This week, Twitter upped their single-tweet character limit from 140 to 280, purportedly based on this interesting analysis of tweet lengths published on Twitter's engineering blog. The gist of the analysis is this: English language tweets display a roughly log-normal distribution of character counts, except near the 140-character limit, at which the distribution spikes:

The analysis takes this as evidence that twitter users often "cram" their longer thoughts into the 140 character limit, and suggest that a 280-character limit would more naturally accommodate the distribution of people's desired tweet lengths.

This immediately brought to mind another character limit that many Python programmers face in their day-to-day lives: the 79-character line limit suggested by Python's PEP8 style guide:

Limit all lines to a maximum of 79 characters.

I began to wonder whether popular Python packages (e.g. NumPy, SciPy, Pandas, Scikit-Learn, Matplotlib, AstroPy) display anything similar to what is seen in the distribution of tweet lengths.

Spoiler alert: they do! And the details of the distribution reveal some insights into the programming habits and stylistic conventions of the communities who write them.