In 1936, Francis Scott Fitzgerald(1) wrote: “The test of a first-rate intelligence is the ability to hold two opposing ideas in mind at the same time and still retain the ability to function.”(Esquire Magazine, February 1936). The same year, Alan Turing laid the foundations of modern computing. Therefore, to achieve this intelligence, a computer (or machine learning) should pass “the Fitzgerald Test”. Let’s call to “the ability to hold two opposing ideas in mind at the same time”, as the capability to select a distinction. If according Niklas Luhmann(2), following to George Spencer-Brown(3), the distinction begins a process of building complexity by reducing the complexity, and if “in machine learning the complexity is in the data”(4), then a universal algorithm should be able to perform this operation of reduction by establishing distinctions. I propose to call this algorithm, the GDSD Algorithm: Given Data Set Distinction, where “set” should be read as “establish”, as an order. It would have the power to apply across all fields of human knowledge.



(1) F. S. Fitzgerald, The Crack Up. New Directions, 2003.
Niklas Luhmann, Introduction To Systems Theory. Polity Press, 2013.
George Spencer-Brown, Laws Of Form. E. P. Dutton, 1979.
Pedro Domingos, The Master Algorithm. Basic Books, 2015.


Niklas Luhmann, La Ciencia De La Sociedad. Universidad Iberoamericana, 1996.



Héctor Corvalán, A Universal Algorithm For A Hypothetical Machine Learning

Published in: on junio 24, 2017 at 9:45 pm  Dejar un comentario