Classifying Complexity and Choosing a Response
4 min read
Lesson objectives
- The learner classifies the uncertainty in a working situation, selects a defensible response, and reclassifies when evidence changes.
Please sign in to use all the tools
Once you sign in, the reading tools are yours: larger or smaller text, typeface, line spacing, focus mode, highlighting and bold, and notes saved to your account and gathered in My study.
An agile mindset does not mean every piece of work needs a sprint or a task board. Begin with the situation: do we understand the relationship between cause and effect, and do stakeholders agree on the need? For a repeatable repair with a known result, a clear procedure and execution check are appropriate. If a problem can be analyzed but requires expertise, gather evidence and seek expert judgment before choosing a fix. If user behavior or the effect of change emerges from many interacting factors and cannot be predicted reliably, run a small safe probe, observe, and adapt. In an unstable incident threatening service continuity, stabilize first; do not make users wait for a classification workshop.
Cynefin is a sense-making framework for a decision context, not a permanent label for a whole project. In a clear context, cause and effect repeat and a standard procedure can work. In a complicated context, experts can discover a relationship through analysis, as when diagnosing database degradation from operational evidence. In a complex context, patterns emerge after safe probes, as when learning how a new customer notification changes behavior. In chaos, act promptly to reduce harm, then gather evidence and reclassify. One initiative can contain all these situations: account protection may be clear, performance tuning may need specialist analysis, and learning customer preference may be complex. When information is insufficient, state the uncertainty and ask what would distinguish the domains instead of choosing a convenient label.
Agile Manifesto values help prioritize people and interaction, a working outcome, customer collaboration, and response to change. They do not abolish plans, documentation, or contracts; they prevent those artifacts from replacing learning and value. When a manager asks for a fixed date for a product whose need is still uncertain, do not answer that planning is forbidden. Offer a near-term plan with assumptions and review points, stating which commitments are stable and which depend on evidence. Nor is ‘complex’ a license to ignore safety. A probe needs an exposure boundary, appropriate permission, and a stop signal if harm appears. In a clear situation, do not spend a week surveying opinions as a substitute for a known and validated procedure.
For an exam scenario, ask three questions in order. What decision is needed now: repair existing harm, choose a product direction, or learn a cause? What do we actually know, rather than assume? Does the answer option provide an action suited to the uncertainty and generate useful evidence? If the result is highly predictable and the context stable, a ceremonial experiment adds little. If customer signals conflict and no stable causal rule is known, do not prescribe one expert's solution as universal best practice. Revisit the classification after consequential changes. A chaotic incident can become a complicated diagnostic question once service is restored and records are available.
An approach-suitability tool is not a vote for the word ‘agile.’ Put clarity of requirements on one axis and clarity of solution on another, and write the evidence behind the situation's position. When people agree on a need but implementation is unknown, specialist analysis or a prototype may be appropriate depending on whether a solution can be inferred. When technology is familiar but user value is uncertain, a longer execution plan does not answer the value question; behavior needs testing. The boundaries of these tools are approximate. Dividing reality into boxes aids discussion but does not prove every situation occupies one box. Name the part controllable by a known procedure and the part requiring learning, then choose a point for reassessment.
Lesson summary
The learner classifies the uncertainty in a working situation, selects a defensible response, and reclassifies when evidence changes.