Wicked Problems and Other Problems -- Outline Note
--
Copyright ( c ) 2003 by Ken Friedman.
All rights reserved.
This text may be quoted and printed freely
with proper acknowledgment.
--
35 Wicked problems and other categories of problems
The issue of wicked problems has come up on DRS and PhD-Design for
several years. Wicked problems are an important category construct in
design. The nature of wicked problems has not been understood
adequately by many of the writers who use the term.
Wicked problems are one class among several classes of problems that
are important in design practice, design research, and design theory.
The major mistake in many articles is the failure to understand that
many design problems are NOT wicked problems. Of all possible design
problems, relatively few are wicked problems.
Many wicked problems are design problems. This is particularly true
of designers who work as policy planners, managers, legislators, and
other actors whose job is to "[devise] courses of action aimed at
changing existing situations into preferred ones" (Simon 1982: 129,
1998: 112) in organizations and large-scale social context.
There are many classes of design problems. While many wicked problems
are design problems, it does not follow that ALL design problems are
wicked problems. Design problems range on a continuum from axiomatic
and algorithmic problems at the most highly structured end of the
continuum to wicked problems and recursively linked wicked problems
on the other. Several other classes of problems are found between the
two extremes. Each of these classes of problems is effectively
infinite.
Since new problems are always emerging within any class of problems,
each class is effectively infinite in size. It is difficult to
compare infinities. Even so, describing the different classes of
problems carefully demonstrates that some classes of problems are
significantly larger than other classes of problems.
Of all design problems in any field or subfield of design, wicked
problems constitute the smallest class. Much design work is routine
work. This is why it is easy to delegate so much studio production
work to junior designers and inexperienced members of a design team.
Wicked problems are not the largest category of design problems.
Nevertheless, wicked problems are among the most visible problems.
This is because wicked problems tend to stand out against the
background of other, less visible problems. This is a natural outcome
of the fact that wicked problems generally involve many stakeholders.
In contrast, large classes of problems of the kind known as tame
problems generally involve relatively restricted issues. Moreover,
these problems often affect a small group of stakeholders or a single
client. Some tame problems are so restricted in scope and so simple
in scale that any solution satisfies nearly everyone involved.
In contrast, many more people have an interest in any wicked problem
than have an interest in any tame problem. The wickedness of wicked
problems is partially defined by the competition of interests among
stakeholders whose interests conflict with the interests of other
stakeholders in the same problem.
Problems involving public policy, ethics, legislation, economics,
law, medicine, the environment, and other highly visible problems
tend to be wicked problems. They stand out precisely because they are
hard to solve. Moreover, such problems affect many human beings,
including large aggregations of individuals, families, communities,
and entire regions or nations as well as groups, agencies, and
organizations of all kinds.
Relatively few designers work with these kinds of large-scale wicked
problems, though many designers function in situations where their
responsibilities may be influenced by wicked problems.
Horst Rittel developed and labeled the concept of the wicked problem
in the 1960s. The classic article in the literature points is a paper
by Rittel and Melvin Webber that appeared in Policy Sciences (Horst
and Webber 1973). This article has been influential in sectors of the
design field ranging from urban planning and policy studies to
information science, systems thinking, knowledge management, or
environmental studies.
In our part of the design field, Dick Buchanan (1992, 1995) wrote the
seminal introduction to the topic of wicked problems. Buchanan's
approach was deep and serious. The difficulty with our literature
since then has been the fact that few of the graphic signers or
industrial designers to write on wicked problems has examined the
literature on which Buchanan based his discussion, let alone
considered the important advances in the decade since Buchanan wrote
his contribution.
The failure to look beyond that one article is reflected in three
common approaches to the concept.
One approach offers a brief and unsatisfactory definition of wicked
problems followed by the assertion that design problems are wicked
problems.
A second approach does not bother to define wicked problems at all.
This approach offers the statement that there is such a thing as a
wicked problem and goes on to claim that design problems are wicked
problems. These authors seem to believe naming something is the same
as understanding it.
The third approach traces the concept of wicked problems in design to
Dick Buchanan with a "what he said!" style of citation. (Some authors
cite Rittel THROUGH Buchanan without bothering to read Rittel
himself.)
Design research scholars in other fields have followed Rittel's work.
While some of these writers work in several sectors, including
industrial design or graphic design, their writings on this subject
have not generally appeared in the journals and proceedings of these
sectors.
As a result, Buchanan's article remains the best piece yet written in
our sectors of the larger design field. Few authors in our sectors of
design have bothered to build on Buchanan or to address central
themes that have been developed in other fields. The most important
of these themes involves work on how to deal with and address wicked
problems.
Instead, those who write on wicked problems from a background in
graphic or industrial design (and from the arts and crafts sectors of
design) tend to offer a common argument.
