Weight a vote for an object based on a user's previous voting history

vote_weight.py

from math import sqrt

def confidence_bounds(ups, downs):
    #stolen from reddit!
    n = ups + downs
    if n == 0:
        return 0, 0
    z = 1.0 #1.0 = 85%, 1.6 = 95%
    phat = float(ups) / n
    confidence_positive = sqrt(phat+z*z/(2*n)-z*((phat*(1-phat)+z*z/(4*n))/n))/(1+z*z/n)
    phat = float(downs) / n
    confidence_negative = sqrt(phat+z*z/(2*n)-z*((phat*(1-phat)+z*z/(4*n))/n))/(1+z*z/n)
    return confidence_positive, confidence_negative

UP_VOTE, DOWN_VOTE = 0, 1

def score_weight(user_ups, user_downs, vote_type):
    # vote_type is either UP_VOTE or DOWN_VOTE
    vote_index = 0 if vote_type == UP_VOTE else 1
    bounds = confidence_bounds(user_ups, user_downs)
    # if less than 50% probability of current vote type,
    # then return the whole vote
    if bounds[vote_index] <= .50:
        return 1

    other_weight = max(bounds[1 - vote_index], 0.5)

    return 1 - 2 * (bounds[vote_index] - other_weight)