Score one for Menzie

 In many cases he is right to do regressions on log. He is looking for disequilibrium between two partitions. Some point where the moments do not match, a sudden collapse or expansion of k, the velocity, really. In a public auction, Menzie can tell when traders are crammed down a hall way. Physicists do the same when they make the quark matrix.  What is the probability of jumping rank in the Chinese remainder game, making the complete sequence longer or shorter, by an exponent.

Logbinomial takes you from Bayes to sample space, the Boltzmann function.  The base in exp determines the samples per element of some finite set. So log number Covid cases tells you the length of the line at the hospital. And you can easily find any jump in line length by regression on the log, and that tells you when the collapse comes.

He is using Newton's short cut, but so what? He finds the breaks points soon enough. He is working ex post, but the method is semi predictable, it gives an estimate of N, the market size.