The Logic of Failure


dietrich dörmer

EXEC SUMMARY: This book systematically and logically diagnoses the process of why we fail. dörner begins in the Introduction by providing snippets of computer situations and how two individuals fail to resolve problems. He then continues by presenting psychological reasons in blunt terms why they failed. He establishes the path to failure in rather stark and almost fatalistic terms. It is as if we are all doomed to fail, unless we heed the correct procedure to problem solving, which is the subject of his book. CHAPTER ONE provides three examples, two computer generated and one very real, that detail the process of failure step by step. Each of the three vivid examples lets dörner probe the path to failure through storytelling and detailed accounts of the events which led to utter failure. Each scenario concludes with blunt assessments of why the participants failed. CHAPTER TWO establishes the nomenclature of why systems fail. He uses the three failures of the previous chapter as his source for examples and proceeds to analyze the process of failure step by step. First all problems are complex, that is that they have many features that are interrelated. Second they are dynamic in that they develop a life of their own. Thirdly we can't always see everything because something just might not be visible. Lastly our ignorance and mistaken hypotheses lead us to failure. The steps in planning and action are presented: define goals, develop a model and gather info, prediction and extrapolation, planning of actions, and review of effects of actions and revision. Each of these steps will be the substance of each of the subsequent chapters in the book. CHAPTER THREE Setting Goals is a complex process. We must set useful goals. We must understand interrelationships, be clear, avoid "repair behavior", be conscious of contradictions in our goals, and above all know our own limitations. CHAPTER FOUR Information and Models: speaks of gathering info and analyzing it to make decisions. One situation is not to be treated like another. One success can't be transferred to another. Constant feedback is crucial. Understanding interrelationships is very important. CHAPTER FIVE Time Sequences outlines the effect time has on systems. Time is the 4th dimension and many aspects of temporal sequences are key to effective systems management. Oscillation, sudden reversals, exponential growth, events in time are not isolated but are related. Timing, patience, time delay, and constant tinkering are key to systems management. CHAPTER SIX is about planning which involves: condition element, action element, and a result element. CHAPTER SEVEN should be reread.


  • Begins with story of town trying to alleviate downtown pollution form cars and at the same time maintain economic boom. Changes were too aggressive. Desired result was not achieved and city suffered economic collapse.
  • failings of human thought and action and arrogant belief that failings are found in other people. We assume we could do better. When planning and making sweeping changes, one must be conscious of effects on the whole.
  • Computer game of West African village was played by the economist and physicist. Disaster to the village and animosity between the 2 ensued.
    CONCLUSION: They made terrible decisions. They did not consider the consequences of one action on another. The Moros are a metaphor of all problems: innumerable interrelated subsystems, and we need to think in terms of these interrelations. It's not about the spotted owl, it is about what the spotted owl represents in the larger picture.
  • We need to see the problem embedded in the context of other problems. We must learn to think about systems not isolated problems. We must consider side effects and long term repercussions.
  • Are we turned loose in industrial age with brain of prehistoric times? Is it too difficult to visualize results? Male dominated society? (serial male and parallel female) Western "analytical" thinking. All of these have been called the reason for failure.
    Page 6-7 speak of cures offered by many.
  • The author suggests there is no quick fix except in our way of thinking. It is flawed. We have developed bad habits. Each small mistake adds up until BANG. But it is not just thinking..
  • We have hidden agendas: values and motivation. Thought helps us achieve our goal but fails to go to the root of the evil. Is it "good thoughts plus stupidity" or "evil intentions plus intelligence"? shortsightedness.
  • computer allows for complex simulations. Failure does not strike like a bolt, it develops gradually. Failure starts with the habit of thought. The key is to learn how to be better problem solvers. Start wisely, make corrections midcourse, and learn from failures.

    ONE: Some Examples
    The Lamentable Fate of Tanaland: a simulation. 12 participants had dictatorial power over area to promote its well- being over a ten year period. They had total power, no opposition, to determine hunting, farming, etc etc. They had six opportunities to gather, plan and decide. Old problems repeated themselves, new solutions brought new unforeseen problems. Catastrophe inevitable: linear.

