On one aspect of science policy based on an uncertain model

We discuss one aspect of the allocation of new scientists to teaching and r esearch careers. In the past, this allocation problem was treated on the ba sis of a model with known parameters and as a classical optimal control pro blem with the allocation proportion as the sole control. The utility functi on took both short- and long-term goals into account. Here, we allow for un certainty in the possibly time-varying system parameters. We account for th e possibility of new scientists choosing careers other than teaching and re search. Finally, we treat the allocation problem not as an open-loop optima l control one, but rather as one of robust control, insensitive to the syst em uncertainties, in order to assure the desired numbers of teachers and sc ientists within a computable horizon.