Requisite variety of design

Posted on by Brandon Klein


The ability to control a situation is directly proportional to the ability to process information about it. In computer science, the “law of requisite variety” (a paraphrase of the above) describes some of the operating characteristics of effective systems. It also describes the characteristics of organizations facing massive change. Standard approaches to managing large projects, however, break complex issues into bite-sized (that is, “team-sized”) chunks. As the complexity of the objective situation is arbitrarily reduced to make problems more manageable, what actually happens is that the organization’s ability to find the appropriate solution is diminished. Separating a tapestry into its component threads means you can no longer see the tapestry.

We are trained to think more as analysts than as designers. Analysis takes a complex situation and seeks to tease out what is going on. This is a necessary – but not sufficient – condition for developing a solution. It is hard to put Humpty-Dumpty together again, but that is what designing a solution is all about.

The variety of the solution must match (or exceed) the variety of the problem.

We often mis-identify our problem, addressing symptoms rather than actual problems. Problems seem to be impenetrably complex. We (speaking broadly) don’t understand the nature of our problems or understand their complexity. We don’t (speaking broadly) know how to address our complex problems, in that we don’t know what we don’t know.


The process of design must be robust and complex enough to match the complexity of the environment in which the solution must operate. The variety of the solution seeking process must match (or exceed) the latent variety of the problem to be solved.

The solution itself may be a paragon of simplicity, but it is likely to be effective only if in the process of development it takes into account all the forces with which it must contend.

In designing, explore all aspects of the situation before moving to closure. Include insights from many different vantage points. Test a wide range of alternatives in a variety of conditions. Design works best when it is iterative with the solution.

Utilize a methodology which has sufficient variety that it enables correct

  • Identification of problems
  • Identification of potential solutions

Design works when:

  • It is iterative, with successive approximations
  • Its intermediate results are subject to rigorous testing
  • It alternates between divergence and convergence
  • It integrates and creates new knowledge, ultimately changing how we understand ourselves and our problems.
  • It questions itself
  • It expands the field/domain of “reality”
  • It recontextualizes issues
  • It changes how we understand problems

What the ASE does is generally “forced questioning,” where the facilitator picks the question. That means it matters what the questions are—we have to pick the questions carefully.

See Lateral Thinking, by Edward DeBono. (Especially Chapter 22 “Description/Problem Solving/Design.”) “Lateral thinking is especially useful in problem solving and in the generation of new ideas. But it is not confined to these situations for it is an essential part of all thinking. Without a method for changing concepts and bringing them up to date one is liable to be trapped by concepts which are more harmful than useful. Moreover rigid concept patterns can actually create a great number of problems. Such problems are particularly fierce since they cannot be altered by available evidence but only by insight restructuring . . . Lateral thinking particularly emphasizes the dangers of rigid patterns which the mind is so apt to construct because of the way it handles information. –Edward de Bono, Lateral Thinking

This notion of ‘rigid concept patterns’ is really important. The way our brains function is to create and use patterns (or models). This is both an important enabler and a thinking trap. For example, when designing vehicles that have to work underwater, engineers create manipulator arms. These have only a limited reach and maneuverability. When rethinking the design of a vehicle that had to excavate a shipwreck 8000 feet deep in the Atlantic, engineer Tommy Thompson rethought the whole idea of a robot arm. “Most manipulators were anthropomorphic, modeled after the human arm, but with only five to seven degrees of freedom. The human arm has twenty-seven. But even the human arm has only two segments, upper and lower, plus an ‘end effector,’ or hand. Tommy wanted three upper arms and five lower arms, and at least twenty-seven degrees of freedom, and he wanted to do things the human arm could never do. He wanted to extend the reach by telescoping the shoulder, the equivalent of changing the TV channel without a remote. Just shoot the whole arm across the room . . . He wanted to pop the eyeballs out with the arm, zoom in with the cameras, and backlight it so the scene was crisp.” --Gary Kinder, Ship of Gold in the Deep Blue Sea

See also Mars Pathfinder by Price Pritchett

Relates To

  • There is no “out there.”
  • No one is more limited than the cognoscenti