The rational way to vote in the CA recall election

There are two parts in the recall election; both, fortunately, have a rational, utility-maximizing way to do it.

First, you need to create utility values for all candidates as well as Davis. The higher the utility value, the more you'll benefit from having the person as governor of California (mind you, this is based on the assumption that you're voting to win and not to make a statement). For example, my utilities at this point are Simon = Issa = 0, Ah-nold = 3, Davis = 5, Arianna = 8, and Camejo = 10.

Now, check the last replacement poll before the election itself, or rather, a combination of the last polls of various papers and stations. Using this poll, construct two related arrays of probabilities:

1) The probability each candidate in the replacement vote has of winning.

2) Given that there is a first-place tie between two candidates, the probability it is between X and Y for every pair of X, Y (i.e. what the probability for Issa and Simon is, what the probability for Camejo and Ah-nold is, and so on). The sum of all such probabilities should be 1, and if P_ij is the probability that the tie will be between i and j, then given any i, j, k, and l, P_ij/P_ik should be equal to P_lj/P_lk. You can approximate these probabilities by asserting that P_ij = K*P(i)*P(j), whereas P(i) is the probability that i will win and K is a constant chosen so that the total sum of probabilities will be 1.

On the recall question, vote according to (1): Multiply each candidate's probability with his/her utility, and add the results. If the total is higer than Davis' utility, vote yes on the replacement; if it is lower, vote no.

On the replacement question, vote according to (2): Multiply each probability by the difference between the utilities of the candidates involved. Then, give the candidate who has the higher utility a positive score equal to this value and give the one who has the lower utility a negative score equal to this value. Repeat for all such probabilities, always adding the result to the score of the better candidate and subtracting it from the score of the worse candidate. Vote for the candidate with the highest score here.


Two notes:

a) P(i) is not necessarily equal to the percent of the people polled who support i. If there are 5 candidates whose percentages are from highest to lowest 35, 20, 15, 15, and 15, then the leading candidate's probability is not 0.35 but about 1. The best way to find P(i) is to use standard errors and normal distribution probabilities, although I don't know enough statistics to give you a complete function that creates P(i) accurately. The best way is to ignore everyone who's more than two standard errors behind the leading candidate and then calculate for those who are close enough to the leader.

b) Make a mental note about utilities right now. If you wait until the last day, you'll subconsciously pick values that justify your preconceptions.

But I love it. :bounce:

...the post explains how to vote like a rational human being and not like a donkey and sheep hybrid spouting "Better dead than red! Dmes or death!"

i practically failed the two courses I had last year so I'm not getting it.

...and I'm majoring in it...

But then again I'm majoring in philosphy and music, I never had much of a head for figures.