A detailed understanding of how network topology constrains network dynamics is one of the fundamental questions in biology. We study the space of boolean dynamics over all small (2 and 3 node) networks. Our work suggests that network motifs are not optimised to perform a single dynamical task, but that some motifs are recurrent because, by a change of rules, they possess a broad range of functionality. We also study the simplest class of network topologies: networks in which each node has exactly one input. We give the exact solution for the number and size of attractors on a single loop and on multiple loops. For a critical K=1 Kauffman model, we show that the minimum number of attractors scales as the number of states of the nodes in loops, considerably faster than was previously believed.