That Nobel was won in 1978, and is based on the fact that in practice we can't be rational. The NP demonstrations that, in theory, updating a Bayesian is a com

That Nobel was won in 1978, and is based on the fact that in practice we can't be rational. The NP demonstrations that, in theory, updating a Bayesian is a computationally infeasible problem was G. F. Cooper in 1990 (for Bayesian Networks). The stronger result that approximating the update is also computationally infeasible was Dagum, P. & Luby, M., 1993. So Simon's work relates to what I said, but isn't based on it.