Julian Sanchez on 31 Oct 2000 03:33:10 -0000

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

[Nettime-bold] Preference Voting & US Elections

> whether that will just ensure a conservative victory. I don't profess to
> understand the US voting system, but don't you have preferences? Isn't it
> possible to vote 1 Nader, 2 Gore, so that in the instance that Nader isn't
> successful his preferences will still flow to Gore rather than Bush? I'd
> really love someone to explain the system to me, because the debate so far
> seems excessively divisive.

No, the U.S. doesn't use preference voting.  Voters in each state select one
and only one candidate for the office in question to vote "yes" on.  Each
state wields a certain number of electoral votes (in rough proportion to the
population of the state), which (depending on the particular state) are
either all given to the candidate with a plurality in that state, or are
distributed proportionally among the candidates.  The candidate with the
most electoral votes- who is usually but not necessarily the one who
garnered the most popular votes- wins. Since second-preferences aren't
expressed (i.e. the system fails to reflect the fact that most Nader voters
ALSO prefer Gore to Bush) we get the third party spoiler effect that people
here are worried about.  Of course, since the 2 major parties are
unambiguously the beneficiaries of this effect, we can't hold out much hope
of a change in voting systems anytime soon.
   But FWIW, preference voting is not without its problems (as Arrow's
Theorem should lead us to expect).  It displays a property that the
political theorists call "population nonmonotonicity"- which in English
means that a candidate can win under one distribution of rankings, and then
lose under another distribution, where the *only difference* between
distributions is that some voters ranked that candidate *higher*.  In other
words, voting for a candidate can actually HARM that candidate.  Americans
are worried that voting for Nader could hurt Gore- but under a preference
system, they'd have to worry that voting for Nader could hurt NADER! Since
this is a bit counterintuitive, take a look at a simplified example:

8 people prefer A (and are indifferent between or abstain from ranking the
3 people prefer B to  A to  C
2 people prefer B to  C to  A
4 people prefer C to  B to  A

Under preference voting, C is knocked out at the first round. B acquires C's
votes, and so defeats A by a margin of 9 to 8.  Now imagine 2 of the
A-voters decide instead to vote the ranking C-A-B instead:

6 people vote A
3 people vote B then  A then  C
2 people vote B then  C then  A
4 people vote C then  B then  A
2 people vote C then  A then  B

This time, B is eliminated in the first round, and A beats C second round.
So A should be glad that his 2 supporters switched over-- even though they
decided to rank him last!

Here's another example, showing a different problem (though obviously one
shared by the American first-past-the-post system):
3 people prefer A to B to C
1 person prefers A to C to B
1 person prefers B to A to C
1 person prefers B to C to A
3 people prefer C to B to A

First round votes are A:4 - B:2 - C:3, so under Australian style preference
voting, B is eliminated and A wins 5-to-4. But in pairwise contests, B would
defeat each other candidate 5-to-4.


Nettime-bold mailing list