I will describe van Hemmen and Palmer's old explanation of the failure of Sherrington and Kirkpatrick's original solution to their model. It involves indeterminacy of the classical moment problem. Then I will try to suggest that for the theoretical physicists they might be aided by knowing moments of the generalized Ulam problem*, which we have started to calculate. *That is the problem of determining how many increasing subsequences there are of a certain length in a random permutation.