efficiency

Cosma writes about power laws and psycho killers;
The usual story - somebody sees a power law when e.g. a log-normal distribution would be a much better fit.

This suggests that three toed sloths and the internet are not fully efficient distributors of information.



Meanwhile, on intrade I saw (Jan 24, 8am ET) some interesting odds for the Dow Jones Industrial index, suggesting that markets are not always fully efficient either:

Dow Jones to close ON or ABOVE 11750 on 31 Jan 2012: probability = 91.5%

Dow Jones to close ON or ABOVE 12000 on 31 Jan 2012: probability = 93.0%

Dow Jones to close ON or ABOVE 12250 on 31 Jan 2012: probability = 97.0%



So I wonder about the odds in a cliché betting market...