The general kind of argument opens by stating three premises.
These are:
1) There is a class of problem known as a wicked problem.
2) Wicked problems are intractable by any reasonable or rational technique.
3) Design problems are wicked problems.
From these premises, the authors generally adduce two or three conclusions.
1) To solve design problems, designers cannot rely on reason, logic,
or any standard problem-solving method.
2) Since designers solve wicked problems without relying on known
methods, they must rely on some special method known to designers and
developed in design practice.
3) If this is how designers solve problems, any form of research or
theory other than this special system of designerly work is of little
or no use to the design field.
Of course, this is a simplification. The real argument as stated or
published takes many forms. In this short outline, I will not address
or develop the arguments, and I will not take them up to clarify or
challenge them.
These invalid arguments should not be confused with valid statements
using related or similar terms. For example, it is sometimes claimed
that the tacit knowledge developed experientially in design practice
is the designerly solution to wicked problems. This statement is
problematic. It is true, however, that designers and professional
practitioners of all kinds rely on tacit knowledge. This knowledge is
special to and embedded within practice. Tacit knowledge serves
practitioners in important ways. In an extended note, I will discuss
some of the valid arguments that have been inappropriately used to
support the invalid premises and conclusions stated here concerning
wicked problems.
In this outline, I will simply say that much of the writing on this
topic in design lists and design journals has been problematic.
Let us briefly consider the general kinds of problems and their
attributes. Since this is an outline rather than taxonomy, I will
state that what follows is not entirely developed. Even though I have
been struggling with the taxonomy of problems for some time, I have
not yet achieved a taxonomy that satisfies me.
Even so, it is possible to indicate the kinds of problems that designers meet.
Several categories of design problems are solvable, at least in
theory. Many are difficult to solve in practice, but this has to do
more with contingencies than with the nature of the classes of
problems. These classes of problems are sometimes labeled tame
problems.
One class of tame problems is the class of axiomatic problems that
can be solved with logical deduction from axioms and postulates.
Euclidean geometry involves axiomatic problems. While these problems
are logical-deductive in nature, many are far from trivial. Einstein
was convinced of the beauty and importance of classical geometry and
his own work made great use of later, non-Euclidean geometries.
At the same time, some theoretically tame problems resist easy
solution. Unsolved axiomatic problems have puzzled mathematicians for
years, and some have gone unsolved for centuries.
Another class of problems involves algorithmic problems. These
problems can be solved using clear systems of responses. Many forms
of algebraic problems are algorithmic.
It should be noted that many such problems are simple at one order of
complexity and difficult at higher levels of complexity that
introduce the need for large-scale computation. The computations
rather than the problems themselves are the challenge.
Some of these problems have surprising and significant importance in
design. One such problem is the traveling salesman problem. This is a
problem of selecting the most efficient route between numbers of
points in a network. This is a central problem in many areas of
logistics, network planning, systems planning, and transportation
scheduling. Every advance in the algorithmic management of the
traveling salesman problem has brought about significant gains with
reduced costs and grater productivity across all of the many
industrial sectors that involve logical analysis.
A massive class of problems permits computational solution. These
problems are genuine and often significant, but they can be solved
using computers or computer-assisted interfaces. Building the program
is the designer's work. Once the program is built, the system itself
can often solve hundreds, even thousands of problems.
Many expert systems now manufacture artifacts to customer
specification with little or no intervention by a human agent. The
customer develops a problem statement. The problem statement elicits
a range of prompts that narrows the range of solutions until the
customer's problem yields a specific solution. Some of these systems
use combinatory tactics in ways that permit a single system to create
millions of semi-customized, articulated solutions.
Today's computer integrated manufacturing environments now generate
billions of items every years manufactured to millions of patterns
created -- designed -- by such systems.
At some point in the theory thread, the claim was made that
logical-deductive systems cannot solve design problems. These three
classes of design problems and their solutions demonstrate that this
is not so. Many of these are small, tractable, problems. Aggregated,
they constitute three huge classes of the design problems that we
solve every day.
At the next level of complexity, we see problems that yield solutions
to judgment-driven algorithms. A corporate design program is a case
in pint, as is a planning code or a building code. Human agents must
decide among solutions in elation to specific criteria, but these
decisions are judgmental decisions within a limited range of
possibilities. Here, too, we see one of the great classes of design
work. Most design studios (including architecture studios and
especially graphic design studios) generate an enormous amount of
their work based on hourly fees charged for the application of
judgment to essentially algorithmic problems.
At a slightly higher level of complexity, we fine well-defined
problems. These problems begin to require expertise. The problem is
well defined, but it may yield several possible solutions. Somewhere
at the border of the well-defined problem, we find the distinction
between the apprentice or junior designer and the journeyman or
senior designer.
