Data paradigms are metaphors we think with.
This talk describes how julia re-united a psychometrician's probability theoretical and technical thinking in "data".

I have thought in terms of
boxes, registers and memory adresses,
value bundles with C structures and Pascal records,
objects with inheritance in C++ and java,
vectors and data frames in R.
All useful but oblique.

Finally julia types liberated from tediously translating theory to implemetation.
More clarity, less mistakes.