The Ways of Wickedness:
Analyzing Messiness with Messy Tools
Bryan G. Norton
School of Public Policy, Georgia Institute of Technology

Epigram: Russell Ackoff (1979): “Managers are not confronted with problems that are independent of each other, but with dynamic situations that consist of complex systems of changing problems that interact with each other. I call such situations messes. Problems are extracted from messes by analysis. Managers do not solve problems, they manage messes.” (quoted in Meadows, 2008, p 1.)
Introduction.
The revelatory paper, “Dilemmas in the General Theory of Planning,” by Rittel and Webber (1973) has had great impact because it provides one, particularly potent, explanation of an emergent consensus—or almost-consensus—across many disciplines.
While different disciplines use their own vocabulary to articulate the result, Rittel and Webber drove home a very important point that gains its generality by its simplicity: Many “problems”, as addressed in real-world situations, involve elements that exceed the complexity of any known or hoped-for model. No descriptive disciplinary perspective or expert system can embody all of the variables and data necessary to understand, predict, and control the functioning of the dynamic systems within which humans struggle with complex problems and conflicts. This insight is forcing a revolution—or at least evolution—in the way we think about social sciences and policy. The changes all reflect deeper philosophical forces associated with the gradual and ongoing eclipse of logical positivism and its view of science.

The identification of “wicked problems” and the commentary by Rittel and Webber should not be considered entirely original and unexpected, despite its arresting articulation. Recognition of the limitations of comprehensive modeling and decision making systems had come under pressure from a number of arguments and results that emerged throughout the 20th Century. Logicians began to see paradoxes and difficulties when they began to systematize logic and mathematics, as paradoxes arose in attempts to formalize set theory and the foundations of mathematics. In 1931, Kurt Gödel published his incompleteness results, which rigorously proved that for any body of knowledge including arithmetic, and any formalized system designed to axiomatize it, there will be truths of that body that cannot be proved within that system. While highly technical in the proof, this result could be interpreted to have broad implications. For example, Rucker (Rucker, 1982) says of Gödel’s work: “Although this theorem can be stated and proved in a rigorously mathematical way, what it seems to say is that rational thought can never penetrate to the final ultimate truth ...” Other logicians, including Gödel himself, have denied such broad applications, and nobody to my knowledge has succeeded in providing a rigorous argument from Gödel’s result to the conclusion that there can be no comprehensive formalization of a complex problem (Barrow, 2004). Gödel’s results have dominated logical thought since published, however, and while perhaps too technical to yield a clear application to knowledge systems as a whole, the result encourages a skeptical attitude toward attempts at complete and holistic explanations of complex phenomena. This skeptical attitude blended with other powerful arguments that called into question the scientistic worldview of logical empiricists, who struggled to create axiomatic systems that would be comprehensive, general in application, and able to generate answers that are both correct and certain. Attacks on logical positivism from a number of sources,including W.V.O. Quine’s attack on empirical reductionism (1953), Rudolf Carnap’s recognition that there are questions that must be asked “external” to any system of linguistic rules and axioms (1956), and Alfred Tarski’s (1944) semantic treatment of “truth” as a predicate of sentences all reinforced the idea that human systems of knowledge cannot be captured in one, complete system of analysis. Thus, while it has proven difficult to articulate rigorous arguments to show the impossibility of comprehensive accounts of complex systems as applied to ordinary language and ordinary discourse, a number of arguments showing that impossibility in formalized systems has created a presumption against comprehensive modeling of natural systems.

While there has been resistance to these ideas from practitioners of disciplines which seek to formalize decision making processes, their consequences are in one way or another, being worked out in disciplines as diverse as philosophy, economics, engineering, artificial intelligence, cognitive science and planning, each field of which has its own jargon and its own literature reporting the recognition. Even if one is skeptical of treating the formulation of Rittel and Webber as a loose corollary of Gödel’s theorem, the simplicity of the argument of Rittel and Webber, and the breathtaking generality of its application, makes the Rittel/Webber formulation arresting. The broad scope of what they called “planning theory”, and the growing complexity of environmental and other social problems, carries the reader inexorably to the conclusion that most problems that matter to society today are wicked problems or at least have wicked aspects.

Many who encounter this work for the first time find that their concept of wicked problems aptly describes many environmental disputes, disputes that seem intractable, engendering endless controversy. For those frustrated with the lack of progress in many areas of environmental protection, Rittel and Webber’s work suggested a powerful explanatory hypothesis: Complex environmental problems cannot be comprehended within any of the accepted disciplinary models available in the academy or in discourses on public interest and policy. This failure is not a matter of inadequate practice, but a matter of principle. Accepting this result implies that the formal analytic methods that are created as aids in decision making can be applied to real problems only after many difficult and contentious steps have already been taken; and these steps have each simplified the original reason for distress. Wicked problems always “spill out” over the edges of any system developed to comprehend their behavior or address them.

