I just meant to include that as a real world example of the effects of upgrading from 2 to 3. But the blog post in question does date back to 2016 so I don't know how that's unfair :)

It's actually comparing 2.7.12 (2016) with 3.6 (2016).

Not that the release date has much to do with the benchmark, mind you...

Which is exactly why it's ok and not unfair to compare the performance improvements over time of Python 3.6 versus Python 2.7?

The point is that even at that time, Zed was advocating staying on py2 because py3 had no future, or wasn't turing complete, or whatever is complaint du jour was.