I like to understand things from scratch so I usually try to go to the bottom from where things started and how the code evolved, the version control helps me with that but for Jupyter Notebooks, storing in JSON format doesn't give the proper representation.
There are packages like nbdime but it's just an extra thing to do, extra command to write, How do you deal with it?