Wednesday, June 23, 2004

Rocky Mountain High(ly-Technical Analysis), Colorado 

Last time, on the Bonassus:
The Denver Post reports that
The wealthy president of a Brazilian university is bankrolling an initiative to end Colorado's winner-take-all presidential electoral system.

J. Jorge Klor de Alva is the major donor to The People's Choice for President - a nonprofit group seeking voters' permission to award Colorado's Electoral College votes proportionally as a percentage of the statewide popular vote.

For example, a candidate who wins 60 percent at the polls could snag five of the state's nine electoral votes, leaving the remaining four to a candidate who wins 40 percent on Election Day.

The group has begun to collect signatures; it needs 67,799 to get the measure on the ballot.

If approved Nov. 2, the constitutional amendment would affect this year's choice for president by immediately permitting the division of Colorado electoral votes.
I discussed the "why Colorado" question in the earlier post, and set forth an argument for why Colorado's citizens would probably benefit from adopting the proposal in the long run.

But the question remains, how much would Democrats and/or third parties benefit if the proposal passes? Your faithful correspondent has taken a look under the hood of the Colorado proposal, and can tell you that:

SHORT ANSWER: Third parties would have a strong incentive to focus on Colorado, since they would only need just over 11% of the vote to win an elector. And, as expected, Democrats are likely to do better than they would under the current system.

BUT (and it's a surpringly big but), due to some oddities in the way the proposal is crafted and due to the probable heavy focus by Greens on the state, Democrats won't do nearly as well as they should. I'm not sure if the proposal's drafters didn't do their homework, but they could have easily drafted a system which would have had greater expected benefits for Democratic presidential candidates in 2004 and into the future.

LONG, BORING EXPLANATION: The problem is rooted in the very obscure question of what mathematical algorithm is used to transform votes into electors. It would seem like a problem with an obvious answer: just divide the electors proportional to the vote. But this isn't as easy as it sounds. Take this hypothetical, not wholly implausible turnout:


GOP Votes978,51348.93%4.40/9
Dem Votes751,55737.58%3.38/9
Green Votes186,7979.34%0.84/9
Libertarian Votes83,1324.16%0.37/9

Is there an obvious, obviously fair way to divide Colorado's 9 electors when faced with these voting results? The question is left as an exercise for the reader, but let me assure you that 200 years of analysis says the answer is "no, dammit!" [By the way, I could generate a gajillion more examples of tough-to-distribute voting outcomes, if you think I cherry-picked this one (which I did).]

So what? So in our hypothetical case, the formula which would be put in place by the Colorado initiative would award 6 electors to the GOP, 3 to the Dems, and none to the other party. In other words, despite getting less than half the vote (and only about 11% more of the vote than the Democratic candidate), the Republican candidate would get 2/3 of the state's electors.

Weird, huh? And what's weirder is that it didn't have to be like this. Every country using proportional representation has to have some kind of rule for turning votes into seats, any of which would be perfectly usable here. And, as it turns out, the 4 most commonly used algorithms all award more electors to Democrats and fewer to Republicans in the hypothetical example above.

Furthermore, if you look at all possible vote distributions with two large parties and a few very small parties, and compare the various rules in use elsewhere with the proposed Colorado rule, a statistical analysis (available on request) shows that the Colorado rule heavily favors the larger of the two parties in almost every case, relative to the other rules.

In other words, if the drafters of the Colorado initiative were trying to help out Democrats (and we assume that Republicans are likely to outvote Democrats in the next few presidential elections) our drafters didn't do as good a job as they could have.

The only possible objection to using one of the commonly-used, comparatively-friendly-to-Democrats rules that I can think of is that they're complicated. But actually, two of the rules are significantly LESS complicated than the proposed Colorado rule.

I don't know what happened here. I'll tell you what, though: I wish that rich Brazilian guy had called me before things got to this point.

This page is powered by Blogger. Isn't yours?

Weblog Commenting and Trackback by HaloScan.com Referrers: