Risk Analysis Under Adversarial Conditions
James Blodgett is giving a presentation this week at the Second World Congress on Risk. This is the abstract.
Risk Analysis Under Adversarial Conditions
Risk analysis can develop adversarial elements. Those proposing an activity have an interest in showing it to be safe. Those exposed to risks have an interest in finding those risks. Some risk analysis tools fail under adversarial conditions, since estimates of risk have subjective elements about which adversaries can disagree. The precautionary principle addresses this issue, but has philosophical and practical difficulties. A recent case study highlights the difficulties. Particle colliders were thought to be very safe, but several safety factors eroded. Nevertheless, application of best practices risk methodologies was strongly resisted. Proposed solutions include special courts (proposed by Judge Richard Posner,) and red team/blue team methodology (proposed by Sir Martin Rees and Francesco Calogero.)
——–
I am adding another solution, Carolyn Raffensperger’s court-appointed advocates for the future. I will have handout material on several aspects. I would like to see development of even better solutions and even better protocols.
If anyone has ideas for improving these protocols, I would be interested. Consider how protocols would apply to issues other than the one being debated on this website. For example, suppose the nanotech folks propose self-replicating nanobots, or the AI folks propose a computer that transcends human intelligence, or (on a more mundane level) a liquid natural gas terminal is proposed for Boston harbor (to receive shiploads of liquid natural gas–think megatons.) How should we vet these things? How do we balance the risks and the benefits? How do we quantify risks if advocates and opponents present reasons to estimate them as quite different numbers?