Ygge and Akkermans: Background

It is a very natural habit of mankind to search for easy answers to hard problems. However, such discoveries are sufficiently rare to be on guard whenever a new one appears, and yet it is all too easy to allow the power of apparent convenience lure a sound mind from careful inspection of the solution. Thus, when Huberman and Clearwater [2, 6] discovered a market-based method for controlling a building environment that considerably outperformed a mainstream controller, few questioned the miraculous improvement and fewer yet wondered precisely which of the techniques used by the two authors effected the improvement. To step back for a moment, it seems a matter of common sense that any centralized approach, given sufficient information and computation time, should produce a result at least as good as any decentralized approach. Thus, the fact that the Huberman and Clearwater’s market-based controller produced a better solution than a centralized controller should have been suspicious. Furthermore, there was little attempt on the part of the authors to explain their results beyond attributing it to the powers of market-based problem solving techniques. Interestingly, despite their use of the techniques from the field of Economics to solve a resource allocation problem, many of the critical details, such as utility and bid formation, were arbitrary enough to preclude any useful analyses using well-known tools from the latter field.