  • WHY? The participants established their modus operandi for Tanaland early and did not alter it much later. Conclusion: more thinking and less action. Redefine the problem. Don't get bogged down on a separate part of the whole thereby losing sight of the whole.
    CONCLUSIONS: acted without prior analysis of the situation, failed to anticipate side effects and long-term repercussions, assumed that the absence of immediately obvious negative effects meant that correct measures had been taken, let over involvement in "projects" blind them to emerging needs and changes in the situation, were prone to cynical reactions.

    The Not-Quite-So-Lamentable Fate of Greenvale: Another computer simulation involves a small town of 3700 people whose main employer is a watch factory. In this scenario, the participant was mayor with a ten year term, dictatorial powers and the watch factory was part of the municipality. This study observes two participants, Mark and Charles. Charles was a "good" mayor and Mark was a "bad" one. Why?

  • "Good" participants in this exercise made more decisions than "bad" ones. All participants made their decisions in eight sessions. "good" participants made more decisions during all of the sessions and especially in the last four sessions. (Tinkering) "Good" participants also made better decisions which had more of an effect on the system, they understood interrelations. They acted more "complexly." "Good" participants were concerned about the factory sooner than the "bad" participants. "Good" participants tested their hypothesis more often. They asked more "why" questions as opposed to "what" questions. They were interested in the network of events rather than in one event. "Bad" participants changed the subject more often, or to be subject to "ad hocism" (easily distracted).
  • Innovation and stability indices measure the degree to which a participant stuck to a topic and for how long. A good participant had a low innovation index which shows they were on the "right" topics while a high stability measures fortitude or stick-to-itness (a terrier), they continued their focus. Bad participants switched from field to field and didn't stay with one too long. "Good" participants were more self-reflective. They structured their thinking and spoke out loud more. "Bad" participants shifted the blame.

    "Chernobyl in Tanaland": I leave this one to the notes that were passed out.

    CHAPTER TWO: "The Demands": here we get the criteria by which we set out to solve problems. From the three situations we find much in common. All are complex with interrelated variables. Intransparency exists, that is we can't see everything. They all are dynamic, they have their own energy. Individuals usually don't understand the systems.

  • "Complexity" is manifest with a great many features. Lots of players, variables, things to consider. Domino effect. Interdependence is crucial. One action has an effect on another variable. Complexity is seen when we have ten variables and five links among them. The more variables and links the more complex. It is subjective. Driving a car is an example. The difference between a new driver and an old driver. The old driver manages complexity well by developing "supersignals."
  • "Dynamics" says that systems have a life of their own. They have their own inertia. This creates time pressure and decisions may have to be made without all of the facts.
  • "Intransparence" is when we may not be able to see everything. Results or expectations may not be visible.
  • "Ignorance and Mistaken Hypotheses": requires that we know "structural knowledge." We should understand how things are now and how they will influence one another. This is called a "reality model." Explicitly the variables are known while implicitly they are invisible (intuition).

    Now that the characteristics are established we will consider guidelines.
    Defining Goals Chapter 3: This is the first step in dealing with a complex problem. This is not obvious at first. This is a constant process of redefining goals as we proceed.
    Formulation of models & gathering info Chapter 4: Obviously gathering info is crucial but time will make it easier said than done. How much is enough? We need to arrange it in a picture to show complexity.
    Prediction and extrapolation Chapter 5: Assessing the info and the model to determine effectiveness in achieving goal.
    Planning actions, decision making, and executing actions Chapter 6: It is time to act. Acting in a reasonable manner is difficult. "Methodism" is a tendency that can impose a crippling conservatism on our activity. Rashness is the opposite. Balance is key. Decisions follow planning and actions follow decisions, and assessment is most important. Replan, rethink, redo. Don't get stuck in the rut. Be prepared to call a failure a failure and fix it. Finding the middle path is hard between sticking to a doomed plan and giving up is not easy. See page 43 for the diagram. Review of effects and revise Chapter 7:

    CHAPTER THREE: Setting Goals.

  • "Requirements of Goal Setting": Why set goals? a reason, to accomplish something. to make it as it should be or to prevent or avoid some end.
  • Goals may be:
  • positive or negative: goals can be positive or negative. pos= to achieve a definite condition; neg= want some condition "NOT" to exist. neg is not clear, try to avoid them.
  • general or specific: gen = single criterion or a few; specific = many criteria.
  • clear or unclear: provides criteria to determine if the goal has been achieved.
  • simple or multiple: mult means you will have to attend to many factors at once.
  • implicit or explicit: implicit goals are hidden and not readily visible but there nonetheless.