At the next level up, we find well-understood problems. These kinds
of problems involve characteristics and issues that we understand
well, but they also involve issues or aspects that may not be well
defined, particularly not is a specific situation. Well-understood
problems begin to be difficult. Even though we understand the nature
of the problem, we may not immediately see the solution, not even the
range of solutions.
Next, we see poorly defined problems. Here we move into genuinely
difficult territory. Heuristics, iteration, and different forms of
pluralistic tactics enable us to solve poorly defined problems by
restating and restating them. The most common range of solutions to a
poorly define problem is to transform the poorly defined problem into
a well-understood or well-defined problem. Here, poorly developed
formal language represents the true difficulty. The expertise of the
designer comes in redefining the problem in a way that renders it
tractable.
Finally, we come to wicked problems. Wicked problems offer a number
pf distinct challenges that are not seen in other classes of
problems. Nevertheless, the very nature of a wicked problem opens up
areas of solution within the large, difficult structure of problem
with multiple variables, constraints, stakeholders, and possible
iterative changes. Stating the nature of wicked problems and pointing
to some of the kinds of solutions is a task I will pursue in the
larger note to follow.
My purpose here has been to address a theme that has come up several
times in this thread. I want to argue that the term wicked problems
has been used in an inappropriate way to define all design problems.
Instead, wicked problems are a specific class of design problem. Of
all design problems that we solve in the working world, wicked
problems constitute the smallest of a series of effectively infinite
classes.
It is the nature of wicked problems to be more visible than other
problems. This makes wicked problems SEEM more pervasive than other
kinds of problems. In terms of genuine pervasive spread, consider
this instead. One company regularly solves a kind problem on one
product that permits a computer-integrated manufacturing system to
generate 2.9 million solutions covering a vast range of customer
needs. The solution takes the form of artifacts that are manufactured
in millions of units. The system that solves these problems is
invisible to he end-users of the manufactured artifacts.
It is worth noting that logical-deductive systems also serve a role
in solving wicker problems. Logic and deduction enable us to rule out
classes of solutions that would not be satisfactory.
A second, systemic implication of struggling with wicked problems is
that we slowly move certain kinds and classes of problems from the
wicked to the poorly defined, from the poorly defined to the well
understood, and from the well understood to the well defined. Kinds
of problems that once seemed to be wicked problems have even been
reduced to algorithmic problems.
It is only through struggling conscientiously with wicked problems
that we begin to recognize the qualities and characteristics that
render some kinds of problems intractable while open a solution space
that allows us to address other problems fruitfully.
One more class of wicked problems is particularly difficult. These
are recursively linked wicked problems of the kind we meet in
geopolitics, diplomacy, and international organizations. There are
other ranges of recursively linked wicked problems. One reference
book catalogues and describes such problems and even links them. This
is the Encyclopedia of World Problems (Union of International
Associations 1994).
Of the many designers I know, only a few dozen operate on problems of
this sort.
There ARE wicked problems that will remain impervious to general
solutions for many reasons. A range of very clear characteristics
identifies these classes of problems, and they embody attributes that
make them theoretically unavailable to general solutions. One must be
skilled at analysis and logical deduction to understand why these
problems are wicked. Simply clapping hands and shouting "Wicked
problem!" is not good enough.
Laurence Peter, the author whose name was given to the Peter
Principle, used to speak of problems that are so complex that we must
be highly intelligent and well informed simply to be undecided about
them.
That is how it is with wicked problems. Simply knowing that there is
such as thing as a wicked problem tells us nothing about the nature
of wicked problems. It tells us nothing about their properties and
attributes, or their theoretical and practical importance.
I will return to this issue to examine the nature of wicked problems
and to discuss the kinds of progress we have made in addressing
wicked problems over the past ten years.
-- Ken Friedman
References
Buchanan, Richard. 1992. "Wicked Problems in Design Thinking." Design
Issues, Vol. 8, No. 2, 5-21.
Buchanan, Richard. 1995. "Wicked Problems in Design Thinking." The
Idea of Design. A Design Issues Reader. Cambridge, Massachusetts: MIT
Press, 3-20.
Rittel, Horst, and Melvin Webber. 1973 "Dilemmas in a General Theory
of Planning." Policy Sciences, Vol. 4, 155-169.
Simon, Herbert. 1982. The Sciences of the Artificial. Second ed.
Cambridge, Massachusetts: MIT Press.
Simon, Herbert. 1998. The Sciences of the Artificial. Third ed.
Cambridge, Massachusetts: MIT Press.
Union of International Associations, editors. 1994. Encyclopedia of
World Problems and Human Potential. Munich: K G Saur.
--
Ken Friedman, Ph.D.
Associate Professor of Leadership and Strategic Design
Department of Leadership and Organization
Norwegian School of Management
Visiting Professor
Advanced Research Institute
School of Art and Design
Staffordshire University
|