Asymmetric Topology of Distributed Problem Solving Networks

Distributed problem solving, which often involves an intricate network of interconnected tasks carried out by hundreds of actors, is fundamental to the creation of manmade systems 1 as well as the organization of work in biological systems 2. Here we analyze, for the first time, the statistical properties of four human large-scale distributed task networks. We find that the distribution of outgoing communication links is scale-free 3-6 (power law decay) with or without a cutoff 7-10. The distribution of incoming information flows always has a cutoff, and when both distributions have cutoffs the incoming distribution has a cutoff that is lower by more than a factor of two. The functional significance of this asymmetric topology can be explained by considering the dynamical interactions that take place in distributed problem solving. The study of many 'real-world' social, biological and technological networks 11 have been shown to be neither completely regular nor completely random. Instead, these networks are typified as " small-world " networks 11, 12 , combining the large degree local clustering of connections characteristic of regular networks with the small average path length of random graphs. Empirical work shows that the total node degree distribution of