[EM] Majority Judgment avoids Arrow's Theorem (paradox)

[Kristofer Munsterhjelm]

On 06/06/2017 01:22 AM, fdpk69p6uq at snkmail.com wrote:
> Arrow's theorem only applies to ranked systems, while MJ is a rated
> system (as are Score/Range, SRV/STAR, Approval, etc.)  Later in life,
> Arrow supported rated systems:
> https://electology.org/podcasts/2012-10-06_kenneth_arrow
> <https://electology.org/podcasts/2012-10-06_kenneth_arrow>
>
> Gibbard's theorem is supposed to apply to all conceivable voting
> systems, though.

Just to be precise, to all conceivable _deterministic_ voting systems. 
Any random method of the type:

- With probability x%, choose the Random Ballot winner,
- With probability (100-x)%, choose the Random Pair winner

is strategyproof. It's also not very good.

To be even more precise, Random Winner is also strategyproof as far as 
strategy from the voters is concerned, because the ballots are never 
looked at. It has an obvious teaming strategy, however. And dictatorial 
systems are strategyproof, but lacking for other reasons.