Tuesday, November 24, 2015

A small leap for Man and a big jump for Math

Recent news that a University of Chicago mathematician may have reduced a NP complexity problem of network comparisons to that akin to P level complexity, signals a jump in knowledge in otherwise stagnant field. Consumed by big data and bigger noise, mathematicians and data scientists have been burning the midnight oil, solving everything under the Sun, using century old techniques and faster computers. In the process, most forget to think and step aside to see the challenge in front of them could be simplified before diving deep.

In the age of cheap hardware and companies plush with cash, innovation appears stagnant. Making a neural net with thousands of computers in a network is not innovation, it is just a show of brawn over brains. Pumping large number of rules through a supercomputer in an attempt to beat a human recollecting random facts is not innovation, it is scaling ignorance. Collecting, storing and analyzing large amounts of noise in an attempt to discover complex heuristics is not innovation, it is just sticking one’s head in the cloud.

Innovation happens but only rarely. Reducing the complexity of a problem class, fits the bill perfectly.