Thursday, March 29, 2018

The Logic of Failure


Books I wish I read earlier in life. The Logic of Failure by Dietrich Dörner. Originally published in German in 1989 and translated into English in 1996, the book’s catchy subtitle is “Recognizing and Avoiding Error in Complex Situations”. It has an excellent cover to match with a large bright red F.



Would it have helped me plan and think through complex issues before I became an administrator? Possibly, but I’m not sure. That’s a good thing, because Dörner doesn’t try to sell the reader a “new” method that grooms leaders into strategic and creative thinkers. He proposes a methodology, but then adds plenty of buts and caveats. The devil is truly in the details because every complex situation is different. We can’t always see when, where and why. There is no substitute for experience, but even experience can cause its own compounding problems. Depressed yet?

How does Dörner study the problem? By putting people in computer simulations designed to model complex situations. Mind you, this is back in the ‘80s. Today’s simulations would likely be exponentially more sophisticated, although I suspect the behavioral results of the participants would be similar today. The setup? There are two. Managing a small fictitious town in an isolated hilly region in Europe. Or managing an African region with different tribes subsisting on farming or herding. Both situations feature complex and inter-related variables, and the point is not just to see who succeeds or fails, but why. As you’d expect, a few do well, many fare poorly, some learn from their mistakes, and others don’t. There are common threads among the successful. The reasons for failure, however, are myriad. But there is a logic to them. A self-help guru would package the common threads into a five-steps-to-success program, but Dörner is much more circumspect. His book closes with a cautionary tale I will discuss at the end.

The introductory chapter includes an analysis of the 1985 Chernobyl “disaster” (although Richard Muller would argue it is less disastrous then commonly thought). From his analysis and observing his simulation participants, Dörner describes four features of complex situations. First, they are complex. His definition: “Complexity is the label we will give to the existence of many interdependent variables in a given system. The more variables and the greater their interdependence, the greater that system’s complexity. Great complexity places high demands on a planner’s capacities to gather information, integrate findings, and design effective actions. The links between the variables oblige us to attend to a great many features simultaneously, and that, concomitantly makes it impossible for us to undertake only one action in a complex system.”

This situation is the bane of scientists designing an experiment to answer a question. We try to keep everything else constant except for one variable, so we can isolate its causes and effects. And when you’re in a complex situation, without the luxury of time to gather all the information you think you need… well, I don’t envy policy makers and administrators (now that I’m no longer one, having some experience being in such situations). It gets worse. “Complexity is not an objective factor but a subjective one,” Dörner writes. “We might even think we could assign a numerical value to it…” Having immersed myself this semester learning about measuring complexity in chemical systems, I’m very much inclined to agree. There are several approaches to measure complexity in molecules and molecular systems, but all the devisers would agree that they are pegged to subjective reference states.

The other three features of complex situations are dynamics, intransparence, and ignorance/mistaken hypotheses. There are many variables for which we have no direct access, thus the “frosted glass” of intransparence. This contributes to our ignorance, and we make assumptions (both implicit and explicit) that might simply be wrong, only to realize later that our actions have made things worse. Dynamics is one feature particularly difficult for us to grasp – how things change over time. Possibly a combination of brain evolution and our education system, we extrapolate linearly when many situations (even simpler ones) might suggest a power law or exponential behavior. I’m certainly guilty of that as a teacher. Just this month I have discussed generating simpler linear models (a common feature in science) multiple times in general chemistry: the Clausius-Clapeyron equation, first and second-order rate laws, and the Arrhenius equation. We drew lots of graphs. Why? Graphs help us translate time into space – turning something dynamic that’s hard to grasp, to something static and easier for our feeble minds to comprehend. And I’m pretty sure most of my students don’t grasp the log scale of a graph axis, even if they can build up the data and sketch the graph.

