Theory

Congestion management is equivalent to removing redundancies in flow.  It involves lossy compression.  Example: If customers are mostly buying four cans of soda, why not force the ad hoc customers to conform and sell in cases of four?

Redundancy minimization is equivalent to entropy maximizing. The second part says entropy maximizing flows have one form of solution, any other correct solution can be adibaticaly warp into the one abstract solution. The proof is via contradiction, showing the redundancy when two solutions are 'relatively prime'. Relatively prime means they fail the adiabatic warping test. They do not scale.

There is one prime generator.

Bounded error, is an assumption, but accepted by observation. Translated it means waiting in line too long exceeds the error bound at equilibrium.  With the virus running around we can see a evolutionary desire to avoid unstructured queues, congestion.  Bounded error every where holds, it is up to the application to show this true. Nash has to hold on this one. Coase shows p in the need for auto traded pits at congestion points.

The unique part is the calculations, we are actually doing the model in real time. The digital transformation follows the theory, we are cutting and pasting from the math labs to the digital trading pits. The math labs only work with the complete sequences, a marketing skill, actually. The quants, us, are simply social security number which do not like to wait in line, but that is exactly what our smart cards do, find the shortest line. The pits we deploy and somewhat like a Hufman encoder, but not always. Coinbase, Amazon, Walmart, grocery stores all play the digitization game, drop transaction costs to trivial, find the congestion points, make a pit. All sandbox.