With the Python community looking forward to Python 3 and PyPy, NumPy+Scipy+Matplotlib seems to be be focused on making further developments in Python 2 (and for some time to come, according to Travis Oliphant's recent post on NumPy, http://technicaldiscovery.blogspot.com/2011/10/thoughts-on-porting-numpy-to-pypy.html). From what I gather, R has a team of ~20 core developers at any given time and transitions among members are generally smooth; with NumPy I've read several times that not enough contributors know about its innermost workings, and that keeping the ecosystem in step new developments in the Python language itself will be difficult. In the end that might mean having multiple Python installations (not that this isn't happening already, especially on Mac OSes), one of which is installed for the purpose of running NumPy+SciPy+Matplotlib. I have already encountered issues with different versions of Python being installed on different network machines, and having to rewrite and maintain several branches of code for different Python versions. I am not sure how this will be resolved and whether you think this is a sustainable language/framework/ecosystem that would be worth adopting for a product, or learning and teaching?