Problematic Phenomenon

A problem is a situation bearing improvement. You cannot jump straight from the phenomenon to the solution without a causal explanation. Diagnosis is the process of deriving causal explanation from behavior in terms of function and structure, and is the inverse of simulation.

Symptomatic

People diagnose problems quickly every day with no teams or analytic tools. They look closely at the detail, asking the question “What’s wrong?” comparing what they see to their experience (case-based knowledge). Symptomatic knowledge needs to be developed in every business. It is fast and effective. You cannot buy it, and it is becoming harder to find. Fault Tree Analysis and Expert Systems are ways to capture such knowledge. If you cannot fix a problem with a symptomatic strategy, then you just don’t know enough and you need to learn more. Our focus is on gaining that knowledge quickly and efficiently.

Topographic

When experts cannot agree, or testing shows causal explanations are wrong, symptomatic must be abandoned in favor of other strategies. The alternative is generically termed a topographic (model-based) strategy. Our seminars are all about the tactics of using system models to converge on the causal explanation. To converge, hypotheses start very general and proceed towards specific. The power of the search depends on the way the system is split, and therefore the key skill is system decomposition. The objective is to simplify and clarify the complex, to the point where symptomatic knowledge can be brought to bear. Topographic diagnostic strategies are independent of information gathering tactics. Tactics are the specific tests carried out, and the analytic method used to generate explanation.

Functional

One form of topographic strategy makes use of functional models. These describe the means-end relationships of a system. Functional models can be qualitative (descriptive), in which case functions can be recognized by verb-noun descriptions, or they can be quantitative. In functional decomposition, models make use of a block diagrams to explain system dependencies in terms of means-end relationships. We teach tactics for exploiting this type of decomposition, both deterministically and probabilistically.

Structural

As well as functional models, we can make use of structural models to help us develop causal explanations. Structural models describe what our products are made of (rather than how they work), and what equipment is used to make them (rather than how they are made). It also extends to decomposing a single part into features, and then into fundamental elements, facilitating extremely powerful diagnostic tactics that we teach.

Deterministic

It is very important to recognize that causal explanations can never be structural, just as they can never be probabilistic. If we make use of structural or probabilistic models, it is because it is a means to an end. It is because it allows us to develop tactics (tests) that progressively converge towards causal explanation, ultimately leading to a workable solution. The most powerful tactics for reaching a causal explanation are deterministic. These are the focus of the How Stuff Really Works seminar.

Probabilistic

System decomposition can be expressed probabilistically or deterministically, but statistics do not describe how things work. Statistics is useful in helping identify patterns and underlying relationships. The focus of the Convergent Diagnostics for Variation Reduction seminar is exploiting probabilistic tactics to reach a causal explanation for the variance in a dependant variable.

Causal Explanation

Causal explanations eliminate opinions, guesswork, and trial and error. The causal explanation has to be functional and deterministic. The phenomenon, or failure, is based on some governing principle or law. Causal explanation should go deep enough to enable you to remove the problem.

Solution

A situation can be both a problem and a symptom. A solution substantially improves the situation that needed improving. This framework describes the thinking processes involved in problem solving. The framework is important to avoid doing the wrong thing at the wrong time, which is the main reason problems take much longer to solve than they should. Refer to The Art & Science of Fixing Things article (click the solution circle).