  • "Management of Goals": attempt to make unclear goals clear; implicit explicit; general specific; negative positive. Use "efficiency diversity" to do this. That is many different possibilities (diversity) for actions that have a high probability of success (efficiency). To clarify a point perhaps "deconstruction" of the goal is needed. Tear it down and isolate. Interrelatedness and order are important. We cannot solve all of the problems at once. organize a list. Central problems will produce common peripheral problems. Money is central. Distinguish between central and peripheral. Rank according to importance and urgency. Resolve conflicts between importance and urgency. Time is a factor which creates "muddling through" which is bad. delegation may help only if problems are independent of each other. Distinguish between delegating and dumping responsibility. Stay in touch with delegated problem. Ranking and delegating are good when we have more than one goal. When we solve one we may create another(s). A third method to solve goal conflict is to reshape the entire system. Sometimes the problem is not wanting to know and not knowing. Pay attention to what we don't want to change.
  • "General Goals and 'Repair Service' Behavior": is that situation when you do not have clear goals. When you start muddling through. Wrong problems get solved in repair service. Don't select just problems you can solve. Disregard for failings and malfunctions that may not exist yet. Don't concentrate only on one problem. "flow experience" happens when one solves a problem only to be met by another (peeks and valleys).. Interim goals may dislodge the primary goal (goal degeneration). Too often we concentrate on wrong problem.
  • "Liberty, Equity, and 'Voluntary Conscription'": Pursue more goals than to pursue one. Contradictory goals are the rule. Vicious circles evolve. Train many for a job and soon many will be unemployed. Cause and effect. Same as dealing with an urgent problem which creates a second problem. "We'll cross that bridge when we get to it." This is "goal inversion." Give up one goal to pursue its opposite. Over steering. Doublespeak. Oxymoronic situations: "voluntary conscription." "conspiracy theories" are another way of resolving goal conflicts: blaming others.
  • "Self-Protection": We need to preserve a sense of our own competence. Know our limitations, the Peter principle. Hard to say we failed. Setting useful goals in complex systems very important.
    Return to Introduction

    CHAPTER FOUR - Information and Models

  • "Reality, Models, and Information": Opens with a pond as a system. Why does it stink?? Inadequate assessment of reality. What is a system?? a network of many variables in casual relationship to one another. (see page 73) Interrelationships: positive feedback, negative feedback, buffering, critical variables, and indicator variables.
  • "positive feedbacks": an increase in a variable produces an increase in that variable. A large population gets larger. This tends to undermine a system.
  • "negative feedback": increase in one variable produces a decrease in another. Perpetuates the status quo. It maintains equilibrium. Predator-prey, thermometers. "well-buffered" incorporates many negative feedbacks. It can absorb many disturbances without coming unstable. Feedback systems consume and may exhaust resource resulting in breakdown temporarily or permanently.
  • "critical variables": are those that interact mutually with a large number of other variables. Key variables.
  • "indicator variables": Of little importance on their own, but depend on many other variables. They are the clues to the overall system.

    We must go beyond just causal relationships but to abstract relations as well. We should know the hierarchy as well. Think in analogies. Explore the unknown via the known. To deal effectively with a system we need to know: 1. how the causal relationships among the variables work; 2. how components fit into a hierarchy; 3. need to know the component parts and into which variables those parts are embedded.

    Acquiring knowledge of a system can be achieved by analogy. As our knowledge of a system grows and we make changes we need to gather further info to assess the effect of the changes.

  • "Solving Problems One at a Time": Reliving the Moros scenario, the author provides in depth accounts of many of the participants. None of the participants realized that they were dealing with a system. They conceived a sequence of problems, did not see side effects and repercussions. If we do not concern self with problems we do not have, we soon have them. Lots of "if then" statements.
  • "It's the Environment": Better to know how variables interrelate. Consider the Greenvale scenario and how one participant created a reductive hypothesis which centered on the fact of "the satisfaction of the citizens." Everything centered or worked from this hypothesis. The problem with a single hypothesis is that other factors emerge and if one returns to "are the people happy?" then nothing else happens. Why is the watch factory not producing? This mayor wanted the workers happy and couldn't see that production was low because of poor technology. A single hypothesis like this also allows one to ignore other information that comes in. We become infatuated with our hypothesis to the point of ignoring other info.
  • "Prime Numbers and Tourist Traffic": uses an example from Mathematician Fermat who devised a way to determine prime numbers. But it was proved wrong later on. Point is he made a mistake of overgeneralizing, a common error in formulating hypotheses. A necessary generalization can easily evolve into an overgeneralization.
    In Greenvale, the promotion of tourism was successful once and was repeated over and over again only to be failures. Why?
    In a complex situation with many interlocking elements, effective measures in one situation may not work again in the same or similar situations. Every situation has to be considered afresh. The fire game is spoken about. A chief must dispatch his 12 brigades appropriately to fight fires. pp 95-7. Strategy is a series of makeshifts, bringing knowledge to bear on practical life. It is the art of acting under the pressure of the most demanding conditions. We must keep track of changes and adapt which runs contrary to generalizations. Before we apply an abstract concept to a concrete situation, we should sub,it to "strategic" scrutiny to see if it is appropriate.

