Wanted: Electoral College Contest beta-testers!
I've been working on setting up the second quadrennial BrendanLoy.com Electoral College Contest, and tomorrow night, during the VP debate liveblog, I'll be doing some testing of my Election Night live-results-posting system. To facilitate this process, I need 10 beta-testers to submit electoral picks today. They can be fake picks -- you're not really signing up for the contest, just helping me test the signup system. So, by all means, pick Obama to win Texas or McCain to win California. It doesn't matter. :)
[UPDATE: Beta-testing is now closed. Thanks, everybody! Stay tuned for tomorrow night's liveblog and live test of the results-posting thingy.]
Incidentally, I'll be changing and simplifying the rules of the contest this year. Long-time readers might remember that there was some controversy four years ago about my scorekeeping system. It didn't affect the result, since Mike Wiser was a perfect 538-for-538, and thus would have won in any imaginable system. However, it caused some consternation that, for instance, Scott Schmidt finished ahead of Brooke Bennett, even though Brooke only got one state wrong (Ohio), while Scott got Ohio and four other states wrong, but had a lower "differential" because his errors roughly "canceled each other out." This year, I'm going to use the "differential" only as a tiebreaker; the winner will simply be the person who gets the lowest number of electoral votes wrong. So, under these rules, Brooke would finish 3rd instead of 7th, and Scott would finish 10th instead of 3rd.
To understand what this means, suppose Obama "flips" Iowa, New Mexico and Colorado from 2004, resulting in a 273-265 victory. Now suppose Contestant A has a perfect map except New Mexico, which he thought McCain would win, thus winning the election 270-268. This means that Contestant A, despite picking the wrong winner, was just 5 electoral votes off from the actual result. Meanwhile, suppose Contestant B has a perfect map except for two mistakes: he thought McCain would win New Mexico, and he thought Obama would win Nevada. Because NM and NV have 5 electoral votes apiece, these mistakes "cancel out" perfectly, and Contestant B correctly predicts, with perfect precision, the overall result: a 273-265 Obama victory. Under last year's "hybrid" scorekeeping system, Contestant B would defeat Contestant A because of his zero differential, even though he got 10 electoral votes wrong (compared to Contestant A's 5 wrong). However, under this year's system, Contestant A would win, because he made the fewest mistakes, as judged by the number of EVs predicted incorrectly. Upon reflection, I think this is fairer, and it's certainly simpler.
Anyway, I'll have more to say about the contest when I officially open it for entries later this month. (And, in the mean time, if someone wants to make the case for a better scoring system, go right ahead. First, though, you may want to review the scenarios discussed here.)
[UPDATE: Beta-testing is now closed. Thanks, everybody! Stay tuned for tomorrow night's liveblog and live test of the results-posting thingy.]
Incidentally, I'll be changing and simplifying the rules of the contest this year. Long-time readers might remember that there was some controversy four years ago about my scorekeeping system. It didn't affect the result, since Mike Wiser was a perfect 538-for-538, and thus would have won in any imaginable system. However, it caused some consternation that, for instance, Scott Schmidt finished ahead of Brooke Bennett, even though Brooke only got one state wrong (Ohio), while Scott got Ohio and four other states wrong, but had a lower "differential" because his errors roughly "canceled each other out." This year, I'm going to use the "differential" only as a tiebreaker; the winner will simply be the person who gets the lowest number of electoral votes wrong. So, under these rules, Brooke would finish 3rd instead of 7th, and Scott would finish 10th instead of 3rd.
To understand what this means, suppose Obama "flips" Iowa, New Mexico and Colorado from 2004, resulting in a 273-265 victory. Now suppose Contestant A has a perfect map except New Mexico, which he thought McCain would win, thus winning the election 270-268. This means that Contestant A, despite picking the wrong winner, was just 5 electoral votes off from the actual result. Meanwhile, suppose Contestant B has a perfect map except for two mistakes: he thought McCain would win New Mexico, and he thought Obama would win Nevada. Because NM and NV have 5 electoral votes apiece, these mistakes "cancel out" perfectly, and Contestant B correctly predicts, with perfect precision, the overall result: a 273-265 Obama victory. Under last year's "hybrid" scorekeeping system, Contestant B would defeat Contestant A because of his zero differential, even though he got 10 electoral votes wrong (compared to Contestant A's 5 wrong). However, under this year's system, Contestant A would win, because he made the fewest mistakes, as judged by the number of EVs predicted incorrectly. Upon reflection, I think this is fairer, and it's certainly simpler.
Anyway, I'll have more to say about the contest when I officially open it for entries later this month. (And, in the mean time, if someone wants to make the case for a better scoring system, go right ahead. First, though, you may want to review the scenarios discussed here.)
posted by Brendan Loy at
10/01/2008 12:56:00 PM
