The first lecture explains the basics of how central banks take econometric models and a policy satisfaction function and use these (with some dynamic linear programming) to set policy instruments. Blinder acknowledges that this involves a whole pile of assumptions — about model selection, lags, and stochastic effects and of a linear economic model and a quadratic policy satisfaction function — and that it has severe limitations, but argues that it's the best that can be done and that the alternative is "asking one's uncle". (He feels obliged to argue for a dynamic programming approach that takes into account predictions of the bank's own future actions; this seemed blindingly obvious to me.)
The second lecture considers which policy instruments banks should use. Blinder argues for using "the estimated real short-term rate of interest" instead of a monetarist approach; he also suggests that the neutrality of monetary policy be measured by comparison with a "neutral real interest rate", even if that is "difficult to estimate and impossible to know with precision". Turning to the "rules versus discretion" debate, he takes aim at a range of theories that central banks will succumb to pressure to reach for short-term gains and keep inflation rates too high, and at proposed "solutions" such as using set rules or financial incentives to coerce central bank decisions. Blinder finds none of them compelling: "The academic literature has focused on either the wrong problem or a nonproblem and has proposed a series of solutions that make little sense in the real world."
In the third lecture Blinder puts the case for central bank independence, from markets as well as from governments and public opinion: "Following the markets too closely ... may lead the central bank to inherit precisely the short time horizon that central bank independence is meant to prevent." He also argues against the "credibility hypothesis", that banks with more credibility in fighting inflation can disinflate at lower social cost just because of that: "Much fascinating theory to the contrary, I do not know a shred of evidence that supports it." And he puts the case for greater openness by central banks about the reasons for their decisions, on both political and economic grounds.
I found Central Banking in Theory and Practice a fascinating entry to a world I know little about. I would have liked a fourth lecture that sketched the history of central banking, and perhaps went into more detail about central bank institutional structures. (Blinder touches in the first lecture on the effects of committee-based decision making on Federal Open Market Committee policy.)