In this short post I’m going to show you how to create a visual model of your complex problem with a 4-step problem mapping method.
This type of modeling can be applied to any complex social problem, like homelessness, poverty or crime. Whether you’re a street activist or long-time systems change practitioner, the systems thinking method I share below can help you gain understanding about the structure and dynamics of your problem and increase your likelihood of making good decisions about solving it.
These are the exact steps we use to start mapping in my online course about systems thinking and solving social problems.
In the post I’ll also talk about the different types of problem mapping, how mapping itself is a problem-solving process, and give an example of mapping the issue of illegal opioids.
It doesn’t matter what you call it
Let’s get this out of the way first. There’s a lot of methods for visualizing connected ideas or systems, including problem mapping, mind mapping, cognitive mapping and issue mapping. There’s also a variety of ways people refer to the results: mind maps, mental models, causal loop diagrams, solution maps, and problem solution maps.
When you’re just getting started, jargon and differences in mapping method don’t matter very much. My intent in this blog is to remove the somewhat artificial barriers that prevent people from regularly creating and using problem maps.
For now, I don’t want you to worry about any of this. Just begin with Step 1 below by putting your ideas about the problem and its causes on paper. You can easily adjusting your map later if you want it to reflect a certain mapping protocol.
Problem mapping as problem-solving process
Mapping is just one of many problem solving techniques. But, it is particularly suited to complex problems with many variables and interconnections. Whereas verbally describing a series of complex relationships is very difficult, a simple picture really can be worth a thousand words.
At the beginning, it’s important to recognize that the goal isn’t to create a perfect representation of reality. That’s not possible and it would be foolish to try. Rather, the process of creating the map is about gaining insight about problem in way that can’t be had with words or equations alone.
Along the way, you’ll make explicit many assumptions you have about the problem, as well as how they are connected. In other words, mapping allows you to simultaneously capture details about parts the problem while creating a representation of the “big picture.” This type of switching back and forth from reductive analysis to synthesis is an excellent complex problem-solving approach.
Mapping isn’t a choice
You might not think you need to formally model your problem, but the question is not whether you should model or not. Mental models are always used, even if only implicitly in our heads or baked into the assumptions we make. The choice is really about whether you want to use an implicit and vague model in your head or you want to use an explicit and detailed model.
I know you already have a rudimentary idea about your social problem and what causes it. This is your most basic mental model of the problem. The question is – how accurate is it? For most people who haven’t written it down, the model in their head is simple, perhaps a linear cause and effect model.
Complex problem mapping example: illegal opioids
As an example, let’s look at the issue of illegal opioids.
The two opioid models and policies I show bel0w are adapted from Narcotics and the Community: A System Simulation. Note that this is a simplified version for instruction purposes only and doesn’t reflect the intricacies of more recent opioid issues likes prescription opioids and fentanyl.
The logic of the model is straight-forward: with less illegal opioid supply, there will be less addiction and thus less addiction-related crime. It’s intuitive and easy to remember. Our minds are full of models like this, each fairly simple because the human brain isn’t able to hold more than a handful of variables in mind at the same time.
While it is hard to disagree with the premise of the simple cause and effect model, mistakes can happen when we assume the effects of our actions will be similarly simple and linear. For example, one common policy recommendation to reduce opioid addiction has been to curb the supply of illegal opioids. Let’s examine the consequences of this policy with a slightly more detailed model that captures its additional complexity.
Adding variables, or nodes
Let’s add three additional variables (or nodes) not in the simple model.
The number of opioid addicts isn’t one static number (a fixed quantity, or “stock” as it’s referred to in a stock and flow diagram). Rather, the number is determined by rates of addiction and attrition (or “flows”). So let’s add “addiction rate” and “attrition rate” to the map.
Because a reduced opioid supply would make the price of illegal opioids rise, let’s add “opioid price” as well.
Causal feedback loops
In the simple cause and effect model, price isn’t considered and any reduction in the supply deterministically reduces addiction and crime. But, adding the price of opioids to the model gives us a counterintuitive feedback loop: because addicts need more money to get the same amount of opioids, reducing the supply could actually make crime worse.
On the other hand, the addition of price to the model also mediates the rate at which people become addicted (represented by addiction and attrition rate nodes). When the price is high, there are fewer new users, which is good.
The point of the more complex model is not to show that reducing the illegal supply of opioids is a good or bad policy. Like all actions aimed at social problems, it is a trade-off between benefits and costs.
