This paper reports an experiment that demonstrates the superiority of groups over the best individuals using various instructions for strategies to solve letters-to-numbers problems. Simulations provide baseline statistics to compare performances. Letters-to-numbers problems require identification of the random coding of the ten letters A, B, C, D, E, F, G, H, I, J to the ten numbers 0, 1, 2, 3, 4, 5, 6, 7, 8, 9 by (a) proposing an expression in letters ( A + B ); (b) receiving the value of the expression in letters ( A + B = DG); (c) proposing codings of the numbers for the letters; and (d) receiving feedback on the correctness of the proposed codings, on each trial. Three-person groups and individuals solved the problems under five instruction conditions to use different strategies. As predicted, the groups generally had fewer trials to solution than the best of an equivalent number of individuals, and the best performance was achieved with the most information-rich instruction, to add all ten letters.