How to Think With Systems: The Importance of Definitions
The first step in solving a problem is to define the items and processes that you deem important. You don't want to define everything ... the goal is to reduce the complexity of the problem by keeping the key elements and dropping elements that don't matter as much. This is not an easy task. Effective problem definition is a skill that takes practice to develop.
This section is repeated from an earlier tutorial ... because it is vitally important.
To create a system model you have to make choices about what to include in the system and what to exclude to the environment. And not surprisingly, there is a human tendency to include things that support the outcome you want and to exclude things that run contrary to your preferences. If you're not careful, the result can be a system that models your fantasies rather than reality.
Successful systems practitioners learn very quickly that wishful thinking is not a very good method. However much you want a particular outcome, or fear a particular outcome, you simply must exclude those feelings from the model building part of the process. The right approach is to build accurate models that are flexible enough to accommodate multiple outcomes ... and then to criticize and appreciate those outcomes.
For the same basic reason, systems models cannot be compared to one another. A systems model is the result of a careful crafting of elements, attributes and relationships designed to fit certain data to a particular situation. Two different people will model the same data differently depending on their particular problem and preferences.
It's worth repeating that the finished model is only the result of the modeling process and not worth much outside of the particular situation that's being modeled. It's the insights that the model enables that matter.
Here's How It Works
Let's say that I am making a pot of spaghetti sauce to go with dinner tonight. I have a recipe, handed down from my grandmother, whose ingredients list goes like this:
- a large can of crushed or diced tomatoes
- 2 T olive oil
- 1 medium onion, diced
- 3-5 cloves garlic, minced
- 1/4 cup chopped herbs: basil, parsely, oregano, thyme, hot pepper
- 1/2 cup red wine
- 1 T sugar
- Salt and pepper to taste
Having cut, chopped and mixed the first seven ingredients and started them cooking, I go to add the salt and pepper ... and discover that I am out of salt.
Now what?
First of all, I do NOT sit down and draw up a systems model and slowly work my way up to a solution ... nonetheless, this IS a system. The ingredients of the sauce interact with one another within the pot to produce the unique flavor that I want. So I cannot just leave out the salt.
I chose this example to emphasize the importance of definitions. I don't have any salt ... so one option would be to go and get some. I can turn off the burner, get in the car, drive to the store, buy some salt, and return.
Or ... I can change the definition of salt. Instead of absolutely requiring the white crystalline substance that I usually use, I can ask the question: do I have anything else on hand that tastes salty? Ah ha! I have a bottle of soy sauce in the refrigerator. Soy sauce is very salty ... and I know from prior use that its taste is compatible with tomatoes.
So by changing the definition of the recipe item from specifically "salt" to "something that tastes salty", I alter the way I think about the problem, and it goes away.
Here's Why It Works
Our brains are used to doing normal things in normal ways. When something unusual comes along, it creates a "problem" ... a situation in which the normal doesn't work any more.
But our brains are slow in this regard. They like normal. They want normal. And they keep thinking as if everything were normal even after it has stopped being that way.
Systems thinking works in large part because the methodical construction of a model, and the continual testing of the model, forces you to stop making assumptions about what "normally" works. That's really the goal. There's no magic involved, and no tricks or advanced theories ... just a way of making yourself think clearly about the important details of the problem.
Making Definitions
It sounds easier than it is in practice. If simply changing the definitions will solve the problem, then why doesn't everybody just do that? Because just randomly changing all the definitions doesn't get you anywhere. In fact it's not the changing of the definitions that's the key ... it's the thinking that you do as you figure out which definitions are important enough to include in your model.
The model will always turn out to have three basic components: things, relationships, and contexts. All of these will need to be defined.
Things (a.k.a. items, elements, objects, nodes)
I am using the vague term "things" on purpose. The system is a network of things and the relationships that occur when they linked up with one another. The things can be objects or ideas or processes ... or anything that you define them to be. They don't have to be real, and you don't have to actually be able to touch them. You create them to help you understand and think about the problem.