Rather, my intent is to show that overly simple mental models of a problem, even when logically correct, can lead us to make decisions that lead to unintended consequences. Better models don’t necessarily help us find the right solution, but they can supply needed insight about how the problem works systemically and give a sense of how our actions may cascade through the larger system.
So, if you don’t already have an explicit model of your problem or if you’re just working from a simple linear model in your mind, here are the steps you should take to create your own.
Problem mapping in four steps
Step 1. Brainstorm primary causes and concepts
Think of a problem and spend some time brainstorming all of its causes, including any other relevant concepts. For now, concepts can be any important variables in your problem: actors, stocks (e.g. quantities) and flows (e.g. rates), or even abstract concepts (e.g. wealth, democracy, etc.). Keep these tips in mind:
- Use nouns and avoid verbs, since actions will be represented in the map with arrows.
- Try to pick things that can go up or down in quantity, strength or influence over time.
- Be as specific as possible. When possible choose metrics over abstract concepts.
You might have a long list, and some causes and concepts may be more important that others. That’s OK.
For our purposes to get started, select the top 3-5 causes. These are the issues you believe are most fundamental in causing your problem.
My example below has only three primary causes for simplicity, but yours may have more. After you get through the steps there will be plenty of time for you to add or subtract variables.
Step 2. Brainstorm second-order causes
Most people think step 2 is about brainstorming solutions. But don’t do that yet! You don’t yet understand how the problem works, so solutions at this point will likely be similarly incomplete. To start getting a fuller picture of how your problem functions as part of a larger system, brainstorm second order causes. To put it simply, what causes each of your primary causes?
You can pull second-order causes from your initial brainstorming list, or brainstorm a new list for each primary cause.
Once you do that, your model may look something like this:
Step 3. Add interrelationships between causes
We’re getting closer to a comprehensive model – just two more steps to go!
In this step you operationalize the biggest insight about complex systems: the relationships are more important than the components. Right now, you still have a fairly simple, linear model. Every node leads directly to your problem, which makes it a kind of hierarchy.
I’m fairly certain that in real life your problem exists in a more complicated web of connections. In this step you connect causes that have a relationship to any other cause, even those that don’t lead directly to the problem.
For example, one of your second-order causes may be related to another second-order cause. That may sound confusing, but your task right now is just to draw connections between any two nodes that you feel might affect each other (see red lines in model below).
Step 4. Define causality of each relationship
The final step is to characterize each of the relationships between nodes as increasing or decreasing (for example, Cause #1a increases Cause #1, or Cause #1a decreases Cause #1).
A plus sign represents increase and a negative sign represents decrease.
Use this question to help you determine the direction of the relationship:
When this component increases, does the other component increase or decrease?
Note that sometimes nodes will have a two-way relationship. For example, in the model below an increase in Cause #1 increases Cause #2a, but an increase in Cause #2a decreases Cause #1. This is a balancing feedback loop.
The first draft of your model is complete. Who-hoo!
Three problem mapping principles
Here are a few words of guidance to keep in mind as you start modeling.
#1. Always model a problem, never a system
Problems themselves dictate the necessary boundaries that formal systems (like the education system) do not.
#2. The map is not the terrain
Your model is only an abstraction of reality and should always be regarded with a degree of skepticism and knowledge that the terrain may change.
#3. All models are wrong and incomplete
The purpose of using a model isn’t to find the solution, but to increase your understanding of the problem and explore the effects of possible interventions.
The whole process is most valuable when you remember that mapping is more of an art than science.
What to do after you’ve created a first draft
Share it with others for feedback
There’s a few directions you can go from here. The first is to talk through your model to a trusted colleague who also has some understanding of the problem, or encourage them to create their own model following these steps.
Differing perspectives can uncover different assumptions about the problem and lead to fruitful dialogue. Your model can be updated based on feedback, or you can work with your colleague to combine models, adding and subtracting nodes and relationships as you see fit.
Convert it into a fuzzy cognitive map
The other direction is to convert your model into a fuzzy cognitive map using computer software. This is super exciting because it allows you to run simulations of potential changes you could make and see resulting changes in the system as a whole.
For example, in the earlier opioid addiction model we could run a simulation of a policy that provides free legal supply of opioids to addicts (in an effort to reduce both crime and long-term addition), and calculate system-wide changes. This helps uncover feedback loops and potential unintentional consequences.
Creating a fuzzy cognitive map only requires a few additional steps. I go through them step-by-step plus how to run what-if scenarios in my next blog posts.
