The Science

Underpinning the Meta-Problem Method are concepts and techniques drawn from a range of scientific and mathematical disciplines, including descriptive psychology, optimization, decision analysis, analytics and systems engineering.

It starts with descriptive psychology and the idea that maybe we should take an action or make a decision to influence the world for the better. Better is in the eye of the beholder, but it always starts from a desire for change. This is the “dilemma” step in the method. (See the Meta-Problem Method page for the high-level steps)

If someone starts from a dilemma like “what should we decide to do” the implication is that our choices may lead to more or less desirable outcome. Choose poorly, and you will suffer the consequences of your mistake.

That leads to the next two steps, “goals” and the “problem space.” Optimization requires us to sketch out our decisions, our single goal (which can be a function of multiple competing goals), and any limitations on our decisions.

However, optimization assumes you can define the whole problem in one fell swoop.  Humans don’t work that way. We only know what we want when we have an idea of what’s possible. Using the Meta-Problem Method breaks the process down into steps where each one informs the next.

Uncertainty and the unknown can derail the best-laid plans. Our human tendency to stick rigidly to the decision we have worked so hard to reach, and ignore any emerging changes that challenge it, often leads to waste and disappointment. The Meta-Problem Method is designed to prevent this by stating exactly what we don’t know.

Once we understand our goals, we shift to asking how we can achieve them. This is where people usually begin the problem-solving process by singling out one version of the problem.

By contrast, in the Meta-Problem Method, we look at a set of possible problems. That notion, that you can consider a set of problems, is inspired by optimization. It helps avoid the mistakes that can arise when you talk about your decisions and the limitations on those decisions as two separate considerations. Problem space combines the two as a region in decision space.

Once you have options, you have to choose between them. In a multi-objective world, some of the choices in problem space could advance more than one of your goals. These are the “high yield problems”.

Now we are ready to tackle our key decision: Which problem should we choose to solve? Borrowing again from optimization we have the notion of a best option given our goals. Systems engineering helps us compare alternatives based on their very multi-dimensional characteristics. Decision analysis (and real-life experience) shows us that people are pretty good at ranking options but are terrible at saying why they like one option more than another.

So the science in this step basically says “people can choose, having concrete alternatives is important to making good choices, and being able to articulate those comparisons makes all of the above explicit so you can have a discussion about the options in front of you.”

It’s a truism that we don’t know what we don’t know. The process of implementing your decision will likely uncover new information, and the context in which you are operating might change. That’s why the last step in the method involves continuously checking that the problem you chose to solve is still the right problem.

It pays to remember that while we can make permanent decisions, there’s usually no reason to do so.