The Problems With Problems: Part 2

• Not all problems are alike: Treating every issue with the same method leads to poor outcomes. 

• Problems differ in structure and predictability, which leads to a typology of four types of problems: technical, predictive, strategic, and wicked problems.

• Each problem type requires a different approach.

• Effective problem-solving recognizes the type of problem first, then matches the solution method.

Not All Problems Are the Same

We all know problem-solving is an essential skill. We teach it in school, train it in the workplace, and expect leaders to excel at it. 

Here's the problem: most problem-solving approaches treat all problems the same.

They’re not.

Some problems have a clear goal and a clear solution path. Others don’t. Some situations are stable. Others are unpredictable. 

Lumping all of these together under a single “problem-solving method” is similar to having the hammer and thinking everything is a nail.

Let’s understand problems by using a simple typology based on two dimensions:

• Structure: Is the problem (or situation) well-structured or poorly structured?

• Predictability: Is the problem deterministic or stochastic?

This gives us four types of problems—each requiring a different approach.

1. Well-Structured + Deterministic

Encountered in technical or analytical problems.  (Engineers and coders love these kinds of problems!)

Definition: Clear goals, defined constraints, predictable outcomes.  When we do the same thing, we get the same outcome.

Examples:

• Fixing a network outage.

• Debugging a software program.

• Fixing a broken machine.

• Planning truck routes in a logistics system.

Common solution methods are logical and use known, established procedures, based on the results of root cause analysis and decision trees.

How it works: You measure, analyze, and respond. The solution is repeatable and verifiable. Efficiency is important.

2. Well-Structured + Stochastic

Encountered in predictive or risk calculations in future planning.

Definition: Clear goals and models, but outcomes have a level of uncertainty or randomness.  The outcome varies even for the same input and actions.

Examples:

• Forecasting sales next quarter.

• Managing financial portfolios.

• Planning crop yields with weather risk.

Common solution methods seek to manage probability and uncertainty, using modeling tools such as Monte Carlo simulation, Bayesian inference, and sensitivity analysis.

How it works: Your objective is to manage risk and adjust continuously.  

Something to watch out for: People often view these kinds of situations as deterministic (have you ever had a chance to sit in on a budgeting meeting and listen to the discussion of sales goals for the next quarter or the next year?).  These are very different.  Noise, randomness, and unexpected external events distort predictions. 

3. Ill-Structured + Deterministic

Often encountered when designing new products and developing strategies.

Definition: Ambiguous goals, but once they are defined, the system responds predictably.

Examples:

• Redesigning a city’s transport system.

• Building a new note-taking app.

• Developing a curriculum (at any level, for a course, an institution, or a national standard).

Common solution methods require considerable framing before solving.  This may require open-ended exploration and understanding of different forms of value.  Design thinking, stakeholder mapping, and systems mapping are common tools for setting scope and framing.

How it works: Need to think in terms of: “What’s the real question here?” This does not follow a straight path (which can be uncomfortable for many).  Iteration and learning loops help find the new pathway.

4. Ill-Structured + Stochastic

These are adaptive or wicked problems, like many communities and societies are grappling with now.

Definition: Goals shift throughout the process, causes interact, outcomes are uncertain, and can even counteract each other.

Examples:

• Addressing challenges for youth, such as unemployment, bullying, and teenage pregnancy.

• Climate adaptation.

• Reforming healthcare in low-income areas.

• Delivering solutions to rural communities.

Common approaches recognize that any intervention changes the system. There’s no single root cause or fixed goal.  Tools such as systems thinking, scenario planning, and continuous dialogue are used to identify insights of the complex situation.  Small, safe-to-fail experiments, that probe the situation can also help.

The trap: Traditional problem-solving doesn’t work here, and situations are often made worse by the belief that the best solutions are quick and decisive.

Stochastic Problems Deserve Extra Attention

Stochastic problems aren’t rare—they’re everywhere. Markets shift. People behave unpredictably. Social systems interact in ways we don’t fully understand.  The stochastic nature of these situations arises from hidden interactions (competitor moves, consumer behavior), delayed feedback (policy effects, long-term risks), and high variability (politics, public sentiment).  

Solving stochastic problems with linear tools, like root cause analysis, is a mistake. Instead, solutions require adaptive models that respond to data in real time.  Test with small probing experiments, and then seek continuous adjustment, rather than a single final solution.

Summary: The Problem Typology Matrix

Final Thought

To get better at problem-solving, we need to get better at recognizing what kind of problem we are solving (a skill that I like to call problem-finding). Not all problems are the same. Not all problems can be solved the same way.