Consider the particle collider

They are conducting a test with N lesser and lesser fuzzy as they get better.

The goal, from a Markov tree, is to get the experimental apparatus up a notch on the tree, and measure events from way farther up the tree, at lower resolution.  But the farther up the tree the experiment, the narrower the bundle of possible source paths. They have an external sense of N density from their general study of beach balls They can do a Bayesian solution.