I've often thought about finance blogs from a binomial distribution point of view. Say, that out of the 25M+ finance blogs (http://bit.ly/ep75e) there are about 100k serious ones. Say that each week for a year each blog makes a verifiable, binary outcome prediction. Also assume that there is a 50-50 chance of that prediction coming true. We could expect 4 or 5 of those 100k blogs to be right 40 weeks of the year JUST BY GUESSING. Let alone some actual insight. Say you were one of those winning blog readers, and week after week this blogger keeps getting things right. Like Bear Stearns going bankrupt, for example. Every now and then (s)he could get something wrong, but there would always be a reason, or (s)he could say that the outcome is just temporary, and that the true outcome is anytime now.

You (and eventually, your friends) would start investing your money along his/her advise, and if the blogger is really, really popular (s)he could start moving the market enough to change the odds from 50-50 to 60-40.

I don't really know how to stop this, but it is interesting.