Of Slinkys And Bathtubs

Of Slinkys and Bathtubs

You think that because you understand "one" that you must therefore understand "two" because one and one make two. But you forget that you must also understand "and".
            -- Sufi teaching story

Have you ever been hit by one of those ideas that just changed the world? I don't mean an "interesting insight". I mean one of those things that so profoundly shifted the way you look at things that you never go back. I want to share an idea with you today that did that for me. It seems a simple thing, at first. The best ones always do. But, follow the thread all the way to the end. Its implications are powerful stuff.

Very early in my career, my first mentor handed me a copy of a book. He smiled his subversive smile and said "Read this." Jack Anderson, my mentor, was a particularly easy going fellow. A recommendation like "read this" was akin to walking down a mountain with a pair of tablets. The book was "An Introduction To General Systems Thinking" by Gerald Weinberg.

The Slinky

We start with a slinky. As a kid, this was just one of my all time favorite toys. I can still idle away hours just sloshing it back and forth from one hand to another. Now, we could break this slinky down and talk all about Hook's Law and spring tensions and such. But, we don't need to do that. It's a just a slinky.

Take your slinky and hold the top half of it still with one hand. Put your other hand, I'll call it the left, under the slinky. Move your left hand. The slinky will bob and bounce up and down for quite a while. Fun game. What made the slinky bounce like that? The temptation is to say it was moving your hand. But, try the same experiment with the box the slinky came in, and you'll see your hand really had nothing to do with it. The bounce came from a property which intrinsic to the slinky. Your hand was an external force which served to either inhibit (when it was there) or allow (when you moved it) the expression of an innate behavior of the slinky.

Slinkys are like Tiggers. They are always bouncy. But, sometimes, they are between bounces.

The Bathtub

Let's take another example: a bathtub.

Imagine a bathtub. Any bathtub will do; I like the big claw foot monsters like my old house had. To complete this experiment properly, we may want to imagine an ample supply of towels as well. Fill the bathtub about half way up. Now, leave the water running and pop the plug out of the bottom of the bathtub.

One of three things will happen next. If the drain is taking water out faster than the spigot is putting it in, the tub will eventually empty. If the water is coming in faster than it is going out, it will eventually overflow. Or, if we fiddle with the knobs very carefully, we can reach a state called dynamic equilibrium where the level of the water stays just the same, even though the water itself is constantly moving through the tub.

The Big Insight

Our bathtub and our slinky are systems. Each has parts – the spigot, the drain, and the basin, for instance. Each of the parts interacts with each others in particular ways (the spigot puts water into the basin, the drain takes it out). And, the net behavior of the system depends on all of the parts, and the individual interactions. Studying just the spigot won't explain the behavior of the systemanymore than studying only the drain will. You have to understand the system, as a system, in order to understand it at all.

But, looking at the big picture, there are some interesting behaviors that we see.

Let's think about when the tub overflows. We can get towels and a mop and even a ShopVac to clean it up. But, the water on the floor isn't the problem. It's just a symptom. (If all we do is clean up the water, we'll always be cleaning up the water because we haven't done anything about where the water is coming from.) The real problem, the thing we need to fix, happened when the dishwasher downstairs shut off. This removed a drain on the house's water pressure, allowing more water to flow to the tub. Now, our perfect equilibrium setting is delivering just a bit too much water. We didn't see the symptom then, because the water level was just half way at that point. One important aspect of systems like this is that the symptoms, the problems, may not show up until long after the occurrence of their root cause. And, that root cause may fall in one of the limiting factors, which is itself a side effect of a completely different system.

To fix this overflow, we can turn the spigot down. But, even if we turn it off entirely, the tub will not immediately be empty. It will take some time to drain. We can't dial the spigot to its perfect equilibrium level at this point – or our equilibrium point will be just at the spilling over stage. We have to dial it down to well below optimum, let it drain, then ease it back up gradually. Another important aspect of systems is that they have inertia. For this reason, you can't just set a value and be done. They require care and tuning and either constant adjustment or internal feedback mechanisms to deal with changing conditions.

Let's review for a moment. Inputs, outputs, queues in between, systems with inertia, constant tuning, problems that show up long after their root cause has come and gone, issues where the root cause of a problem may not lie in the machine or software that exhibits the problem at all…..does any of this sound familiar? From where I'm sitting, systems thinking is the very heart and soul of performance work.

Further Reading

Books, theses, dissertations…whole libraries have been written on the topic of systems thinking and analyzing things from a systems point of view. Many of them are particularly dense works. But, I've found two to be decent and interesting reads. One is the original Weinberg book, above. But, even its silver anniversary reprint edition can be a little hard to find sometimes. The other, the one I'm reading now, is "Thinking In Systems" by Donella H. Meadows. Using simple things like bathtubs and used car lots, she communicates the important points along with some effective tools for dealing with them.

I can, and at some point probably will, carry on for hours about some of the tools and techniques that come with the systems approach. But, for now, I'll just give time for the idea to take root. Let me know what you think.