[samfriedman]

>what about machine learning?

The results are probably real, but maybe not for the reasons the author suggests.

[andrewprock]

And here we have the clearest demonstration of confirmation bias.