The document discusses the complexity of computing Nash equilibria in anonymous games. It presents previous work showing approximation algorithms and hardness results for certain cases. A key result is that the document proves that computing a 1/exp(n)-approximate Nash equilibrium in anonymous games with 7 or more strategies is PPAD-complete by reducing from polymatrix games. The reduction works by embedding the payoffs of a polymatrix game in an anonymous game in a way that "breaks the symmetries" to relate players' mixed strategies to observable partition probabilities.