The article mentions two paradoxes that are worth considering: > How could a great mathematician not want to study these things? [these things = Lie groups, Ri

The article mentions two paradoxes that are worth considering: > How could a great mathematician not want to study these things? [these things = Lie groups, Riemannian manifolds, algebraic geometry, algebraic topology, global analysis, or the deep ocean of mathematics connected with quantum mechanics and relativity theory] > > This suggests the fundamental question: How much, or how little, must one know in order to do great mathematics? > [...] > The second Erdos paradox is that his methods and results, > considered marginal in the twentieth century, have become > central in twenty-first century mathematics. I never had the feeling that his results were 'marginal', but the fact that he never got a full position anywhere got me thinking---maybe Erdos was just interested in these positions (which, given his personality, might be likely), or maybe he did not 'sell' his work well enough. As someone who hates advertising their own work, I can see how tragic this would be.