| |
BGonline.org Forums
My favourite probability problem
Posted By: playBunny In Response To: My favourite probability problem (joe freedman)
Date: Friday, 28 August 2009, at 11:45 p.m.
In that case it's important to know as exactly as possible what chances the Predictor has of getting it right. All we know at the moment is that it's near-certainty, at least well above 99% success rate. I'm happy to use logic when we're talking about a "near-certainty" but if the discrepancy were great enough then a bit of maths would be in order.
When accepting the $1000, how much risk are you taking if the Predictor is correct exactly 50% of the time. 50% of the time there'll be an empty envelope and 50% of the time it will be $M.
How much risk are you taking if the Predictor is correct exactly 99% of the time. 99% of the time there'll be an empty envelope and 1% of the time it will be $M.
But this Predictor is nearly certain to predict your action. Not predicting a hypothetical action but your actual action. That's what it means for it to be a predictor! If the "well above 99% success rate" is not bogus information then you must project your future action back as part of the Predictor's prior knowledge.
And then you act accordingly. You decline the $1000 and have forced the Predictor to give you the $M by virtue of its success rate - which rate is, effectively, your success rate! ;-)
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.