On Sat, Mar 30, 2024 at 9:54 AM Matt Mahoney <mattmahone...@gmail.com> wrote:
> ...We can measure the fine structure constant to better than one part per > billion. It's physics. It has nothing to do with AGI... In private communication one of the ANPA founders told me that at one time there were as many as 400 distinct ways of measuring the fine structure constant -- all theoretically related. As with a recent controversy over the anomalous g-factor or the proton radius, the assumptions underlying these theoretic relations can go unrecognized until enough, what is called, "tension" arises between theory and observation. At that point people may get serious about doing what they should have been doing from the outset: Compiling the measurements in a comprehensive data set and subjecting it to what amounts to algorithmic information approximation. This should, in fact, be the way funding is allocated: Going only to those theorists that improve the lossless compression of said dataset. A huge part of the problem here is a deadlock into a deadly embrace between scientists need for funding and the politics of funding: 1) Scientists rightfully complain that there isn't enough money available to "waste" on such objective competitions since it is *really* hard work, including both human and computation work that is very costly. 2) Funding sources, such as NSF, don't plow money into said prize competitions (as Matt suggested the NSF do for a replacement for the Turing Test with compression clear back in 1999) <https://gwern.net/doc/cs/algorithm/information/compression/1999-mahoney.pdf> because all they hear from scientists is that such prize competitions can't work -- (not that they can't work because of a lack of funding). There, is, of course, the ethical conflicts of interest involving: 1) Scientists that don't want to be subjected to hard work in which their authority is questioned by some objective criterion. 2) Politicians posing as competent bureaucrats who don't want an objective way of dispensing science funding because that would reduce their degree of arbitrary power. Nor is any of the above to be taken to mean that AGI is dependent on this approach to such pure number derivation of natural science parameters. But there *is* reason to believe that principled and rigorous approaches to the natural sciences may lead many down the path toward a more effective foundation for mathematics -- a path that I described in the OP. This may, in turn, shed light on the structure of the empirical world that Bertrand Russell lamented lacked due to the failure of his Relation Arithmetic to take root and, in fact, be supplanted by Tarski's travesty called "model theory". ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Teaac2c1a9c4f4ce3-M83ab3a14c8c449d907b6fcbc Delivery options: https://agi.topicbox.com/groups/agi/subscription