>>20
Yes, but 10% is a complete
joke. I'm not trying to flame, and I hope you won't take it personally; it's a statement of fact.
In the end we rewrote some algorithms so they were (almost) embarrassingly parallel, and dropped cash for a couple racks of blade servers. Performance improvement was somewhere in the vicinity of 1000x, not 1.1x. So you can see why 10% isn't of much interest.
How many non-artificial situations can you think of that 10% will make a difference? Especially when psyco is neither proven nor particularly stable?
Psyco was an interesting development, but any benefits it will have will occur in the knowledge that was acquired during its creation, and used in things like Pypy. I have high hopes in that regard.
Also, I'd like to add that as nice as Python is, there are languages out there that are both a lot faster and more powerful.
Their main problem is the usual: too few libraries.