Word of the Day: Systems and Wicked Problems

Lots of wires, but ENIAC didn't replace GodWe all deal with systems—computers, cars, or communities, for example—and a few concepts may help see things a little more clearly.

This is an excerpt from another book of mine, Future Hype: The Myths of Technology Change.


Perfection means not perfect actions in a perfect world,
but appropriate actions in an imperfect one.
— R. H. Blyth

Systems are difficult to work with, and seeing things for what they are is an essential first step.  Horst Rittel in the late 1960s distinguished between “tame” and “wicked” problems.  This is not the distinction between easy and hard problems—many tame problems are very hard.  But wicked problems, while not evil, are tricky and malicious in ways that tame problems are not.  The unexpected consequences we’ve seen have been because systems problems are wicked.  We will understand systems better—and why they spawn unexpected consequences—if we understand a little more of the properties of wicked problems and approach them with appropriate respect.

Tame problems can be clearly stated, have a well-defined goal, and stay solved.  They work in a Newtonian, clockwork way.  The games of chess and go are tame.  Wicked problems have complex cause-and-effect relationships, human interaction, and inherently incomplete information.  They require compromises.

For example, mass transit is a wicked problem.  Everyone likes mass transit—unless it comes through their neighborhood, it consumes road lanes, or they have to pay for it.  The difference between something that works in the lab, on paper, or in one’s head versus something that works in the real world and is practical to real people is a characteristic only of wicked problems.

Tame and wicked problems differ in many ways.*  See if the traits of wicked problems as described below sound familiar, either with the examples mentioned here or with situations you have experienced yourself.

  • Problem Definition.  A tame problem can be clearly, unambiguously, and completely stated.  Math problems are tame.  By contrast, there is no absolute statement of a wicked problem.  To state a wicked problem means to also state its solution.  That is, the problem can’t be stated without a proposed solution in mind, and coming up with a new solution means seeing the problem in a new way.  Avoid locking in a problem definition too soon.
  • Goal.  A tame problem has a well-defined goal, such as the QED in a proof or the checkmate in chess.  With a wicked problem, you could keep iterating and refining your solution forever—or go back and consider other solutions.  After all, if a wicked problem is something you can’t define, how can you tell when it’s resolved?  You don’t stop because you’re done (you’ve reached the goal) but rather because of external constraints (you’ve run out of money, time, or patience, for example).  You must strive for an adequate solution, not a perfect one.
  • Solutions.  Solutions are unambiguously correct or incorrect with tame problems.  The solution to a wicked problem is not judged as correct or incorrect but somewhere in the range between good and bad.
  • Time.  The solution to a tame problem can be judged immediately (that is, there is no maturation time), and the problem stays solved.  Euclid’s geometry proofs are still valid today.  Evaluating the solution to a wicked problem takes time (because the results of implementing the solution take time to be appreciated) and is subjective.  Is that a good design?  Maybe, but maybe not.  Like the response to art, different people will have different answers, and the solution causes many side effects (unintended consequences), like medicine in the body.  Additionally, a “solved” wicked problem may not stay solved—wicked problems aren’t solved but are only addressed; they’re treated, not cured.  Your perception of how good the solution is may change over time.
  • Consequences.  Trial and error may be an inefficient approach with a tame problem, but it won’t cause any damage.  Implementing or publicizing a proposed solution doesn’t change the problem.  With a wicked problem, however, every implementation changes reality—it’s no longer the same problem after an attempted solution.  After a failed attempt, the solution you realize you should have tried may now not work.
  • Reapplying Past Solutions.  A class of tame problems can be solved with a single principle.  A general rule for finding a square root or applying the quadratic formula will work in all applicable cases.  By contrast, the solution to a wicked problem is unique.  We can learn from past successes, but an old solution applied unchanged to a new problem won’t produce the old result.  Many unexpected consequences arise when we rush to reapply (without customization) a particular solution we’ve seen before—there will likely be unseen differences between the old and new problems.
  • Problem Hierarchy.  A tame problem stands alone.  It is never a symptom of a larger problem, but a wicked problem always is.  For example, if the cost of something is too high, this can be a symptom of the higher-level problem that the company doesn’t have enough money.  Often, we can’t see the higher-level problem (“This new software is terrific!  I can’t imagine what could be better.”).

Systems are large, complex, and real-world, and they are the domain in which technology is applied.  Industry’s dreams and expectations for its new high-tech products are formed in the lab, but it is in the system of society that they’re put to use.  This brief summary of wicked problems as well as these cautionary examples give some insight into the inherent difficulty of meddling with systems.  This is not to say that we can’t address systems problems but that they should be approached with caution and respect.

Let’s end this chapter with a final example of unexpected consequences due to technology.  In the 1954 short story “Answer,” Fredric Brown envisions many great scientists working for years to build a giant computer network, connecting the computing power of billions of planets.  As the inaugural question for this technological marvel, the gathered dignitaries ask, “Is there a God?”

The computer doesn’t hesitate before answering, “There is now!” 

Everything has both intended and unintended consequences. 
The intended consequences may or may not happen;
the unintended consequences always do.
— Dee Hock, president of VISA

* Rittel and Webber, “Dilemmas in a General Theory of Planning,” Policy Sciences, 4:155–169, 1973.

Photo credit: Wikimedia

Related posts:

"Yes, we are an ignorant species.Yes, the bible God condones slavery."

Christianity’s Unbroken Record of Failure
"“...That is different than celebrating racists and taitors.”Taitors. I love them with butter and sour ..."

Christianity’s Unbroken Record of Failure
"what would be your response if an apologist assumes that the man in the tomb ..."

The Bible Defeats Its Own Resurrection ..."
"You may be too young, but years ago there was a PBS documentary by the ..."

Christianity’s Unbroken Record of Failure
What Are Your Thoughts?leave a comment
  • MNb

    Ha! I’m a chess player and it seems to me that chess isn’t that tame.

    Human interaction: check. It’s is the very nature of the game that humans influence the outcome. One reason Anand failed to defeat Carlsen is that he was not capable of steering the game towards positions that appealed to his strength and to Carlsen’s weaknesses.

    Complex cause-and-effect relationships: check. See next point.

    Inherently incomplete information: theoretically not, but as there are more possible positions on the chessboard than atoms in the Universe practically yes. Both latter points are reflected in the horizon problem, well known by engineers and designers of chess computers and programs.

    Problem definition: check. In many positions (but indeed not all) it’s hard to define what thet problem exactly is.

    Goal: ambiguous. Since long it has been a maxim that those players who aim at mate don’t understand the game. Moreover I have played several games where the goal was just a draw, ie avoiding losing. The main point though is that before those goals are reached many provisional goals must be reached and they often are far from clear.

    Solutions: I grant you that, though the verdict “unclear” is very common in chess analysis.

    Time: ambiguous again. Several chess problems were thought to be solved until some smart guys pointed out that the solutions were wrong. But yeah, lots of endgame positions have clear cut solutions.

    Consequences: I don’t advise to apply trial and error in chess games – it will cost tons of points, the worst damage thinkable. But granted, when analyzing or training it is a somewhat clumsy, but legitimate approach.

    Reapplying Past Solutions: ambiguous again. There are no hard, fast rules in chess to find the best move(s). Every position has its own solutions – indeed, often more than one. There are thumbrules, but every single one of them has its exceptions.

    Problem hierarchy: check. One of the hardest things in chess is facing two problems to solve and deciding which one has priority.