Matthew Yglesias has gone too far this time.
He reads a report on the effect on 2 year Treasury yields of public speaches and interviews of Fed Open Market Committee members measured by the change from 15 minutes before the announcement to 2 hours after. It finds effects on the order of one basis point per announcement. The average absolute effect of Bernanke's communications is 1.5 basis points, that is 0.015% on one interest rate definitely lasting at least 2 hours (and maybe forever or growing or something).
Yglesias concludes "I think this highlights the fact that monetary policy works largely through expectations and communications ...". If monetary policy works largely through effects on medium short term interest rates on the order of a basis point, then it doesn't work at all. This is, pretty much, my view of the current situation. I have the radical idea that monetary policy in a liquidity trap is almost ineffective.
The alarming thing is that Matthew Yglesias who is very very very smart, doesn't seem to have noticed the scale of the graph at all. He doesn't distinguish "detectable" from "important." His comment shows no effect at all of quantities.
I feel as if his evil (really innumerate) twin Skippy takes over when he writes about monetary policy.