T Taylor Rule in maximum entropy showing connected queuing with my sloppy math

John Taylor, Economics One,  wants this rule implemented as Law:

The term ‘Reference Policy Rule’ means a calculation of the nominal Federal funds rate as equal to the sum of the following: (A) The rate of inflation over the previous four quarters. (B) One-half of the percentage deviation of the real GDP from an estimate of potential GDP. (C) One-half of the difference between the rate of inflation over the previous four quarters and two. (D) Two.

What is the maximum entropy solution? That means, what is the minimum span graph that adds up the finite log of inflation quants, in the optimum base. In other words, pack inflation things in a box such that all inflation things get counted up with the minimum number of nodes.

His equation look to be like a quadratic operator on a recursive sequence of measurements. He is measuring drunken, aggregate, retail  price changes.
The Shannon uses base 2, since we are finite and integer and quadratic in variation.

 Sloppy, handwaving math:

p = log2(1 + Q(dn)/Nsquared) or
2^p-1 =  Q(dn)/Nsquared

The Noise and quadratic root set the decimal point. Inflation things are good to about 3%, so make this a six bit counter of inflations.  Q is quadratic in dn, dn being the look back in the sequence of inflations. The Noise sets the discount, so you put inflation things in the fractional end and they take two bits to count up to one inflation thing, your sampler is Poisson (I think) over a set of two measurements..

Then we take the square root of the polynomial, real double root to make it easy. We make that a decimal point, I think,  on the power digit series. The discount comes from channel noise. In other words, two parameters, the weighting or quant per sample, and the decimal point.

This digi series:

sum of {2^(d* i/2)}, i is integer 1 to 8 bits,
Each digit has two valued coefficient, 0 and 1.

 This gives us the minimally redundant inflation yard stick, and is a rank three network of bankers, each banker operating in a market.  No queue will exceed a maximum (2) or go empty. The inflation is measured, becomes the sum of the power series in two digits, goes right into the rule set to compute rates, as a yield curve from about 1 to 3 months. This is Poison with a bounded random arrival time?

The thing it measures is a gaussian variable about r, r being the real root of the polynomial, that is (delta inflation +- root). It does so with the minimum of banker salaries.