[EM] Election thinking,

[Dave Ketchum]

Seemed to me Mike left out some important thoughts - can we do better?

On Apr 21, 2012, at 3:41 PM, Michael Ossipoff wrote, as:
        Article, with the added paragraph and some better wording
>
> Adrian and EM:
>
Elections are important to many organizations - and important that  
they help the voters express their desires effectively.  Important  
enough that voters should see to it, whatever it takes, that they get  
the information they need and that their thoughts find their way  
correctly to whoever is responsible for responding.

This article's topic is election methods.

Normally candidates get nominated, and can campaign as needed.  Even  
with these, write-in voting should almost always be permitted - there  
is almost always the possibility of a nominated candidate becoming  
unsuitable too late for formal replacement.
>
> Our current voting system, of course, is the vote-for-1 method.  
> Also  called
> "Plurality", or the "single mark method".
>
> In our Plurality elections, we often hear people saying that they're  
> going
> to vote for someone they don't really like, because he/she is the
> "lesser-of-2-evils". Note that they're voting for someone they don't  
> like,
> and not voting for the people they really do like, because the  
> people they like are
> perceived as unwinnable.

A related possibility is voting for the unwinnable candidate and  
letting the "worst-of-2-evils" win.

A possibility that helps, sometimes, is to be permitted to Approve as  
many candidates as the voter likes best - protecting against the  
"worst-of-2-evils" winning.

This Approval method is a trivial expense and trivial improvement over  
Plurality voting.

>  The candidate with the most
> "Approved" ratings wins. The result? Well, we'd be electing the most
> approved candidate, wouldn't we.  Who can criticize that?

The voter who did not have equal liking for all Approved.  There are  
many voting methods to choose from, so we will only mention a few here:

.     Condorcet - really a family of methods - variations on a design  
using ranking.  One can use a single rank value for one candidate  
(same value as Plurality), or several (same value as Approval).  A  
voter can also use different ranks, using higher ranks for those most  
preferred, and leaving unranked those least-liked.

Here each pair of candidates is in  a two-party race counting how many  
voters rank one, or rank one higher than the other.  The candidate  
winning all of its races wins but, if none, the one coming closest wins.

.     IRV - a Condorcet method, though a voter can use each rank  
number only once and the counting is different.

Considering only each voter's top rank, see if there is a winner.  If  
not, discard the top rank for the least-liked candidate and move each  
such ballot to next candidate.

The discarding sounds good, and usually discards truly least-liked.  
Trouble is. the truly best-liked may have been hidden behind lesser- 
liked by enough voters to have been discarded as least-liked.

..     Score - voters rate each candidate and ratings are added to  
determine winner.  Tricky because making a rating higher or lower can  
affect who wins.

DWK