Elswhere we have presented a traveling-wave analysis of HIV transmission on a tightly self-interactive, geographically-focused core group social network (Wallace R. Soc. Sci. Med. 32, 847, 1991; Soc. Sci. Med. 33, 1155, 1991; Environ. Plan. A. 26, 767, 1994; Wallace R. and Fullilove M. Environ. Plan. A. 23, 1701 1991). Here we reanalyze the problem in probability space and recover a close analog of the Shannon Coding Theorem of information theory. Subsequent direct application of information-theoretic methods provides striking insight regarding the spread of disease along the sociogeographic networks of marginalized subgroups, suggesting that 'risk behaviors' for infection may constitute essential components of a behavioral code for the transmission of information within the noisy channel of a marginalized community's social networks. The code's form, including the incorporation of risk behaviors, arises as a direct consequence of the external oppressive forces which structure marginalization. This viewpoint suggests an explanation of the sometime-observed rapid transmission of both infection and of control strategies for infection along the same network, but suggests further that if risk behaviors are indeed parts of a behavioral code for the transmission of group norms, statements of individual worth or resource sharing, then high rates of relapse are inevitable, given the persistence of the external oppression which gives those behaviors symbolic value. We suggest that violent acts in particular may emerge as key behavioral symbols for 'sending a message' in socially disorganized communities, implying that school-based or other individual-oriented harm reduction strategies for violence prevention, in the absence of a comprehensive, multifactorial reform program, cannot significantly reverse the effects of continuing economic and social constraints or of public policies of planned shrinkage and benign neglect, factors primarily responsible for the disorganization of urban minority communities within the United States.
