I think the worst part is them not optimizing CPython earlier, which means that people now rely a lot on CPython internals, which means stuff like Pypy isn't compatible with the Python ecosystem. And then you end up with everyone having their Python optimization: Dropbox, Instagram, Pypy.

It's when you put all these small optimizations together that it leads to something remarkable. It's analogous to video codecs: there are a bunch of individual optimizations that alone don't look that impressive, only saving ~1% here or there. But once they all are working together, you see savings of 10-50%.

I think what's surprising is how small these optimizations are. A few proposal sped up Python a lot, but were partially incompatible. Meanwhile OCaml is removing the GIL from the language, adding multicore and added stuff like Flambda in the last decade, all of that while mostly keeping backwards compatibility and not sacrificing single-core performance. All of that while having way less people working on it than Python.

Yeah but Python is on the order of 100x slower than "fast" languages. 50% faster would obviously be nice but it's not really going to make Python a fast language.

I'm looking forward o seeing numbers per python release, for what the optimizations result in taken together. Mark Shannon is going at it with optimizations now.