Calculating risk and payoff in planetary exploration and life detection missions

Abstract A framework for quantitative assessment of different mission architectures is described using historical data and formal (Bayesian) information value measures. The science value of the result is argued for binary questions (e.g. 'is there life on Europa?) to be proportional to the logarithm of the posterior likelihood ratio of the answers, and can be derived from estimates of the false positive rates of instrumentation and of the presence (P D ) of biosignatures at a given site. The expectation payoff is the product of the sought result with Markovian success probabilities of the required steps of launch, landing, sample acquisition etc., and historical planetary mission data are reviewed to derive (sometimes dismaying) estimates of these probabilities, e.g. historical landing successes rates are of the order of 66% and when landing is successful, the conditional rates of individual sample acquisition/analysis/return have similar values. The history of seafloor exploration on Earth is used as an analog, and indicates that in the absence of close reconnaissance data, P D may have rather low values of the order of 1% or less. The data acquisition success framework is demonstrated on the value of single versus multiple landers, on the choice of flyby altitudes for multiple plume fly-through missions, and on the value of surface mobility, which for small values of P D multiplies the science return by the number of sites visited. Bayesian reasoning requires encapsulation of prior information: while such estimates (of biosignature presence, false alarm rates, etc.) are inevitably subjective, the decomposition of that information onto specific factors affords transparency into their contribution to the final result and provides a basis for rational mission evaluation.