Open Question: What Does it Mean to Understand a System?

I recently began working on a new app. It’s one part design tool, one part programming environment, and lots more too. But at its core, it’s a medium for creating, thinking with, and understanding complex systems. Of those goals, understanding a system is probably the most important, but murkiest to me.

What does it mean to understand a system? Is it the same thing as “reading” a system? How do you go from not understanding to completely understanding one? What does the threshold look like?

That’s just for a single system, so how do you generalize these principles to all systems? What does it mean to be “fluent” with systems?

I happen to have a few ideas on how to answer these questions which I’ll post in the discussion section below, but I’m curious to hear your answers too. Please feel free to use examples, to link to papers and books I should look at, etc. I’m really curious how you think about this topic.

Join the Discussion 👂🤔✍️

Jason Brennan
What does it mean to understand a system? Some guesses:

  • Knowing what components make up a system, and what they each do.
  • Knowing the states the system can be in and how it flows from state to state (though I worry that the idea of a “system state” might be too unrealistic / clear cut: for example, what “state” is my circulatory system “in”? What “state” is US democracy “in”?)
  • Knowing the system’s goals, if it has any? (And knowing if it doesn’t have any)
  • Knowing the system’s limits. When does it work well? When does it break down? What can it recover from?
  • Being able to predict what a system will do with given input.
  • Knowing how the system interacts with other systems.
I’m also wondering, does understanding a system mean you can answer those questions (or others) in your head? If you have to use an apparatus to understand a system, do you really understand it? (I suspect needing to use media doesn’t disqualify you from understanding something, but it might be interesting to debate)

Some references that have informed my thinking (though I could use re-reading them):

  • Donella Meadows’s Thinking in Systems has an overview of Systems Thinking, including “stock and flow” notation and an overview of system anatomy.
  • Bret Victor’s Media for Thinking the Unthinkable provides a good overview of media required for understanding systems, such as seeing behaviour, creating powerful representations, etc.
  • Maryanne Wolf’s Proust and the Squid is about book literacy, how the brain learns to read, and an overview of what book fluency looks like.
I feel like systems are too complex and abstract a notion to be understood in any formal sense.  Therefore I take the Monte-Carlo approach.  Imagine a circle in a square.  The area in the circle is understanding the system and vice versa.  I pose a large quantity of questions about the system.  These questions are random dots placed in the square. Answerable one's land in the circle and vice versa.  If the number of dots that land in the circle is N times larger than those that land outside the circle I have "understood" the system.  Then I just have to find a reasonable metric for determining which N is a good threshold for..

wait nevermind..
Nicky Case
When it comes to complex/chaotic systems, I think PREDICTION is a fool's errand. It's like... if you knew you were in a thunderstorm-prone area, and you wanted to avoid getting hit by lightning, so you pull out a supercomputer to model the weather patterns and PREDICT where exactly the bolts will hit. Then try to avoid those areas.
A better approach than "predicting", I'd say, is PREPARING. Instead of doing the futile task above, just install a lightning rod. Or wear rubber boots. Or just don't go out in the rain.
Just because you can't predict "the" future, doesn't mean you can't prepare for possible future*s*!
I think THAT's what we should define as "understanding" a system -- not predicting specifics, maybe not even white-boxing the thing, but being able to practically act and work with/within the system to achieve our goals. Or something like that
Jason Brennan
Alex, that’s an interesting approach! How do you imagine this questioning goes? Is it person-to-person (say, mentee and mentor)? or do you see it more as person-to-system?

Nicky, that’s very thought provoking indeed! Are you suggesting “you understand a system well when you’re able to make useful preparations for interacting with it”? (or something like that?) In my head, I’m picturing this as literally having a handle on the system (that is, the system looks like a briefcase that you’ve put a handle on, and are now carrying; I may have taken the metaphor too far (metoophar?)).

Some examples of systems I’m interested in (or have been thinking about) are typical ones Alan Kay uses in Etoys, like:

I’m curious, what does it mean to be prepared for systems like those?
In lieu of a qualitative method of defining system understanding I used a quantitative one.  Like practice exercises at the end of a chapter in a textbook.  I'm not sure if there is a term for posing an array of pointed questions that test specific instantiations of the system.  Perhaps if you call those questions experiments, than that term is .. the scientific method, ha.
As for who to pose questions about a system to, I suppose one good candidate would be a computer.  I read a great article on Bret Victor's site about using dynamic data visualization to improve your intuition of some process.  It was called the ladder of abstraction.  I usually really on people like Bret Victor or Allen Kay to put these higher order concepts more eloquently into words.  When I do it I just sound like a pothead!
I'll be tripping over my words, but:

Understanding systems involves recognizing that there is a system in the first place, loosely mapping it, understanding that interactions can be loosey-goosey rather than strictly regimented. It's recognizing that there are patterns, triggers, responses, and other kinds of feedback loops. 

I think the harder challenge is how to discern/recogne that something is systemic in the first place, even more than the bits that comprise it.
Martijn Brekelmans
Let's begin with something I assume, but take as fact: systems do not exist outside the human mind, they are a construct that is helpful to us in some way.
In a similar manner, "understanding" something doesn't really exist outside of humans, animals can "understand" how to approach a given situation, but I don't believe they are able to do any kind of meta-analysis of their problem solving skills.
Because of this, I believe that understanding a thing means to know enough about a given thing so that it can help an individual reach her goals. Substitute thing with system, and my understanding of "system" becomes:
Understanding a system means to know enough about the system to reach a goal that is important to an individual or group of people.
Where the group of people = anything ranging from small teams to complete societies.
Everybody has different goals, to me, understanding for example a game -- which is a system -- means to find out about new interesting interactions and mechanics so that I can have fun.
My definition might be rather abstract, but I don't know how you could do with something less abstract, given the immense complexity of your question.

Please read the Discussion Guidelines before replying.

☑️ Email me when someone replies.

Speed of Light