Dörner devotes a whole chapter to Time Sequences, and another to Information & Models. Other parts of his book discuss the important twin features of Goal Setting and Planning. There are many, many, many places to Fail. I recommend reading the eye-opening examples in his book. There is a logic to them, and I’m not sure this logic could have been extracted without running the simulations. If anything, I felt mildly justified by many youthful hours spent on long complicated games, my favorite being the ‘80s Avalon Hill boardgame Civilization. Its descendant, the ‘90s Sid Meier computer game of the same name (the first version) is probably the last computer game I’ve played. Spending most of my working day in front of a computer, I’m not interested in using it for leisure. I still have my old ‘simulation’ boardgames, but no longer have the time to play them. The shorter ones strip out some of the complexity, and are still fun, but do not teach the lessons of complexity. The variables are fewer. I’ve recently pulled out a few older Reiner Knizia favorites. His designs are exquisite and force tough choices even with a simple ruleset.

In the book’s final chapter, Dörner asks: “How can we teach people to deal effectively with uncertainty and complexity?” The problem: “There is probably no cut-and-dried method for teaching people how to manage complex, uncertain, and dynamic realities, because such realities, by their nature, do no present themselves in cut-and-dried form.” While on average, experienced leader-managers performed better than students in these simulations, there were more than a fair share of significant failures among the experienced.

Dörner’s last example, however, is sobering. Before one of the simulations, participants were divided into three groups. Here are his descriptions. “The strategy and tactics groups received instruction in some fairly complicated procedures for dealing with complex systems. The strategy group was introduced to concepts like system, positive feedback, negative feedback, and critical variable, and to the benefits of formulating goals, determining and, if necessary, changing priorities and so forth. The tactics group was taught a particular procedure for decision making...”

First, the self-evaluation results of each group after the simulation (which was conducted over several weeks). “The members of the strategy and tactics groups all agreed that the training had been ‘moderately’ helpful to them. The members of the control group, who had received training in some nebulous, ill-defined ‘creative thinking’, felt that their training had been of very little use to them.” Just think of all those snake-oil salespeople selling the latest workshop in innovation or creativity for leadership. Heck, I even sold the idea to a group of sophomores where we would experiment combining creativity and chemistry. We’ll reflect on the results at semester’s end.

But how did the actual participants do in Dörner simulations? (In this case it was being mayor of the small fictitious town.) No difference in performance. You heard that right. There was no difference in actual performance! Despite the training. Yet, participants receiving the training thought it helped them. Why though? Dörner’s answer gives me goosebumps.

“The training gave them what I would call ‘verbal intelligence’ in the field of solving complex problems. Equipped with lots of shiny new concepts, they were able to talk about their thinking, their actions, and the problems they were facing. This gain in eloquence left no mark at all on their performance, however. Other investigators report a similar gap between verbal intelligence and performance intelligence… The ability to talk about something does not necessarily reflect an ability to deal with it in reality.”

Does this bring back memories of listening to a well-spoken leader or administrator who turned out to be ineffective or incompetent, or worse? I’ve even done the speaking myself, just a few times, when I’ve had to be on a panel or say something in public. (I try to avoid these situations when I can.) I’ve also been sent to leadership training sessions. I think I learned something from them, but maybe it’s simply just a better vocabulary. I read voraciously to learn, but maybe I’m only acquiring better ways to sound knowledgeable. At present, I’m slowly working my way through Thinking in Complexity by another German author, Klaus Mainzer. It is subtitled “The Computational Dynamics of Matter, Mind and Mankind.” You can bet that I am able to discuss such concepts in Astrophysics, Biology, Consciousness, as smooth as ABC. I know all about non-linearity, attractors, neural nets, and feedback loops. And I am trying to solve the riddle of the origin of life, a complex problem about how complexity arises.

I’m likely not to succeed in solving the riddle. The problem is full of intransparence with too many interdependent variables. It’s a toy problem to work on, and it keeps my mind active. Real-world practical problems, however, are subject to time-sensitive decision making. Delaying a course of action is itself an action, for good or ill, whether one be collecting more data or concentrating on some other urgent priority. If anything, reading The Logic of Failure has made me more circumspect. I wish I read it earlier. Perhaps it could have helped in better decision-making in the past, but perhaps not. There is a logic to failure, but the situations are so diverse that a lifetime might not be sufficient encounter to learn how not to fail in the next complex situation.

No comments:

Post a Comment