[EM] Why We Shouldn't Count Votes with Machines

[Jonathan Lundell]

On Oct 5, 2008, at 5:38 PM, Dave Ketchum wrote:

>     While many methods, including Plurality, have no trouble  
> correctly picking the winner when there are only two candidates,  
> Plurality restricts voters unacceptably when there are more than two  
> candidates and many voters want to show more than one as better than  
> the remainder - which happens often.

The issue is not the number of candidates, but rather the number of  
seats to be filled. Yes, it would be fine to have a better method than  
plurality to fill the very few necessarily single executive seats that  
we vote for, but that's a minor matter compared to the different  
between single-member districts and multi-member districts with PR.

Suppose we could contravene the laws of mathematics and invent a  
single-seat method that was Condorcet-compliant and satisfied LNH/H in  
the bargain. The degree of representation achieved by such a method is  
dramatically worse than any decent PR system.


BTW, it seems to me that there's a relatively straightforward solution  
in principle to the problem of computerized vote counting, based on  
the use of separate data-entry and counting processes. Let voters vote  
on paper, either by hand or with an electronic marking machine, enter  
the ballot data, perhaps by scanning, in such a way that the resulting  
ballot data can be verified by hand against the paper ballots, and  
permit counting by multiple independent counting programs.

There are nontrivial details to be resolved, in particular ballot  
secrecy and the resolution of conflicting results, but it seems to me  
that it's a fairly contained set of problems.