Can we learn algorithms from people who compute fast: An indirect analysis in the presence of fuzzy descriptions

Abstract

In the past, mathematicians actively used the ability of some people to perform calculations unusually fast. With the advent of computers, there is no longer need for human calculators - even fast ones. However, recently, it was discovered that there exist, e.g., multiplication algorithms which are much faster than standard multiplication. Because of this discovery, it is possible than even faster algorithm will be discovered. It is therefore natural to ask: did fast human calculators of the past use faster algorithms - in which case we can learn from their experience - or they simply performed all operations within a standard algorithm much faster? This question is difficult to answer directly, because the fast human calculators' self description of their algorithm is very fuzzy. In this paper, we use an indirect analysis to argue that fast human calculators most probably used the standard algorithm. © 2012 Springer-Verlag Berlin Heidelberg.