I'm game to play the game, but I'm not sure how to game the game.  In 
short, it would be better if we could make sure we are comparing apples 
to apples, pcs to pcs and so forth.  Perhaps the challenge ought to 
include, at least:

	1) Problem statement and requirements (ok here)
	2) Underlying test data, if any (I didn't see it anywhere to repeat 
the experiment at home)
	3) The problematic Smalltalk code (ok here)
	4) The original code (in this case Python) with which to 
compare <=== missing

At the end of the experiment, I would be quite surprised to see Python 
outperform Squeak substantially in this arena.  (As is apparent already, 
Yoel appears to have his Squeak oddly misconfigured -- but this doesn't 
mean that Python on his machine isn't also running slower, and hence the 
Squeak *is* the culprit.)  At the moment, the "challenge" creates a 
nonfalsifiable proposition -- for there is nothing with which to 
compare.  Let's see the original code and the test data on which they 
are compared -- in this way we could get a sense of whether we are 
measuring system against system, or problem solution approach against 
problem solution approach.

At any rate, this "challenge" as stated doesn't clarify for me what I am 
trying to do in Smalltalk that is supposedly more naturally or 
efficiently done in Python.