
>    may eventually be forced.

People will be free to change their ratings at any time. If someone  
thinks a new proposal does not synthesize their support of the old  
one, then they can give a negative score to the synthesizing proposal;  
or alternately (depending on how we implement it), they can give a  
synthesizing score to the old proposal against the new one.

But more importantly, synthesis scores are meant to bring synthesizing  
proposals to peoples' attention, not to force them into acceptance.  
Synthesis scores should help good proposals rise to prominence, but  
they still have to garner a broad consensus to become policy. Simply  
getting a high synthesis score won't be enough.

>> ... We prefer that a proposal die (or be synthesized) if it cannot
>> garner consensus. This allows us to have no limit on the number of
>> proposals, with the expectation that most proposals will fail.
>
> So a proposal that fails to garner a consensus ends up dead?  Let's
> say a sub-group of 4,000 people supports a proposal - and would agree
> to it if they had the chance - but the rest of the 40,000 participants
> do not agree.  Then the proposal ends up dead (in some sense).
>
> If the proposal dies, what happens to its formal support?  Suppose it
> still has the assent of 4,000 people.  Is that information lost to the
> system?  Is that what "death" means for proposals?

It fails to pass in the larger community, for the time being. However,  
by their shared support of a proposal, those 4,000 people have defined  
themselves as a potentially distinct sub-community. They can make the  
proposal a policy of their smaller community and/or they can work on  
new methods of convincing the larger community to pass the resolution.  
The way to do the latter would most likely be to synthesize the  
original into something that has larger support. Thus we expect  
proposals to migrate up and down communities of different sizes until  
they find their natural fit.