On 08/24/2012 04:56 PM, Michael Ossipoff wrote:
I've printed the program out, and now I feel that I've probably found and corrected all of its errors.
Do you have any test vectors? Having tests would make it a lot easier to see whether you have actually corrected all the errors.
In the case of an election method, those test vectors would be elections where the method you're describing gives different results than as many other methods as possible. You'd have expected input (the ballots) and output (the winner or result ordering).
Even better ones would also exercise edge case handling, like tiebreakers or different parts of the logic of the method.
---- Election-Methods mailing list - see http://electorama.com/em for list info
