Regarding the past:
Because their initial reviews didn't contain requests to change the variable names to match the previously existing version?
Regarding the future:
Any time a new version comes out, it needs to have agreed upon variable name uniformity enforced. So when something gets created in "nupic.research", if it introduces a new variable name or changes a variable name, then "back-porting" those changes to their corollaries, needs to be enforced.
Regarding the present:
A decision body needs to be convened to enact changes to all basic algorithms (including their use in frameworks such as OPF, or Network API) - and variable names need to be normalized across all expressions of their types.