The systematic study of non-perturbative (and other) effects in string theory often leads one to the study of large systems of diophantine equations. I will give examples, relate these issues to Hilbert's 10th problem, and discuss implications for computability of the landscape.