Anatomy of Agile Enterprise

Janne J. Korhonen

There is Not a Simple Solution to Every Problem

Vote 0 Votes

"Everything should be made as simple as possible, but not simpler." This maxim, attributed to Albert Einstein, bears particular relevance to how problems should be approached. There is not a simple solution to every problem. Simple problems can be solved with straightforward techniques; complex problems call for more elaborate methods.

Simple problems can generally be solved with a high degree of success by following a recipe: a process, a guideline, a best practice. Complexity arises from the number of nodes and their linkages. For instance an information system grows more complex, as the number of function points increases. Allenby calls this Static Complexity. Cause and effect relationships are readily apparent, and structured, linear techniques and processes are applicable. Examples of such techniques include single-point forecasting and operational procedures. The decision model is to sense incoming data, categorize the data, and then respond in accordance with predetermined practice (Kurtz and Snowden, 2003).

Dynamic Complexity arises as the nodes of the system interact in new and unexpected ways, changing the relative position of nodes. As cause and effect relationships are separated over time and space, complicated problems require coordination and/or expert know-how. Analytical methods and systems thinking are called for. Examples of applicable techniques include experiments, expert opinion, and scenario planning. The decision model is to sense incoming data, analyze the data, and then respond in accordance with expert advice or interpretation of that analysis (Kurtz and Snowden, 2003).

Whereas simple and even complicated problems can be addressed with linear thinking, techniques and tools, these means render inadequate when dealing with complex problems of Wicked Complexity. This type of complexity arises from the reflexivity, intentionality and evolution of human systems and institutions that creates contingency and unpredictability: "Information developed by humans about human systems is, by definition, a new part of the human system that it arises from, and thus it immediately affects and changes the underlying human system." (Allenby, 2009). Complex problems are non-linear, i.e. inputs and outputs are not directly correlated. The solution is a part of the system and it cannot be applied as a recipe to other, like problems. Cause and effect relationships between interacting agents are mutual. Emergent patterns can be discerned, but only in retrospect. The decision model in this setting is to create probes to elicit the patterns, then sense those patterns and respond by stabilizing the desirable patterns, while destabilizing the undesired ones (Kurtz and Snowden, 2003). In this context, narrative techniques are powerful, as they convey a large amount of knowledge or information in a very succinct way. Other tools for managing in a complex context include large group methods, simple rules, experimental probes, and "ritual dissent" (Snowden and Boone, 2007).

Finally, Anthropogenic Complexity arises "because humans are impacting not just local environments and resource regimes but the global framework of physical, chemical, and biological systems is new and challenging, in that no discipline or intellectual framework enabling rational understanding at that scale yet exists." (Allenby, 2009). Chaotic problems pertaining to this type of complexity have no systems-level cause and effect correlation. The decision model in this space is to act, quickly and decisively, to reduce the turbulence, sense the reaction to the intervention and respond accordingly (Kurtz and Snowden, 2003). Immediate action and direct, clear communication are of great importance.

A common failure is to consider complex and chaotic problems as reducible to their constitutive parts and to apply the same linear thinking and rational planning that work in the simple and complicated contexts. Any attempts to categorize or analyze problems in a structured way are futile. As depicted in Figure 1, the cost/effort then becomes prohibitively expensive very soon.


Figure 1. Linear techniques are not applicable to complex problems.

As illustrated in Figure 2, the problem solving techniques should be contingent on the level of complexity. Whereas linear techniques are applicable, and affordable, in the face of simple and complicated problems, non-linear techniques are needed to adequately address problems of complex and chaotic kind. Although these problems may still be expensive to deal with, non-linear techniques significantly extend the range of problems that can feasibly be solved.


Figure 2. Requisite techniques are contingent on the level of complexity.

"We cannot solve our problems with the same thinking we used when we created them." Ergo, there is much need for non-linear techniques.


Thank you for sharing. I agree with all your statements, the only think I wonder is why you draw the figure 2 as you draw it. What is the metrics behind stating that "solutions" in the complex problem space are expensive and in the chaotic problem space even worse?

I would agree that finding the ONE best solution might be expensive or prohibitive, but that can easily be said also for simple and complicated problems. Selecting one "Probe-Sense-Respond" or "Act-Sense-Respond" Solution doesn't necessarily translate into costly. Actually most of the innovations I observed in the last years have been very simple (and cheap) answers to VERY wicked chaotic or complex problems.

Can you elaborate that a bit further?



Thank you for your note. My implicit assumption was that the scope of the problems increases along with their complexity. Simple problems are typically local and contained, complicated ones less so, and complex and chaotic problems are of very large, even global, scope. Solutions in the complex and chaotic space are also idiosyncratic in nature, so you cannot scale up best and good practices. Hence, the inherent effort to solve the increasingly complex problems is progressively higher, yet the cost/effort ratio is evened out by non-linear techniques that are better suited to complex problems than the linear ones, which may actually create more problems (and costs), when misapplied.


I understand your argumentation chain, but I am not fully buying it. Assuming you apply linear techniques then your cost/effort curve is a good way to represent it (figure 1), but I believe that figure 2 can be dramatically different with even less cost/effort on chaotic problems than on simple problems, due to the nature of one act-sense-respond cycle. If that is a "good-enough" hit already it is most likely a very cheap answer.

Janne J. Korhonen provides insights into how information technology can be applied strategically to catalyze organizational change and responsiveness. Drawing from both theory and practice, he discusses agile enterprise and its governance.

Janne J. Korhonen

Janne J. Korhonen is an independent business and IT consultant,specializing in enterprise architecture, business process management,service-oriented architecture and pertinent governance models. He has over ten years of experience as an architect and consultant in a variety of extensive and mission-critical IT projects. With strong theoretical underpinnings, his consulting encompasses systemic co-development of business, organization and information technology.

Recently Commented On

Recent Webinars

    Monthly Archives