Example #149 of why it’s difficult to specify bets...
Louie texted me a screenshot showing that Zagat had given an opinion on Subway (the fast-food chain). My girlfriend said “No way,” so we both specified a bet that if we went to the Zagat website, we wouldn’t be able to find a Zagat rating for Subway. She said 40% and I said 65%. When we checked, it turned out Zagat had conducted a survey of people who visit fast food joints, and Subway had been one of the restaurants they got survey results for. So does that count as Zagat giving Subway a rating? I don’t know. I was just thinking of “official Zagat ratings,” rather than survey ratings, but it’s technically true that there’s a rating for Subway on the Zagat website because of that survey of random people who eat fast food.
What i really need is a panel of 5 trusted judges to decide whether my bets are right or wrong, in contested cases.
Digging behind the bet, I think you were betting that Louie had been spoofed. If the screenshot Louie sent was really sourced in zagat and not spoofed, then zagat did indeed have the opinion lukeprog thought it didn’t.
If the purpose of bets is measure your model against the world, it seems to me that the more valuable lesson is how often you are surprised in the process of evaluating the bet than how often you are correct. If you put 40 or 65% on the hypothesis that the restaurant falls in a particular bin, you aren’t surprised either way by the answer, but you both erred in believing that there were just two bins.
Example #149 of why it’s difficult to specify bets...
Louie texted me a screenshot showing that Zagat had given an opinion on Subway (the fast-food chain). My girlfriend said “No way,” so we both specified a bet that if we went to the Zagat website, we wouldn’t be able to find a Zagat rating for Subway. She said 40% and I said 65%. When we checked, it turned out Zagat had conducted a survey of people who visit fast food joints, and Subway had been one of the restaurants they got survey results for. So does that count as Zagat giving Subway a rating? I don’t know. I was just thinking of “official Zagat ratings,” rather than survey ratings, but it’s technically true that there’s a rating for Subway on the Zagat website because of that survey of random people who eat fast food.
What i really need is a panel of 5 trusted judges to decide whether my bets are right or wrong, in contested cases.
Digging behind the bet, I think you were betting that Louie had been spoofed. If the screenshot Louie sent was really sourced in zagat and not spoofed, then zagat did indeed have the opinion lukeprog thought it didn’t.
If the purpose of bets is measure your model against the world, it seems to me that the more valuable lesson is how often you are surprised in the process of evaluating the bet than how often you are correct. If you put 40 or 65% on the hypothesis that the restaurant falls in a particular bin, you aren’t surprised either way by the answer, but you both erred in believing that there were just two bins.