Where American Transcendentalism Meets Russian Mysticism

In some Youtube of a Berlin event, Chelsea Manning, a political prisoner at the time of this writing, after being pardoned once by a state actor, said we’re all machine learners, or learning machines.

In the home country of Kraftwerk, such “we are the robots” admissions are not out of place. She merely meant that we’re like devices equipped with Bayesian algorithms. Which means what again?

The Bayesian feedback loop, often coupled with Markov chains, applies weight to prior beliefs, then modifies them depending on how freakishly anomalous an event was, based on prior belief predictions.

“If someone with your beliefs would be unlikely to predict events that are happening more often, maybe these specific beliefs of yours are unlikely to hold water, so give them slightly lower weight.” And repeat. Over time, you’ll fine tune, as hindsight wisdom piles up. That’s if you don’t selectively ignore specific feedback. Blind spots remain a challenge, no matter whether you’re a Bayesian or a Frequentist or somewhere in between.

“If these anomalies keep happening, it’s time to jump ship, paradigm-wise, but then that’s a big if.” That’s not an unusual way to think, right?

How did those priors (existing biases) get established in the first place?

Not against a gale of counter-evidence we might surmise.