Rather than argue about whether this insight constitutes a consensus among theoreticians, I will accept the explanatory hypothesis and ask, “Then what?” What should we conclude about the future of social improvements, and about the possibilities for rational discourse leading to cooperative action, with respect to this huge number of pressing public, environmental problems? Are we left with the depressing conclusion that rational analysis of these problems cannot be comprehensive and coherent, that it is impossible even to formulate them clearly enough to allow rational public discourse? In this paper, I will accept the above hypothesis in all its explanatory generality and then examine the prospects for rational discourse and progress toward social improvement in a policy context shaped by—one might say “limited” by—a recognition that most environmental problems are wicked problems. What can, and what should, change in our discourse about environmental protection? Can we find ways to address environmental problems—granting they are wicked problems—that improves the ability of communities to respond creatively and rationally to them? I will offer the beginnings of an answer to this question in 4 parts. In Part 1 I will discuss in more detail the theoretical and practical consequences of declaring complex environmental problems to be wicked, noting the stringent limitations it places on the methodologies available to analyze and “solve” such problems. I will argue that, while the Rittel-Webber critique requires us to abandon many of the assumptions associated with a positivistic view of science and its applications to policy analysis, it also points a more productive direction for the future of policy analysis. In Part 2 I note that, while the consensus around the ideas of Rittel and Webber requires some reaction by all policy analysts, the response has differed significantly across fields and research traditions. I will explore the ten characteristics of wicked problems in Part 3, subdividing them and treating each as examples of four particular aspects of wickedness. I will then note that, in a process guided by heuristics, it may be possible to bring systematic thinking to bear upon aspects of wicked problems, even if they cannot be “solved” in the ways once hoped. Finally, in Part 4, I will introduce “boundary critique”, developed within Critical Systems Theory (CST), an approach that offers some reason for optimism in dealing with some aspects of wickedness.

Part 1: The Significance of the Rittel/Webber Distinction
I fully agree with the implication of this issue of the journal, that it is useful to understand most intractable environmental problems to be wicked problems. This practical value, taken together with the generality and power of the theoretical considerations behind the distinction only intensifies its importance, as I have argued elsewhere (Norton, 2005, esp. Section 4.1).

Rittel and Webber identified “wicked” problems as a special type of problem and they listed ten characteristics of them, eschewing a formal—or even informal—definition. Although they did not emphasize the point, I think these authors clearly recognized that, while the characteristics they list for wicked problems are quite disparate, the class of wicked problems are all expressions of diverse and conflicting values and interests, which cause individuals to view problems very differently. Different participants in public discourse, acting on very different interests and diverse values, will not only differ about the ends and the means toward social improvement, they will also differ regarding how to formulate, or “frame” what is the real problem to be addressed. When these views are sufficiently diverse, it is said that different discussants in public discourse have different perspectives. Benign problems, on the other hand, do not involve differences and deep disagreements about values, and hence answers obtained, once verified, are no longer contentious. For this reason I think that Rittel and Webber go a long way toward understanding why some public environmental problems are so intractable and perplexing. Conflicts over values, and disagreements caused by value conflicts, are often hidden in the different ways discussants formulate a problem. In a diverse community, with diverse values, it is not surprising to find different perspectives, which in some cases create miscommunication and misunderstanding about the problem itself.

A supplementary point is also well made by Donella Meadows, who notes that we usually, in contentious situations, define a problem as the lack of our favorite solution, rather than providing a thoughtful analysis of the values threatened and the reasons the values are under-produced or under-protected by the currently functioning system. Examples: “The problem is, we need to find more oil. The problem is we need to ban abortion.” She suggests: Listen to any discussion, in your family or a committee meeting at work or among the pundits in the media, and watch people leap to solutions in ‘predict, control, or impose your will’ mode, without having paid any attention to what the system is doing and why it’s doing it.” (Meadows, 2000, p xx) In such situations, no clear view of the “real problem” will emerge, and rational analysis, however sophisticated, cannot be brought to bear upon these situations.

