This is an interesting topic and has many analogies. If you gave mathematicians some general exam and have all universities participate, I would also not expec

This is an interesting topic and has many analogies. If you gave mathematicians some general exam and have all universities participate, I would also not expect top universities to score the most. However, if you asked instead what the money value is of their research, then MIT, Oxford, et al. would score highly. I don't want to extrapolate on the economist.com article, and whether it's true or not, but rather just illustrate that this happens in other contexts. Another example is salaries. Are the best programmers in the areas that pay the most? I think no, but the ones whose work make the most money for their employers probably are. A case in point is the alpine Emacs coder that featured in a comment a few days ago, Thierry. The main issue is that if you focus on the money part, then you focus on ideas that have settled. Many small places or small universities are still at the distillation step and the distillate tends to move to where the money is. I use this argument as one of the main arguments for fundamental research, and also for why it has cognitive dissonance with earning potential.