  • "The Pale Cast of Thought": suggests that if we gather a lot of info we will have more trouble coming to a decision. The more we know the more we know we don't know. Once we have made a decision, relief comes since we have relaxed and are finished. new Info comes in and we tend to ignore it. Success is relative to time, info gathering, and acting. If time is short gather lots, ask few questions and act. If time is long gather lots and ask lots and act and question again. We combat uncertainty by either acting rashly with minimum of info or gathering too much which may inhibit.
    Return to Introduction

    CHAPTER FIVE Time Sequences

  • "Time and Space": We live and act in a four-dimensional system. In addition to 3D of space the 4th is of time to the future. In space we recognize configurations like chairs etc, but in 4D, time we can get confused. Melody is a config in time. It extends over time. A time config is available for examination only in retrospect. It is relative to now not then. Life, business, ed must pay attention to patterns in time, trends. Strategic planning must take this into account, time.
    Interesting how time worked in the following scenario: A got mad at B. C, B's friend attacked A rather rudely. C lost face etc with larger group and later in meeting did not get support needed by group. Did A set up C early for a later battle knowing C overreacted often? Did A exercise old adage: "Make your opponent mad; then maybe he'll make a mistake."
    Spatial patterns like parking lots are easy to see incompleteness and as easy to complete based on symmetry (asymmetry), repetition, etc. But in time configs we tend to treat temporal development as individual events. They are not, they are interlocked. Building of schools is an example.
    We rely on a few mechanisms of prognostication to gain insight into the future: extrapolation from the moment is that aspect of the present, anger, worry, or delight which play a key role in our predicting for the future. 2 factors then come together in extrapolations from the moment: 1. limited focus on notable feature of present and 2. extension into future is based on present knowledge. Fixation on now affects new. Too much significance on now. Thus how do people form their ideas of the future?

  • "Lily Pads, Grains of Rice, and AIDS": see pp 110-11 lily pad and grain of rice math problems. Both deal with "exponential growth". Exponential growth is different from linear growth. Exponential is a multiple whereas linear is the same amount. Exponential growth is similar to rate of growth, a percentage. One's estimate and actual are light years apart. To say something will grow an average of 2.5 % per year is a lot over time. Look to mutual funds, your pension fund, your savings account, interest rates.
    In 1985, an author reported 262 cases of AIDS in Germany. In mid-August there had been 230 cases. Of those 109 died. The author concluded this was a small number compared to cancer and all else which kill mankind. Author was like king in grain of rice story, condescending to numbers and exponential growth. Rate of growth is a %. In this case it was 130%. So 262 becomes 16,863 in 5 yrs, 1,085,374 in 10 yrs and that is Germany. (using compound-interest formula). Three different predictions followed. One way under another way over and the third with the facts said no worry when it proved a 183% annual growth. page 116. Clearly real world assessment did not provide writers and readers with the ability to predict correctly. What is the rate of growth over TIME?

  • "A Premature All Clear?": Be weary of deceleration.

  • "Laymen and Experts": looks at persons or institutions which have to make predictions under time configs. Why do group predictions fail. 1st they gather info and make math functions to fit the data.

  • "Twenty-eight Is a Good Number": deals with time delay. oftentimes when we act we have to wait for the effect. To expect immediate result is a failure. Patience is required. It becomes a ritual as we bounce back and forth between extremes looking for the median.

