The problem domain you're working in is also important. For example, if you're doing heavy duty number crunching for finance or image processing, Python has Nu

The problem domain you're working in is also important. For example, if you're doing heavy duty number crunching for finance or image processing, Python has NumPy, for which AFAIK there is no equivalent for Ruby. Then again at very large scale you don't want to restrict yourself to one language in which case you can lash together various components using a distributed job scheduler like Gearman (http://gearman.org/ http://gearman.org/).