Sumner on optimal vs feasible

In my food policy book, I discuss the difference between advocating an optimal policy and advocating only among the list of feasible options. Per has often mentioned policy analysts' disillusionment with policy makers who don't enact their policies - chalking it up to lack of political will - when the fault more accurately lies with analysts who didn't consider the full set of solutions.

Scott Sumner is grappling with this issue as well. The policy he usually puts forward - having the Fed targeting 5% NGDP growth - assumes on average 3% real growth and 2% inflation. Today he notes that his 'ideal' inflation rate is actually either less or more than that, depending on whether we're in optimal world or feasible world:

If we were to assume the Fed adopted NGDP forecast targeting ... then we would not have to worry about “liquidity traps,” i.e. the zero-rate bound on conventional monetary policy. In that case we’d be better off with a lower inflation rate. ... [Zero percent or negative percent inflation] are sensible ideas if we have sound monetary policy, as inflation is a tax on capital, and lowers the rate of economic growth.

On the other hand if we don’t have a sensible monetary policy regime, then low inflation makes an economy more susceptible to bumping against the zero-rate bound. ... So let’s suppose you have a central bank full of meek, timid souls. What sort of inflation rate is optimal? I’ve mentioned that you’d probably be better off with an inflation rate even higher than 2%.

He then goes on to praise Australia's 3-4% inflation that has helped keep them out of a liquidity trap and a great recession, or indeed any recession for 20 years.