May 6, 2007

The No Free Lunch Theorem

"Anyone who markets a heuristic as universally good may be full of more than just good intentions", says Scott E. Page, author of The Difference. Before coming to this statement, the author explains: "Given that there are so many types of heuristics, a natural question to ask is which heuristic works best. It has been shown that comparing heuristics across all problems is a fool's game. No heuristic performs better than any other across all possible problems. This result is known as the No Free Lunch Theorem. We can read this theorem in two ways. It means that for any given problem there will be good heuristics and bad heuristics. It also means that for any heuristic there will be problems on which it performs well and problems on which it fails miserably." .... "As powerful as this heuristic may be in some contexts, the claim that this heuristic works everywhere bumps against the logic of the No Free Lunch Theorem.
What does this have to do with the solution-focused approach? Well, as you may have seen, I have been writing enthusiastically about the solution-focused approach and I have been trying to present it in the form of a heuristic, or an algorithm, a few times. See for instance this and this. No matter how enthusiastic I am and how useful 'the' solution focused heuristic may be, I am nevertheless convinced that it has its limitations, like every heuristic. No matter how great the approach may be, it would be dangerous to think that it is always the best approach to follow, or even, that it should always be followed. What I think is:
  • In some cases SF will be superior to other approaches
  • In some cases SF will be roughly equally effective as other approaches
  • In some cases SF will be inferior to other approaches
  • In some cases a combination of SF and other approaches may lead to best results
We don't know where the limits of the applicability of the solution-focused model are. My hunch is that the more complex a system is (complex as in complex adaptive systems), the more likely it is that the solution-focused model may be superior. That is just a hunch, however. (More hunches are welcome). What does all of this means for our practice? First of all, it means we have to remain modest at all times. We have to keep an open and exploring attitude and not judge other for preferring different approaches. We have to learn more about SF. Second, it may be necessary to develop different versions of the solution-focused model for different contexts (therapy, coaching, management, training, teamwork, ....). Third, we should not pretend to know exactly what the solution-focused approach is.


  1. Coert,

    I think it's good that you often point out that SF is not for everything. This will keep you open to signs that you should not be using SF in a particular situation and others who read it can keep from being "solution forced."

  2. Research on the effectiveness of different kinds of psychotherapy reveals the same thing, and a couple of theories have sprung up that basically say it isn't about the theory, it's about the client.

    It's also important to have a breadth and depth of knowledge regarding other heuristics and algorithms, to help the professional match the customer. The great thing about SF is that used as an approach, it can piggyback on other ways of working.


Enter your email address:

Delivered by FeedBurner