The Six Laws of Root Cause Analysis

What are the fundamental principles behind successful root cause analysis of difficult large-scale social problems? Our research found nothing close to an answer to that question, so we have provided one in the form of six laws.

More than six laws could be included. But we preferred to keep the laws as focused and simple as possible. These six seem to be the minimum needed to correctly analyze difficult large-scale social problems with root cause analysis. The six laws also provide the scientific rationale behind Social Force Diagrams and the System Improvement Process.

The laws cannot be understood without reference to the Standard Social Force Diagram, shown below. Note the forces R, S, and F.

Here's an explanation for each law.

1. All causal problems arise from their root causes.

A causal problem occurs when problem symptoms have causes, such as illness or a car that won’t start. All social problems are causal problems. Example of non-causal problems are math problems, scientific discovery problems, information search/organization problems like criminal investigation, and puzzle solving.

We know from Newton's Third Law that for every action there is an equal and opposite reaction. All effects have a cause and all causes have an effect. Problem symptoms are an effect. Therefore, all causal problems arise from their root causes.

This law is so fundamental it's The Law of Root Causes. The remaining laws build on this one. If you are learning root cause analysis, this law is where your new mental model must start.

2. Superficial solutions fail because S < R.

The reason they fail is force S is directed at an intermediate cause. Force S is always less than force R, because root causes exert much more force on intermediate causes than superficial solutions ever can. This explains why low leverage points exist.

Examine the Standard Social Force Diagram above. Note forces S and R. The equation S < R means that S is always less than R.

A Root Cause Force cannot be changed significantly without changing what's at the bottom of the causal chain, the root cause. S cannot change R because it doesn't push on the root cause. That's the way physics works.

For example, consider the Authoritarian Ruler Problem, diagrammed below.

The superficial solution of revolutions didn’t work because it was a weaker force than the root cause force. Revolutions are to be avoided at all costs, because they are so destructive and unpredictable. Even if they work, most of the time the new good ruler goes bad or is eventually succeeded by another bad ruler. Superficial solutions don’t work because S < R.

This is a HUGE insight. If solutions to a difficult social problem are failing, it’s usually because S < R, or less often because the problem is insolvable, as presently defined.

3. Fundamental solutions can succeed because they can be designed such that F > R.

The reason they can succeed is because force F is directed at a root cause. This explains why high leverage points exist.

Continuing our example, this law explains why the invention of modern democracy worked. The force created by the Voter Feedback Loop was so much greater than the root cause force that it tipped the system into a mode change and solved the problem, because F > R.

The key to understanding what F > R means is to grasp how changes to a system's structure allow mode changes. R represents the force of the current feedback loop structure that contains the root causes. The high leverage point represents where in that structure a change can be made without the system resisting the change successfully, i.e. a point where F can be greater than R. The fundamental solution is the actual change mechanism that changes the structure, at the location of the high leverage point, by exploiting the fact that F > R.

This is exciting stuff. We’ve boiled all of root cause analysis down to two simple equations: S < R and F > R.

4. If analysis shows no F > R exists, the problem is insolvable.

When this situation is encountered the problem should be redefined such that at least one F > R exists, and analysis should start over with the new equation(s) in mind. Or solution should not be attempted and the problem declared insolvable. But now you know exactly why it cannot be solved and will not waste any more effort on solving it.

How do we know if a tough social problem is solvable or not? If we follow the causal chains down from the problem's symptoms, and we cannot find a case where F > R, then the problem is unsolvable.

Returning to our example, for thousands of years, it looked like Authoritarian Ruler Problem was unsolvable. Revolutions didn’t work for long. It didn’t take long for a good ruler to go bad, or for bad rulers to get back into power, and then another revolution was needed. That’s why there were so many revolutions, and uprisings, and assassinations, and coups! The problem looked insolvable because activists couldn’t find a solution where F > R. But in retrospect we know the problem was solvable. It’s all a matter of applying the right tools.

What happens if you’re working on a problem that you’ve really got to solve, and discover that no F > R exists? You redefine the problem by relaxing the problem definition. For example, if NASA discovered they couldn’t put a man on the moon in ten years, they could have switched to twenty years.

5. Difficult large-scale social problems have multiple root causes.

This is why the System Improvement Process decomposes the one big problem into smaller subproblems, each of which has one or more root causes. Without this decomposition correct analysis is impossible. Analysis uses one Social Force Diagram per subproblem.

Consider climate change. We know what we should do. We should get atmospheric CO2 down to below 350 parts per million. But we’re not doing it, which is change resistance.

All difficult social problems contain at least three subproblems, the original problem, change resistance, and solution model drift. One reason they're difficult is because change resistance is high. If it was low the problem would be relatively easy to solve. The other reason they're difficult is the system should be managing them well but is not, due to solution model drift.

Each subproblem has one or more root causes. Therefore all difficult social problems have multiple root causes.

The System Improvement Process incorporates this law by providing the standard three subproblems found in all difficult large-scale social problems.

6. Due to lock-in, difficult systemic problems can be solved only by correctly engineered mode changes.

Lock-in occurs due to the unrelenting strength of a system’s dominant feedback loops. The desired mode change requires reengineering the system’s structure such that when force F is applied, a new force R is created, the mode change occurs, and the system’s current dominant feedback loops are replaced by new ones.

Feedback loops are the most powerful forces in the social universe. If you don’t understand a problem’s dominant feedback loops, then you don’t understand the problem and will be unable to solve it. Understanding dominant feedback loops begins with understanding mode lock-in.

Mode lock-in occurs when a system becomes locked into an overall pattern of behavior for a period of time. Small impacts on the system will not knock it out of that mode due to the presence of strong balancing feedback loops. Only radical impacts like invention of the radical new technology of agriculture can do that. Mode lock-in is usually good because it provides stability to a system. However, once a system slips into an undesirable mode it can be surprisingly difficult to snap the system into a desirable mode.

Difficult social problems are invariably systemic. This causes strong mode lock-in due to the way the system's feedback loops are structured. Thus the only way to solve a difficult social problem is to reengineer its present structure, which consists mostly of its dominant feedback loops. Correct reengineering will cause a mode change. The reengineered system will have a different feedback loop structure with inherently different behavior, hopefully the behavior you want!

Discussion

These six laws are what’s behind the powerful simplicity of Social Force Diagrams and the System Improvement Process. The laws let you cut through the mind-boggling complexity of social problems by focusing on what matters and ignoring the rest. Or as prognosticator extraordinaire Nate Silver would put it, the laws let you avoid mistaking the noise for a signal.

Here's what Nate wrote in The Signal and the Noise on page 196, while reviewing predictions of the Great Recession of 2008: (Bolding added)

"And this is exactly how the financial crisis played out. Not only was Hatzius' forecast correct, but it was also right for the right reasons, explaining the causes of the collapse and anticipating the effects. Hatzius refers to this chain of cause and effect as a 'story.'

"In contrast, if you just look at the economy as a series of variables and equations without any underlying structure, you are almost certain to mistake noise for a signal and may delude yourself (and gullible investors) into thinking you are making good forecasts when you are not."

Let's think like Nate:

Solutions based on the Six Laws of Root Cause Analysis are forecasts about how a system will behave in the future after the solution is applied. To make an accurate forecast, one must grasp the full chain of cause and effect involved, and deeply understand it as a "story" of how the structure of the system works. Otherwise you will mistake noise for a signal, and will delude yourself (and gullible fellow activists) in thinking you have a good forecast, a solution that will work, when in fact you do not.

As far as we can tell, making accurate forecasts about how a solution will affect a social system's behavior requires something much like The Six Laws of Root Cause Analysis.