The spiral problem in particular

 The Fed tax, it grows according to the deficit. But the regulated banking loss to shadow banks grows as the tax increases.  The tax income of the swamp ios proportional to regulatory banking.

That is the correlation problem, the central banker is not innovative, it does not respond first in line because it is correlated with its major customer.  It is like jump off to the (2,t,z) chain of Markov, and the shadow banks are the alternative node on the (1,y,z) chain. The 1,t,z chain is the most efficient for S/L banking, and the central bank wants to minimize the inherent correlation with Congress.  That is kind of its main goal, minimize entropy spent on the 2,y,z line. That means predicting another small adjustment to its contract in fifteen years or so. That is some sixty trades on a quarterly basis. It is a provable contract, all parties can estimate their own risk ex ante. It is the hidden Markov model.

On the 1,y,z chain the pit boss want 1 out of N coin tosses, the rest split between S and L. It is really working against N+1, if it includes itself. In the 2,y,z modes, the traders are working two interleaving pit bosses, which are themselves partitioned.  I am thinking that the binomial split by twos, the coin toss allocation suffers when 2/K implies entropy in the pit boss. If transaction costs are trivial, the just break this into separate 1,y,z pits, the traders responsible for splitting entropy.

So the goal for the New Fed is to split duties.  There are two pits. One where we are negotiating asynchronous S/L and the other where we negotiate inflation tax.  If N went to infinity and we were lognormal, then we have a regular economy with an independent limited Gaussian sales tax from the Swamp. Every 15  years or so we negotiate that tax in the legislative auction house with restricted entry and exit. Keep the banks out of it as N goes to infinity. But in the finity we get Senators who evaluate the small sales tax and adjust it accordingly.

The Treasury can apply this tax on the bond market, by bonds with the true fiat. It can spend it. Spend it on the senators state capitals. Pay off the small states. Splitting the cash by state is a tax on the House of representatives, that is the point. The House then knows the cost of state imbalances, the inefficiency losses from ear marks. That little blunder in the Constitution is liquid, we can move around it.

If the cash is sent to state capitals, then it is a tax on big states and tends to balance N, and thus entropy over the long term.