What kind of theory sets itself the goal of maximizing a variable that it takes to be unbounded above?!
Clearly this does not make much sense. The resolution of this blunder is rather simple. Insofar as modeling the uncertainty in Info-Gap’s uncertainty model is concerned, the value of α is unbounded, and can be so till kingdom come! But the point is that to determine robustness, the value of α is implicitly bounded by the performance constraint r(d,u) ≥ r*,∀u∈U(α,û). So α is bounded whenever robustness is bounded and it does have a worst case after all.
