cypherpunks-legacy

Download

cypherpunks-legacy@lists.cpunks.org

November 1998

1 participants
12 discussions

blunder, or forecast?
by Adam Powell 05 Nov '98

05 Nov '98

Once again the Net had it first, this time by accident... http://www.freedomforum.org/technology/1998/11/4abc.asp ABC News election-eve Net 'mistake' accurately predicted most close races By Adam Clayton Powell III World Center 11.4.98 It did not happen, but just imagine if it had: The night before Fox television aired the World Series, the Fox Web site posts a complete set of inning-by-inning box scores for the upcoming Yankees-Padres games. After leaving the results online for a while, Fox pulls them down and says it was all a big mistake. Then the Yanks and the Padres play the Series, and nearly 90% of the innings show the same scoring as in the "mistake" posting. Sports fans, reporters and bookies would all cry foul, right? Now consider what happened this week: On Monday night, ABC News accidentally posted complete state-by-state election results on its Web site, hours before any votes were cast yesterday. ABC withdrew the numbers by mid-evening Monday, saying they were all a mistake and a test of their systems. "It wasn't our finest hour," Michelle Bergman, manager of communications for ABCNews.com, told the Associated Press yesterday. And, indeed, in the days before each national election, the major national news organizations hold "rehearsals" for reporting election night, using "dummy" numbers. Or maybe ... With almost all election districts reporting, those "phony" ABC News test numbers on Monday accurately matched the outcomes of the Senate and governor races in 61 of the 70 contests ^× 87%. It would be difficult to find a political analyst, pundit or bookie who even came close. While every major analyst on Sunday was predicting the Republicans would pick up anywhere from one to four Senate seats this week, ABC's test numbers on Monday had it right on the money: a 55-45 GOP-Democrat split, for no net change. Even more remarkable, in some of the most closely watched contests, ABC News election eve "test" numbers matched the final vote count almost precisely ^× within one percentage point. In the Florida governor's race, Jeb Bush beat Buddy MacKay by 55% to 45% ^× the exact final result rehearsed by ABC News on Monday. In Texas, Jeb's brother George won by 69% to Mauro's 30% ^× the very result used in the ABC rehearsal on Monday. ABC News rehearsal numbers also matched the exact final results, to within one percentage point, of the governors' races in Alabama, Colorado, Wisconsin and Wyoming. All told, ABC's "error" had the correct candidates winning the governors mansions in 32 of 36 elections yesterday. The only gubernatorial contests where ABC had the wrong candidate winning were in Hawaii, Iowa, New Mexico and Minnesota, where few predicted Reform Party candidate Jesse Ventura would be the new governor. In the Senate elections, ABC's test numbers matched the winners in 29 of 34 contests, including all of the major races. ABC on Monday had posted "WIN" indicators next to Boxer, Schumer, Fitzgerald, Murray and Hollings, all of whom won close races in California, New York, Illinois, Washington and South Carolina, respectively. ABC also had the correct result in Wisconsin. The final vote totals this afternoon showed Russ Feingold narrowly won reelection by 38,410 votes. On Monday, ABC's rehearsal numbers showed Feingold winning reelection by a margin of 39,000 votes. But ABC is not claiming these numbers were the result of any new forecasting models or special analysis. "It was completely random," Bergman told free! this afternoon. -------------------------------------------------------------------------- POLITECH -- the moderated mailing list of politics and technology To subscribe: send a message to majordomo(a)vorlon.mit.edu with this text: subscribe politech More information is at http://www.well.com/~declan/politech/ --------------------------------------------------------------------------

1 0

blunder, or forecast?
by Adam Powell 05 Nov '98

05 Nov '98

1 0