Assign each real-valued score X from 0 to N the value B^X where B is a chosen positive real in the open interval (0,1). Allow voters to score each candidate with discrete integer scores from 0 to N, where no score indicated is set by default to 0.

Next, assign each candidate C the metric

M[C]=SUM{all scores X given to C} Nc(X)B^X

where Nc(X) is the number of X scores given to candidate C. The candidate with the largest metric is eliminated from the race, and the Cardinal Baldwin renormalization recalculates scores.

The process repeats until two front-runners remain, and then there is a STAR runoffâ€”the front-runner who is scored higher than the other more frequently among the voters is declared the winner.