[EM] Arrow's axioms & an alternative to elections

[Philippe Errembault]

>>> The biggest problem I see is, who gets to define the rules for what
>>> gets decided at which level?  If the authority for that is too
>>> dispersed, you get a logjam.  If too centralized, you risk devaluing
>>> certain levels which would seemingly defeat the whole purpose of the
>>> arrangement.

>> Yes, ok. that can be a problem. I suggest the following strategy :
>> - Decision are taken a the top level,
>> - if citizen disagree with a decion, they can change their 
>> representaion at any time
>> - If representatives think they are not sure about a decision, then 
>> they report the qusetion to the n-1 level.

> Hmm, that sounds like a recipe for chaotic behavior - if there's too 
> many links, potentially one small change can disrupt things at any 
> time.  Of course, it might be a fun and worthwhile computer simulation 
> to see what constraints would be necessary to guarantee some measure of 
> stability.

Yes, this is a risk, but

1/ it can be attacked by tuning the momentum in the system.
i.e. you can add progagation times or some kind of low pass filtering between the tiers.

2/ with the step 3, the impact is limitated to specific topics, which drastically
reduces the potential catastrophic impact of such a behaviour. (especially as there 
is no more concentration of power)

3/ Chaotic behaviour is the only way to have "some kind of intelligence" in the system
Intelligence is inherently chaotic.

The system being chaotic, simulating it with a computer would be quite difficult. I don't know where
we would find good simulation for human beings and there is quite a risk to miss important behaviours
of the system. Instead, I would suggest to implement it in parallel with existing institutions, to keep a 
control, make tuning and see what happens.

Philippe Errembault