[EM] What is the goal of a "better" election method?

Steve Eppley SEppley at alumni.caltech.edu
Tue Jan 10 11:37:23 PST 2017

Good question, Sennet.  In my opinion the 
most important goal is to give candidates who 
want to win a strong incentive to take 
positions responsive to voters' preferences 
on many more issues, so that lobbyists for 
wealthy special interest minorities will get 
their way on far fewer issues.

With traditional primitive voting methods, 
many voters choose among the candidates based 
on one or a very small number of issues.  For 
example, a significant number of voters in 
the U.S. base their votes on candidates' 
positions on the abortion issue, and it 
doesn't matter much to them what the 
candidates will do on other issues. This 
leaves the candidates free to serve special 
interest minorities on most issues (for 
example bank regulations).  Candidates can 
win election by cobbling together an 
electoral majority coalition based on just a 
few issues. (The coalition can even consist 
of minorities on those few issues, because 
several minorities can add up to a majority.)

In theory, a voting method that elicits all 
pairwise majorities and pays attention to the 
majorities' sizes (affirming the larger 
majorities' pairwise preferences when 
majorities cycle) ought to meet the goal.  
Here's why: Suppose candidate Smith takes 
position X on some issue, and a majority of 
the voters prefer position Y over position 
X.  The risk for Smith is that some other 
candidate, say Jones, will enter the race, 
taking position Y and copying Smith's 
positions on all other issues, because in 
that event a majority would tend to prefer 
Jones over Smith... even if it's an issue 
such as bank regulations that most voters 
don't care about much (relative to other 
issues).  The more unpopular X is compared to 
Y, the larger the majority who should prefer 
Jones over Smith. (If the issue is 
one-dimensional and Y is the voters' median 
position, then the further X is from the 
median, the larger the majority who would 
tend to prefer Jones over Smith.)  Given a 
voting method that elicits all pairwise 
majorities and pays attention to their sizes, 
why would Smith risk taking an unpopular 
position, if Smith wants to win?

A flawed argument often made by proponents of 
proportional representation methods is that 
representativeness, by definition, requires 
proportionality.  But that's not the only 
possible definition, and proportionality is 
really only a means to an end. Suppose the 
end is to have a legislature that produces 
policies similar to the policies the voters 
themselves would produce in a 
well-functioning direct democracy. (By 
"well-functioning" I mean voters somehow 
aren't too busy to represent themselves, and 
they choose policies using a voting method at 
least as democratic as the Robert's Rules 
"pairwise single elimination" majoritarian 
method for voting on motions and amended 
motions.)  In other words, the closer a 
legislature's policies are to the policies 
the people themselves would choose, the more 
representative is the legislature.  With this 
definition of representativeness, a 
non-proportional legislature may be more 
representative than a proportional 
legislature.  A legislature elected using a 
voting method that gives candidates a strong 
incentive to take majority-preferred 
positions on many more issues would be very 

On 1/8/2017 9:49 PM, Sennet Williams wrote:
> I follow this list when I can, but it is 
> not really clear what the benefit would be 
> of a different election system, except 
> possibly "counting more votes" or 
> "representing more voters." It seems like a 
> waste of energy unless it will accomplish a 
> goal that the most eligiible voters will be 
> motivated by.
> -Personally, I want "better govt."
> -China is working to end "multi-party 
> politics" because it results in bad govt. 
> (I think that the Chinese govt. is 
> referring to parliament rather than the 
> two-party system)
> -The Berkeley Daily Cal editorialized for 
> IRV because "IRV lets you vote for who you 
> really want"  (something to that effect, 
> that was a long time ago and memory fades)
> The point is, comments on this list might 
> be more effective if you declare
> 1-what benefit to the organization that 
> your voting system will accomplish.
> Keep in mind that lots of people DO NOT 
> want to enfranchise more voters.  They 
> might not see the benefit of it.  More 
> people voting reduces the influence of the 
> previous voters.

More information about the Election-Methods mailing list