http://www.shroomery.org/forums/showflat...;o=&fpart=1
At the above post, there was alot of discussion about the unfortunate fact that we have a two party system here in the US. What wasn't talked about (much) was why we have only two viable parties.
The answer is simple: our system of one person, one vote inevitably leads to a two party system.
The analysis that follows applies only to winner take all elections in which there is only one elected representative, like the election for president - different analysis is necessary for multiple candidate elections
To see how this works, first we need a model of how people vote.
Let's say that for each candidate, each voter has an idea of how much he likes them.
For example, a voter could like Badnarik alot, be sort of neutral to Kerry, and hate Bush. Obviouslly, this voter could vote for Badnarik, who will lose, or vote for Kerry to keep Bush out of the White House.
Since few people will vote for Badnarik, whether they like him or not, the libertarian party can only become a threat for the white house if a major scandal or other disruption causes one of the other parties to become unviable.
The point is that this voting system makes a two party system very stable.
Consider the case where there are three equally viable parties, and that there is no correlation between who prefers one party over another. This year, and as long as all three parties are equal, there will be three people who could be elected president. However, as soon as one party starts to lag in the polls, people will start to desert that candidate for their second favorite, so that their most hated choice won't win - this has the effect of quickly eliminating the weakening party.
On the other hand, we see the effect of a two party system - it's very hard for one party to replace another, and even if it does there'll still be a two party system, just different parties.
Now, Alcalagon and the third party candidates favor Instant Runoff Voting (IRV). In this system, each voter ranks the candidates according to his preference. For example, the voter above could rank Badnarik 1, Kerry 2, and Bush 3 (ignoring other candidates).
The votes are counted as follows:
1) The candidate with the fewest -1- votes is eliminated
2) Voters that voted for the eliminated candidate have their -2- votes moved to -1-
3) If there is only one candidate remaining, he is the winner
4) Go back to 1)
In the example above, the voter could vote for Badnarik without hurting Kerry, and would.
However, this represents a special case where Badnarik CANNOT WIN the election. If the race was a genuine three-party race, the voter could hurt Kerry by voting for Badnarik.
Consider: some people vote -1-Badnarik -2-Kerry -3-Bush, some vote -1- Badnarik -2-Bush -3-Kerry, some vote Kerry, Bush, Badnarik; some vote Kerry, Badnarik, Bush.
In the first round, Kerry is eliminated, but not by alot. In the second round, Badnarik is eliminated, and Bush wins. It is possible that the voter above (who chose Badnarik first and Kerry second) could have kept Bush out of the White House by voting Kerry first.
When the 'third party' is small, IRV encourages people to vote for it; but, when it gets to be a threat to the other two candidates, it still has the potential to cause the party it is 'closest to' (the above example probably would have been more convincing with the Green party, where most Greens prefer the Dems to Reps but many Dems prefer the Reps to Greens) to lose to its 'most hated' opponent.
So, IRV enables a party to gain support, to a point, but it still strongly favors a two party system and makes a three-or-more party system highly unstable (remember, this only applies to those votes where there is a single candidate - the story is very different for multi-candidate elections like city council).
For a concrete example of this, look at Australia - they use single candidate IRV, and have two major parties - the conservative Liberal Party, and the liberal Labor Party.
Are there other choices?
Yes - two better choices are Approval voting and Condorcet voting.
In approval voting, each voter votes yes or no on each candidate, and the candidate with the most yeses wins.
Approval is better than IRV because it eliminates the bias toward a two party system. However, it is not perfect because it can result in the election of a second choice candidate. For example, say that most people prefer Badnarik to Kerry, but hate Bush, and a few people think Kerry's ok but don't care much for Badnarik or Bush. Kerry will get elected, but most people would have prefered Badnarik.
This problem results from the fact that people don't get to fully express their preferences under Approval voting - Which in my oppinion is still better than IRV where people fully express their oppinions but their oppinions aren't properly counted.
As far as the voter is concerned, Condorcet voting is identical to IRV (they number their preferences, 1, 2...). However, there is a difference in how the votes are counted. Each candidate is considered pairwise against each other candidate. For each voter, the higher ranked candidate gets a vote. The candidate that wins all the pairwise contests wins the election.
In the example above, where a voter in an IRV election enabled Bush to win by choosing Badnarik over Kerry, the voter could vote his preference without fear that he was enabling his most hated candidate to win. There are no conditions under which he would elect Bush by choosing Kerry over Badnarik or visa versa.
Condorcet suffers from the problem that it doesn't always determine a winner. Consider a vote in which there are three candidates (A, B, and C) and three voters, who vote as follows:
Voter 1 -1- A -2- B -3- C
Voter 2 -2- A -3- B -1- C
Voter 3 -3- A -1- B -2- C
The analysis:
A vs B: 2 votes A (1,2) 1 vote B (3) B loses
A vs C: 1 vote A (1) 2 votes C (2,3) A loses
B vs C: 2 votes B (1,3) 1 vote C (2) C loses
As you can see, no winner can be determined using this method, and some other method needs to be used. In real life, I expect that the situations where Condorcet fails to determine a winner will be fairly rare, because they will only occur when the vote is evenly split and the preference for one candidate over another has no correlation to other preferences (how many people actually voted Nader-Bush-Gore in 2000?).
So, if there's no clear winner, another method needs to be used to sort out the remaining candidates. But, we are in no worse situation, and generally better (candidates will have been eliminated), than we were before applying condorcet.
IRV is an alternative method of voting that's gaining some steam in the US. It is favored by both third-party candidates, and has some support among the less party-bound Republocrats.
It is better than our present method for enabling a more open voting and political system; however, it's still pretty shitty - one of the worse alternatives for a single candidate election, in fact.
If you want viable third parties, it is vital to change the voting system. Rather than moving to the mediocre IRV, which will perpetuate a two party system, though making it easier to change the two parties, let's switch to approval voting, or even better, Condorcet.
Edited by phi1618 (10/03/04 09:44 AM)
|
Threaded
Previous
Index
Next
phi1618 
1stimer 
Edit 
Reply 
thealkaloid
infidelGOD
wingnutx
SpecialEd
DigitalDuality
1 2 3 all )
Learyfan