I'm not talking about positive externalities, I'm talking about actual work being done. Some structures-- early-stage startup companies generally included-- tend to involve disproportionately large amounts of work relative to others, which is made up for by the potential for very high, if delayed, payoffs.

The "revealed preference of the economy" is difficult to actually determine in such cases given the uncertainties involved, though it could likely be estimated using reference class forecasting and similar predictive techniques.