IV. Adjacent Territory Essay 29
Fast Variables, Slow Variables
Why the thing you watch is not the thing that decides
There is a number on the wall in the workshop. It changes every day. It is the number of things that went out of the door, and it is up and down and all over the place, and when it goes up everybody is pleased and when it goes down everybody looks at it. Meanwhile, in the same room, a number that has not moved in three years is quietly deciding how many of those things there can ever be.
Almost every system has this shape. A fast part, which moves constantly, is easy to see, is easy to chart, and is mostly noise. And a slow part, which barely moves, is hard to see, is usually not charted at all, and is doing most of the deciding.
The most consequential variable in a system is very often the one that has not changed in a long time, and it is almost never the one you are watching. The habit of looking at the fast variable is not stupid. It is exactly what the fast variable is designed to do, because a fast variable produces data and a slow variable does not.
Take a small workshop that makes and sells furniture. The owner watches three numbers weekly. How many pieces went out. How many came in as enquiries. What the average sale price was. All three move. All three are on a chart. All three are discussed in the Friday meeting.
Now the slow variables. How long it takes to make a piece, which is set by the tools in the workshop and the size of the room and how good the people are. How many pieces the room can hold at once, which is set by the floor. Which pieces the workshop is able to make at all, which is set by the equipment and what the workshop has always made. And what people believe a piece from this workshop is worth, which is set by a decade of work and changes over about a decade.
Here is the interesting thing. The owner can double the number of pieces going out this month by working harder, and the chart will show it, and the meeting will be pleased. The slow variables will not have moved at all. The room is still the room. The tools are still the tools. The value of the work is still what it was. And what will actually have happened is that the workshop has spent three months of capacity on making more things, which means less time on the slow variables, which means the next year is worse rather than better, and the reason will be invisible because the number everyone watches went up.
There is a well known comparison between the weather and the climate, and it is worth using with care rather than borrowing lazily. The weather is the fast variable and the climate is the slow one, and the climate is not a different thing from the weather, it is what the weather has been doing for a long time, and you cannot see it in a single day's weather. The workshop is the same. The weekly number is the weather. The three years of not moving is the climate, and the climate is what the outcome is made of.
What makes a variable fast
Fast variables move quickly relative to the length of time you are watching. They are noisy, which means they move for reasons that have nothing to do with the thing you are trying to control, and they are legible, which means they can be counted and charted and put on a wall.
Slow variables move slowly relative to the length of time you are watching. They are quiet, which means they do not announce themselves when they change, and they are often unmeasured, which means nobody has a chart, and sometimes means nobody has a number at all.
There is a trap in the middle, and it is the one that catches people who have learned about slow variables and think they have solved something. A slow variable will move faster if you push on it, and pushing on it can feel like progress and produce nothing. The classic example is the water level in a bathtub with the tap running. You can adjust the tap. You can adjust it very precisely. The level is a slow variable from the point of view of the person in the bath, because the tap dominates it. But the level is not what the bath is for.
What the slow variable actually is
Here is the useful way to think about it. A slow variable is one that is upstream of the fast ones, and the fast ones are downstream of it. When you are looking hard at a fast variable and adjusting it, you are adjusting the output of the system, and the thing that is generating the output is somewhere behind you and is not in the room.
Which means the fast variable you are watching is partly a symptom of the slow variable you are not watching. A workshop's output this month is partly a function of its capacity and its reputation, and both of those are slow, and both of them were decided years ago by things that felt like small decisions at the time.
And this is why the fast variable can be so misleading. It is not noise. It is a real signal, of the slow variable, mixed with everything else that happened this week. It is a noisy reading of the thing you actually want to know. Taking the reading as if it were the thing is how a good year gets mistaken for a good decision.
What if the workshop measured the slow variables instead? It would measure how long a piece takes to make, and how many pieces the room can hold, and how long it takes to train somebody to the standard, and what people believe the work is worth. None of these move week to week. The charts would be almost flat, which is exactly what slow variables look like, and the charts would be nearly useless for deciding what to do on a Friday.
Now change the condition. What if the workshop measured both, and used the fast one for short term decisions and the slow one for the annual ones? This is the actual answer and it is unglamorous. The weekly meeting deals with output, and the yearly meeting deals with capacity and training and reputation. The mistake everybody makes is using the weekly number in the yearly meeting, because the weekly number is there and the yearly number is not.
"Fast variables do not matter." They matter enormously, and they are where most of the work of a week happens. The claim is only about leverage. A fast variable is a good thing to manage on a short timescale and a bad thing to build a strategy on, and the confusion between those two uses is the source of most of the wasted effort in organisations that measure things carefully.
The other misconception: "if the slow variable is what matters, just change the slow variable." You usually cannot, not quickly. A slow variable is slow precisely because it is held in place by things that are also slow. You change a slow variable by changing something upstream of it, over years, and by accepting that the results will not show up in the fast ones for a long time, which is a very hard thing to do to a person who is measured weekly.
Which one are you watching
There is a quick and quite reliable test. Ask what would have to be true for this number to be high, and then ask how long ago that would have had to be arranged.
If the answer is a few weeks back, the variable is fast. If the answer is a few years back, it is slow. If the answer involves a thing that has not changed in a long time, the fast variable is downstream of it and the leverage is in the slow thing.
And then the harder question, which is the one worth sitting with. Look at the list of numbers your work or your organisation actually watches. Look at how many of them can change tomorrow. Look at how many of them can only change if something you decided years ago was different. The ratio between those two numbers is the ratio between how much of your attention is busy and how much of it is aimed.
Limits
Fast and slow is a matter of timescale, not a property of a thing. A variable that is fast compared to a week is slow compared to a year, and a variable that is slow in a company is fast in a market, and a variable that is slow in a society is fast in a geological process. The classification only means something relative to the decision you are making, and saying a variable is slow without saying slow compared to what is not much of a claim.
Second limit: slow variables are not always the ones you should care about. A business where nothing ever changes is not a stable business, it is a dying one, and a system with no slow variables has nothing holding its shape together. The point is not that slow is good. The point is that the thing you can change is usually not the thing that decides.
Third limit: the distinction is easiest to see in systems with a clear timescale, like weather and climate or a workshop's output. It is much harder to see in systems where everything moves together, and hardest in systems involving people, where the timescale of a habit and the timescale of a career are tangled up in each other. The tool is more of a prompt for attention than a piece of analysis, and it is best used to make you ask the question rather than to answer it.
Stuck
You are stuck at the system level with a fast variable when you are managing the number and the number will not move. You have tried harder, you have set a target, you have put it on the wall, and the chart is flat or falling, and you have started to think that the measurement is broken.
The way out is to look upstream rather than harder. Ask what the number is a symptom of, and then ask what that is a symptom of, and keep going until you reach something that is not a symptom of anything, which will be a slow variable and will be a number you have never measured. Measure it. That is usually the end of the search, and it is usually somewhere uncomfortable, because the thing at the top of the chain is not something you can change this month.
Sources
- The distinction between variables that move quickly and variables that move slowly, and the point that the slow ones dominate the outcome, is the central idea of the systems thinking literature. No specific author, book, or date is cited here.
- The comparison with weather and climate is used in its ordinary sense as an illustration. No claim about climate science is made in this essay.
- All examples are ordinary and hypothetical, and no figures are given.