Dave Ketchum wrote: > In simulation there is value, and sometimes excessive temptation, in > tailoring test cases to favor a desired result.
Maybe try an open simulator. Make the "electorate engine" pluggable so experimenters can try different voting behaviours. That should protect against bias. > In vivo, as I proposed, you get all kinds of test cases exposed to > multiple formulas, but not necessarily a good variety of test cases. It's nice to go live, but the up front costs will be high. It's risky too because you have to follow the crowd. Sites will offer alternative voting methods and electors will vote with their feet. There's no telling where they'll be attracted, or whether it'll jive with the test plans. -- Michael Allan Toronto, 647-436-4521 http://zelea.com/ ---- Election-Methods mailing list - see http://electorama.com/em for list info