[EM] Divided Majorities

[Kristofer Munsterhjelm]

On 05/13/2014 06:52 AM, DNOW1 at aol.com wrote:
> Single winner case - Divided Majority
>
> 26 ABZ
> 25 BAZ
> 49 Z??
> 100
>
> Note that Z has 51 votes in third (last) place.

In other words, you'd like a method that passes mutual majority.

> Thus - remove losers -- right to left until there is a majority winner.

That sounds like IRV. As a patch to Plurality's "get confused by very 
small minorities", IRV works. But the problem with fixing methods by 
patching a defect is that it may very well lead to another problem being 
exposed a little further ahead. (See Burlington.)

> Note in real elections both the majority and minority groups might have
> multiple choices -- with some confused overlapping with the *middle* voters.
>
> Some of the Z voters might obviously prefer the lesser of the A or B
> *evils*.
>
> ANY election reform has to be really simple to get past the low low low
> math brain cells of SCOTUS and the media.

Alright, you'd like low math? How about this: Each voter gives each 
candidate a score. A candidate's majority score is the highest score a 
majority agrees that he deserves (by rating him that high or higher). 
The candidate with the greatest majority score wins.

The logic? A majority agrees he deserves at least a 5 (or B, or whatever 
scale you use), so he can claim to deserve at least a 5. But he can't 
claim to deserve a 6 because no majority backs him on that claim.

Voters for Z can express their lesser evil because their contribution to 
A and B's scores don't affect their contribution to Z's score. 
Furthermore, this method can handle Burlington scenarios without blowing 
up (unless very many voters indeed strategize).

If you think the media has even fewer brain cells, just use Approval. 
Each voter approves ("likes"?) as many candidates as he'd want. Greatest 
number of approvals win. The voters can no longer decide upon a lesser 
evil, that's true. And like IRV, the method can get confused in 
Burlington situations, but unlike IRV, the voters *can* fix it by 
foregoing their lesser evil.

Approval is not my favorite, but its simplicity is hard to beat. Just 
count every vote, even the overvotes. No need for elimination or for 
sorting ballots in different piles as you cross out eliminated 
candidates. No need to send messages back and forth between the 
districts and a central tallying location to decide who is to be 
eliminated next.

> For legislative body elections a choice would be elected when he/she got
> a Droop Quota --
> Each winner would have a voting power equal to the final votes he/she
> gets (direct and moved votes).

If you call Meek "low math", then you have a very odd definition of "low 
math" :-) But I get what you mean.