Recognizing the importance of Rittel and Webber’s provocative conceptualization, I also endorse a slightly different version of the same insight by Ackoff, quoted above in the epigram to this paper. Ackoff says that managers face “messes” not problems and that we extract problems by “analysis” from messes. I see only a semantic difference here in comparison to Rittel and Webber’s distinction. If, however, we accept Ackoff’s version, but then ask: by what method of analysis do we extract problems from messes? Then we are in a quandary. As argued above, getting to a fruitful formulation of a problem involves simplifications—all disciplinary models are at best partial pictures of reality—but employing any disciplinary model of analysis will involve already making many assumptions associated with that model and its home field. So, using Ackoff’s phraseology and reasoning: we need a form of analysis that will get us from messes to problems, but the near-consensus expressed in “Dilemmas” seems to imply that no specific methodology can provide a unique analysis that will apply at the stage when a community or a society faces a mess. The profound point being made by Rittel and Webber, Ackoff, and others, can be stated briefly: With respect to wicked problems, we face an analytic void. In the process of taming a wicked problem, messes are simplified in ways that determine what discourse and what analytic tools will be used. Therefore, when we address wicked problems there will always be an uneliminable element that must be addressed outside any of the particular models used to understand it, and in this world of messes, special interests, politics, and power may skew the formation of problems before they reach the policy agenda.

Approaching the problem from the viewpoint of policy analysis and applied decision theory, it is possible to formulate the general result outlined above as follows: For any complex public environmental controversy of importance to diverse interests, it will not be possible to “model” that conflict in such a way as to provide an algorithmic solution to it. Another way to put this point is to say that, accepting the judgment that no pre-theoretical standpoint exists from which to conceptualize complex problems, all deep analysis must be contextual, rather than a-contextual.

A-contextual analyses of decisions:
A. evaluate decisions by evaluating expected outcomes, and

B. understand decision analysis as a search for algorithmically generated (computational) solutions identifying a single best answer.

A-contextual systems of analysis include:

•Rational Choice Models;
•Microeconomics; Cost Benefit Analysis as a decision tool;
•Optimizing programs in Operations Research
•Competitive Game Theory

All of these approaches, sharing the two characteristics, A and B, are "A-Contextual" in the specific sense that they analyze outcomes by measuring and aggregating some "objective" measure of behavior, such as utils or preferences (wtp), which are value measures that require—and carry with them—no information about the decision context. Advocates of contextual analyses, on the other hand, do not expect that unique solutions to problems can be generated by an algorithmic calculation; indeed, contextual analysts assume that contextual aspects of problems will affect the way a problem is formulated in specific, local contexts, and take this into account.

As a shorthand, I refer to analyses that aspire to embody these two characteristics (A and B, above), as “algorithmic” systems. I am suggesting that the consensus result recognized by Rittel and Webber can be generalized: No complex social problems—ones that involve multiple interests and values—can be comprehensively represented in a system that would allow the computation of a best solution. None of the much-touted formal models can provide a solution to a problem of conflicting values and wicked problems.

Prior to the recognition of the special characteristics of wicked problems, decision sciences generally operated on the assumption that, given enough data, any problem could in principle be given one unique solution. This assumption anchored a strong preference among analysts for acontextual analysis (Lasswell, 1970; Torgerson, 1983; Norton, 1990; 2005). Whether or not a systems analyst uses contextual analysis is closely tied to a choice of whether to envision the system models created as open or closed systems. Closed systems are assumed to contain all of the factors affecting behavior of that system. Open systems interact with an “outside”, meaning any complete analysis must extend beyond the system boundaries, which explains the “spillover” metaphor introduced above. In an open system, any “problem” will have effects beyond the system specified. Indeed, the decision to seek contextual analyses can be seen as a formalization of the choice to model processes as open systems. In self-organizing systems, the crucial aspect is process, and subsystems and systems survive not as eternal entities, but as temporary patterns that remain as long as systems dynamics support them. Contextualism in policy analysis has thus been associated with the central axioms of General Systems Theory and of the ecologists’ version of GST, hierarchy theory (HT).

The Rittel/Webber result implies that no complex social problems involving multiple interests and values can be comprehensively represented in a system that would allow the computation of a best solution. None of the much-touted formal models can provide a solution to a problem of conflicting values; this means that a number of approaches to decision making must be re-thought. Re-thinking these approaches necessarily involves looking into the “void” described above.

This re-thinking has occurred in lockstep with a re-thinking of the positivistic view of science as an objective description of the world, with reality being reported and hypotheses supported, or denied, by observation and experiment. Loss of faith in “objective” science during the 20th century has left the “describe-and-then-evaluate” model unable to take a first step, unable to describe phenomena without first choosing a vocabulary and linguistic rules; the objectivity that came from ignoring context and evaluative attitude collapsed. This process, by policy analysts, of flirting with, and then rejecting, positivism as the model for policy analysis and policy science is well-described by Torgerson (1985). Torgerson shows how the founder of policy sciences, Harold Lasswell (1970), who initially touted a “science of policy” that included a technocratic and overly scientized approach to policy, came to realize during his long career that the dream of a comprehensive and consistent analytic framework for fitting scientific data into policy analysis was an impossible dream.