I had a personal conversation with Larry Hastings (the presenter) at PyCon. Here are a couple of notes from that chat, phrased neutrally. Some of these points may be reïterated in the linked video:

- We can view this work is as a revisiting of Greg Stein's GIL-removal attempt in Python 1.4:

http://dabeaz.blogspot.com/2011/08/inside-look-at-gil-remova...

It seems wholly reasonable to revisit the approach in light of how the language and ecosystem have changed since 1999.

There are demands made of CPython core developers to remove or address the problem of the GIL, and these efforts demonstrate how much work is necessary to do that successfully.

- Comparing single-threaded performance in a GIL implementation against single-threaded performance in a GIL-less implementation is considered an unfair comparison. A GIL-less will do extra book-keeping that will necessarily result in slower single-threaded performance.