Tuesday, July 04, 2017

Ockham's Guillotine

It's been:
  • 2 years since Google DeepMind's AlphaGo defeated a professional Go player
  • 11 years since Marcus Hutter sold his Ferrari (probably a joke, but I'm not sure) to underwrite the €50,000 purse for the compression prize I suggested, and I suggested students get their PhD's under Hutter.
  • 80+(!!) years since Harold Jeffrey's work based on Russell and Whitehead's Principia Mathematica as a precursor to the Turing formalism's foundation of Solomonoff's papers.
The above provide strong support for lossless compression as the test for quality of theory:

The formalization of Ockham's Razor.


The Electronic Frontier Foundation's AI Progress Measurement seems to be going vertical. 


Yet the best we seem to be getting out of computational unified social etiology is stuff like Pathogens and Politics: Further Evidence That Parasite Prevalence Predicts Authoritarianism which, while impressive given the studied ignorance of disciplines like political economy, is still abjectly narrow. 


In proposing compression of a broad range of longitudinal social data, as an approach to unified social etiology, I seem to be operating in such a rarefied environment that there is precisely one active researcher who "gets it" out there.  His response was sympathetic if not enthusiastic. 


NGOs are rightly worried about "bias" arising from "big data" and "AI", but they're clearly not firing on all cylinders.  Indeed, if this public position by the Ford Foundation is any indication, they're coasting on momentum. 


"Bias" is exposed by cross-checking measurement operations and cross-checks are, ultimately, dependent on a variety of relatively independent measurement instruments being brought into consilience.  This is precisely what cross-disciplinary data, metadisciplined by Ockham's Razor, accomplishes.


Who are the serious individual philanthropists that should be supporting something like this?  Are there any?  


It is, of course, too much to ask that well endowed institutions like Harvard, let alone the government, support the destruction of their own theology.  In prior centuries, "elites" such as these would be put to the guillotine.  Indeed, it may be that the one thing that may save them from a new French Revolution is what I have proposed above:


Ockham's Guillotine