Commonly, participants of benchmark studies in control are supplied with knowledge of the true system to be controlled. Exact knowledge of the plant, however, tends to influence the choice of tuning variables and discourages legitimate use of estimated or guessed information. This paper reports an alternative benchmark, in which participants did not know the true system, which was supplied in the form of scrambled simulation code. The paper describes the problem statement, reveals the true system and surveys the results obtained by nine different research teams. Their approaches include indirect and direct adaptive control, model reference adaptive control, adaptive predictive control, H(infinity) robust control and robust stability degree assignment. The results display the trade-offs inherent in the different approaches and highlight the interplay between robust and adaptive control.