  • "Predators and Prey": Besides oscillation we can have sudden reversal in the direction of development. Growth in economy is interrupted by recession. A stream suddenly dries up. Catastrophes really aren't sudden. It is really cyclical. predator-prey. Lots of prey the predators thrive. Prey disappears predator disappears and prey come back so too the predator and so on and so on.

  • "The Moths of Kuera": speaks of understanding processes. Altering one thing effects another. Watching the one before acting with the second is about timing. Conclusion is that few know how to deal with time-dependent systems.
    Return to Introduction

    CHAPTER SIX Planning
    If we want to deal with a complex problem, first we define our goals. Then we make a model. Then we plan

  • "Go Make Yourself a Plan": In planning we don't do anything we consider what we might do. Essence of planning is to think through the consequences. If individual actions won't work we consider sequences of actions. We should consider the consequences of individual actions. It is sending up a series of trial balloons. We consider long chains of imagined actions which consist of individual links. These three links are a condition element, action element, and a result element. "Given such and such conditions, I could take this or that action and achieve this or that result." Plans can branch out because of circumstances beyond control, or can loop, or be dependent on other action. Plans can go in reverse too. If the goal is to get someplace by train perhaps the last connection should be considered first. Use both forward and reverse to find best. Clarifying goal is crucial in planning esp reverse. Another strategy is "hill climbing" which includes steps only in a direction to achieve the goal. This keeps one from detours but you may be on the wrong hill. Set "intermediate goals" before the main goal. This way not too much has to be redone. It is like always saving your work. We cannot apply all methods to all problems. No clear goal can't use reverse. Real long range plans use intermediate planning.

    Narrowing problem sectors gives us manageable problems. Trial and error is good if possible. Cull unsuccessful strategies to see if they may work or may work if modified. This keeps us from being entrenched in same old same old. An important method is "thinking by analogy." One jumps into the fray and then figures out what to do next or sink or swim. In complex and ever changing situations delegate, meet often, and allow independence.

    Plans can be too crude or too detailed. The trick is finding the balance.

  • "Rumpelstiltskin": Planning requires three elements: condition, action, and result. Keeping all three juggled is hard. Certain actions require certain conditions or we may need to create the condition to act. Act takes time and effort. result may be different from expected, requiring more action, different condition. Since we can imagine we are set. Rumpelstiltskin said: "today I bake, tomorrow I brew, the next day I take the Queen's child." But he didn't keep his name to himself. His plans go awry. Remember "frictions." They are the irksome little things that come up, the flat tires, the down computer, the closed door with no key, the late bus connection.

    Methodism is the system of using the same method of one success on another problem. Macros. Problem is if something changes methodists falter because they are tracked.

    Bad problem solvers used: constantly, every time, all, without exception, absolutely, entirely, ... while good problem solvers used: now an then, in general, sometimes, ordinarily, often, a bit, somewhat, among other things, .... Good problem solvers analyze and try to find reasons, bad are dogmatic and assert.

  • "Learn by Making Mistakes? Not Necessarily!": Unexpected results should always cause us pause. Wrong info? Incorrect method? bad goal? Those who ignore negative results are called "ballistic" like the cannonball. Once it is fired we are out of the scene. If we don't look at consequences of our behaviour we think we are competent. Another is "dosage" which deals with the intensity with which a measure is carried out. Using 3 tons of fertilizer on an acre when 2 tons is needed. A third measure is of deviation from moral and ethical standards. Thusly, "ends justify means."
    Return to Introduction

    CHAPTER SEVEN So Now What Do We Do?
    sums it all up. People fail because they fail to formulate their goals, don't recognize contradictory goals, to set clear priorities, badly mishandle temporal developments, and fail to correct their errors. The first reason for failure is the slowness of human thinking. Our slowness obliges us to take shortcuts and prompts us to see our scarce resources to save time and effort. Instead of clarifying interrelationships and variables, we select one variable as central. Gather info and save time later. Set rules, deal with temporal configs. A second reason for inadequacies lies in the realm of cognitive processes. We as self-protecting and won't take chances. Methodism is a factor here. Also we only solve problems we can and forget about those we can't. A third reason is a difficulty with dealing with time. We have a limited amount of info we can take in at a time, although we can store large amounts. A fourth reason is that we don't think about problems we don't have. In short we are captives of the moment. "Operative intelligence" is knowledge we have from experience. The key to solving problems is thinking about our own thinking. One way to solve problems is through simulations. Review 197-8.
    Return to Introduction