Script
I really like Eli Goldratt’s philosophy, the Theory of Constraints. What’s it about?
Goldratt’s goal was to teach the world to think. He studied physics and thought the methods of the hard sciences should be used in other fields too. He developed ideas about thinking methods and applied them to business management. He thought businesses could get huge benefits from more rational, rigorous, organized thinking. He was a successful business consultant and sold millions of books.
Theory of Constraints, or TOC for short, says we need to focus our attention and efforts on the most important issues. To do this, look for constraints, a.k.a. bottlenecks, which are parts of a system that limit its throughput or success at your goals. Most factors in a system have excess capacity – they aren’t the weakest link – so improving them won’t result in better system performance. Find the factors that are limiting your results and focus on improving those.
Don’t optimize local optima. That means avoid improving individual parts of a system that won’t lead to better results at your goals. Most parts of a functional system are already good enough.
TOC also has ideas about problem solving, conflict resolution, contradictions, simplicity and much more. I’ve found TOC’s ideas really helpful for creating a deeper understanding of epistemology. Goldratt emphasized a lot of concepts that Karl Popper didn’t write about, and his ideas are mostly compatible with Popper’s, so the two thinkers complement each other well.
I wrote an essay covering around twenty TOC ideas. This video will discuss a few key ideas from the essay.
The Goal
Goldratt’s most popular book, The Goal, is a novel about the manager of a struggling factory. He improves the factory and illustrates TOC thinking methods and concepts.
The book emphasizes figuring out what your goal is, then figuring out what limiting factors are reducing your success at your goal. Most potential improvements won’t actually help with your goal much because they don’t address limiting factors, a.k.a. constraints or bottlenecks. Related to this, he explains why equal capacity for each workstation is the wrong way to design a factory.
Constraints
Marris Consulting focuses on TOC. Their logo illustrates a constraint with water flowing down through tanks.
Our goal is to get water to the bottom. The middle spout is the constraining factor for water flow. The other spouts have excess capacity, so making them bigger wouldn’t increase the output of the whole system.
If you look at each spout individually, making it bigger seems useful. But that’s the trap of local optima. Looking at the bigger picture, we see that improving some spout sizes won’t help with our goal.
Focusing Steps
TOC’s five focusing steps are a method for making improvements.
First, find the constraint. Then optimize it. Stay focused and don’t optimize other parts of the system.
If you need further improvement, organize the whole system around the constraint. Change other parts in ways that make things easier for the constraint. Doing two hours of work somewhere else to save an hour at the constraint can be efficient.
To improve further, increase the capacity of the constraint. Notably, don’t do this earlier. It’s usually more efficient to make other changes first.
After adding resources to the constraint, check if it’s still the constraint. If the constraint moved, return to step one.
Changing Steps
Changing can be a lot of work. Especially when a lot of people or money is involved, we want to make the right change on our first try.
Goldratt’s method begins with considering what to change. You want to pinpoint core problems related to constraints. You can use the effect-cause-effect method explained in my essay.
Second, consider what to change to. Look for simple, practical solutions. You can use Goldratt’s evaporating clouds method from my essay.
Third, changes need to be done in the right way. One good approach is using the Socratic method. You can ask questions to help people figure out the change themselves. If you just tell people your answers, they often won’t create enough understanding to implement the change well.
Balanced Factories
A balanced factory design tries to avoid excess capacity in order to be efficient. And it tries to avoid having a bottleneck or weakest link by giving every part of the factory just the right amount of capacity.
Goldratt’s insight is that balanced factories are inefficient. The main reason is statistical fluctuations. Production won’t actually proceed in an orderly, balanced way due to random variations in output. A worker, who averages producing ten widgets per hour, will sometimes produce zero, five, fifteen or twenty.
Also, the costs for different kinds of production capacity vary. We don’t need to carefully minimize the amount of capacity for cheap stuff. We should focus our attention and optimization on the expensive types of capacity.
Statistical Fluctuations
When you have a chain of dependencies, and statistical fluctuations, bad luck propagates down the whole chain, and good luck doesn’t cancel out prior bad luck. This is a mathematical fact that can be simulated on computers. It can also be understood with the simple matchstick game from The Goal.
What’s the solution?
A factory’s constraint should have a buffer in front of it with extra input parts so it never has to waste time waiting when there’s a delay at a workstation that comes earlier in production.
And each workstation besides the constraint should have excess capacity.
Workstations before the constraint need to work faster than the constraint to catch up and refill the buffer when bad luck depletes it. That means they have unequal capacity compared to the constraint. When the buffer is full, they should work slower or halt work part of the time. Although it may be counter-intuitive, that’s efficient.
Workstations after the constraint need excess capacity so they can keep up when there’s good luck at the constraint. That also means they won’t operate at maximum capacity most of the time.
Conclusion
The world is complex and you can read TOC books for more details. But, generally speaking, real situations involve a small number of key factors and a large number of other factors that have excess capacity and don’t need to be improved. A system where a large number of factors all needed lots of attention and optimization would actually be unmanageable. To be effective, we need to focus our attention on a few things.
My essay has more TOC ideas that this video didn’t cover. The link is below.