Setting our vision for the 2017 cycle

Note that Python did not standardize on numpy early. numpy is the result of many failed older projects. Some of it’s predecessors wore written by core teams staffed by the biggest names in python including Guido van Rossum. The second attempt was Numeric but it did not serve all use cases so Numarray was released. Many packages supported both for a long time. NumPy was a N+1 atempt to salve it. While not the first, it was successful. (Note I was not involved then. This is based on interviews of the involved parties.)

TLDR. NumPy is a great example of the advantages of a unified interface. But it is not obvious how a community gets to that state quickly. Python did not get there by the top down method, despite trying. And the bottom up method sometimes takes a long time.