Relationships (a.k.a links, communication, connections)
The things don't just sit there. They relate to one another. The relationship may be physical, as with two items that bump up against one another. Or the relationship may be visual, as with two people who see one another across a room. Or the relationship may be virtual, as with two computers connected by the internet. The relationships are what produce the behavior of the system ... they are what shows you how it works.
Contexts (a.k.a inputs, outputs, sinks, sources, the environment)
Since your goal is to take a complicated situation and simplify it so that you can think about it more clearly, you can't put everything into the model. In fact you want to put as little as possible into the model – just the key items: the stuff you need to understand the problem. Everything else goes into the environment and is summarized as a collection of inputs and outputs to the system model.
For example, in the tomato sauce problem I did not include the stove burner in the model. I put it into the environment where the heat it provides can be defined as an input to the system. Similarly, the steam that the boiling sauce gives off as it cooks would not be included in this model but would be an output from the system to the environment.
But ... and read this carefully ... suppose I had a slightly different problem. Suppose that I had all the ingredients, but that in earlier attempts to cook the tomato sauce, it had burned. In this case the steam would be a key element of the problem and would have to be included in the system. I would need to put a lid on the pot to keep the steam in, or make an arrangement to add liquid during to cooking to replace the steam that escaped. The definitions depend on what you are trying to do ... never forget that.
Learning to Observe
If you grew up in an industrial or "high tech" culture, then you are probably not very good at observation. That may sound like a critically negative thing to say, but if you stop to think about it, you will see that it is a natural result of our way of life.
Observation is the art of paying very close attention to what is happening around you, noticing all the details, and spending time thinking about how they all fit together. Modern and post-modern cultures ... I will just call these "current cultures" from now on ... do not encourage these activities.
For one thing, current cultures contain a lot of "packages" ... useful items whose working parts are hidden from view. You probably own a television, for example. But the odds are you have never looked inside of it. Nor are you likely to have any detailed knowledge about how it works. The same holds true for your iPod, camera, cell phone, and on and on.
This packaging of items is a fairly recent phenomenon, but it is now widespread. In the 1950s, to pick one example, any reasonably competent person could take an automobile apart and investigate it. Cars in those days were completely mechanical. Today, however, a typical automobile contains upwards of fifty computer processors. They monitor and control the engine, the brakes, the climate controls, cruise control, and other parts of the car. Even if you could take the car apart ... and that is difficult to do because some part are manufactured as plug-in units that cannot be easily disassembled ... you would not be able to understand what the parts do without a thorough knowledge of computers.
A second feature of current culture also discourages observation. The proliferation of mass media products teaches us to expect data to arrive in a preprocessed, ready to use format. Most of us spend the bulk of our time doing two things ... talking to other people, and interacting with mediated communications devices. (And frequently, we now do both of those at the same time.) While this is highly efficient and very useful, it discourages the practice of looking closely at our environment and thinking about how it works.
In fact the task of doing observation has mostly been handed off to experts. Scientists of all kinds observe the natural world and try to understand the details of how it works. Social scientists do the same with human behavior. Computer scientists study abstract patterns of logic and try to replicate these as software.
Current culture has more experts than any previous human civilization, but if you have not gone to school to study specific techniques of observation, you probably are not very good at it. I say "probably" because observation is mostly a matter of curiosity, paying close attention to details, and thinking about what you see. Anyone can do it, and some people learn it on their own. Which means that you can learn it, too, no matter what your background.
If you go to a bookstore site on the web and type "observation skills", you will find a collection of books that mostly relate to specific fields of study: police investigation, science, and teaching, for example, all require good observation skills. The best book in terms of describing systems observation that I know of is: An Introduction to General Systems Thinking by Gerald M. Weinberg ... but there are not many to choose from. Observation is just not a basic skill in our culture.
Using Definitions
Your first set of definitions is usually just that ... a first try that you expect to update and improve as you build and test the model.
The definitions are inside the box labeled "build model". You can see how it will work - you build a model by defining and organizing the data that you have collected. Then you criticize the model and update it, then you test the decisions it gives you and possibly update it some more. Adding, deleting and revising decisions is part of this updating process.
Examples
The following examples illustrate the contents of this tutorial:
Example 1. College Life
Example 2. Fossil Fuels