Read his latest post, which outlines many key but usually unstated assumptions behind monetary theory and policy. It is one of the most instructive econ posts to appear in some time.
That said, on the policy issue I think one of Krugman’s earlier posts (I can’t find it) is closer to the mark. With or without a liquidity trap, monetary policy can’t fix negative real shocks and — here is now the earlier Krugman — monetary policy can’t make insolvent (or potentially insolvent) banks whole. That’s my take on why the Fed is relatively powerless, not because of a liquidity trap. If you believe, as a Keynesian would, that insufficient aggregate demand is the problem in the first place, you will be relatively worried about liquidity traps. If you believe, as a neo-Austrian would, that malinvestments and coordination problems are the key issues, you will look toward other factors which limit the power of central banks to restore order.
In my view sometimes the Keynesian perspective is relevant, but not so much today. As the contraction of credit spreads through the Fed-regulated banking sector, however, and the broader money supply aggregates come under stronger negative pressure, the Keynesian perspective is likely to become more relevant. That is in fact my major medium-term worry and we probably should be pessimistic in this regard.
There is a separate and very important liquidity issue about restoring the markets and valuations for bank loans, but this is not a liquidity issue in the sense of Keynes’s portfolio theory or the traditional liquidity trap.
Addendum: Brad DeLong adds comment. Another way of putting my point is this: in the situations where a liquidity trap might be binding, there is usually some even worse constraint which is more binding, thereby making the potential liquidity trap not so much a problem at the relevant margin.