Tyler Cowen posting on web bot movie banking

He send us to Kottke which sends us to Tim Wu at the New Yorker. Here is the summary:

Tim Wu writes for the New Yorker about how Netflix uses a ~70/30 combination of data and human judgment to determine their recommendations and what shows/movies to make.

So, no doubt we have the web bot managing the power spectrum of human movie choices.  The discussion was about how much inventory slack in movie selection should the bot maintain and how much self correlation among the humans it should retain. It boils down to the frequency at which a network of humans watch movies.

If humans are assumed movie stable and movie makers perfectly prescient, then the bot wants to sample human movies choices at twice the rate, assuring perfect causality a constantly equalibriated market.

If the bots want to maintain connectivity among humans such that one choice it additive with prior choices, then it samples human choices at Phi, or rather continually estimates the Phi among the humans.

If the bot wants to ensure that no human crowd onto a particular movie, then it sample at 3/2, the optimum congestion rates. This assume movie makers are perfectly adaptable to human desires.

The correct Black Scholes result will be (Phi-1)/2, or about 31% of the movie recommendations come from the bot. This matches movie watchers, movie makers and external markets.

Regarding those sample rates.

That is about where I started on the idea of a band limited vacuum some two years ago.  It did not get the picture back then, and some of this is not complete.  But I learned a  lot from the mathematicians on the web, and I asm entirely grateful.  At this point, I have a hard time seeing anything but the Lagrange, Markov, Fibonacci, Wythoff, Weiner, Group, and Schramm experts getting together. I seems to me these are the folks that make the formal proofs, and set everything straight in the Theory of Everything. I would think they are right there, staring and contemplating the center where everything meets. I gush over these folks, I think they are brilliant.
In a bu8bble space I considered the sample rate to be the exchange and relation times, an asymmetric, self clock. So, in the Shannon model, information rate, channel rate and noise rate were all self equilibrating. They adjust, as in measure theory, such that the tinies measure themselves, and the entire thing is a self adapting rules. Some finite tinies making the optimum theory of themselves.