The "first generation" of research studies on the prediction of violent
behavior found such prediction to be highly inaccurate. Many social policy
changes were implemented or recommended on the basis of that research. More
recently, a second generation of research and theory on violence prediction
has begun to develop that emphasizes the limitations of the existing body
of research, points to possible improvements in predictive technology, and
evaluates public policies involving violence prediction only in the context
of the feasible alternatives to those policies.