The Unfinished Question: On Complexity
Do we even agree on what complexity actually means?
"The limits of my language mean the limits of my world."
— Ludwig Wittgenstein
I recently read an article that drew a distinction between problems that are complicated and problems that are complex. The author’s argument was thoughtful. A complicated problem, the author suggested, can be broken into parts. You can solve those parts individually and then reassemble the solution. A complex problem is different. The parts interact with one another. The relationships matter as much as the pieces themselves. Understanding the whole requires seeing how everything moves together in a shifting, unpredictable dance.
The author went on to argue that this moment in history is exposing the limits of a certain kind of thinking. For centuries, we have rewarded minds that work in neat, sequential lines—solving one piece at a time, moving step by step through a problem. But complex problems, the author suggested, call for something different: the ability to hold multiple streams in parallel, to see patterns across systems rather than only within them. I found that idea compelling. And yet, as I read, another question appeared—not quite a disagreement, more like a loose thread I couldn’t stop tugging:
Is complexity a property of the world, or a measure of our current understanding?
The question stayed with me for days.
One reason it stayed with me is that the experience of complexity seems strangely personal. Two people can look at the same Rubik’s Cube and experience it in entirely different ways. One sees a puzzle that yields to pattern and practice. The other can’t complete even one side and gives up. The cube itself has not changed. The moves are the same. The constraints are the same. What has changed is the solver’s relationship to it—their tools, their vocabulary, their ability to recognize structure. It suggests that when we call something complex, we may sometimes be naming not the problem itself, but the gap between the problem and our current understanding.
The more I thought about it, the more I realized I’ve been quietly asking some version of that question for most of my life. I've spent most of my life working with my hands—electrical systems, equipment, railway operations, home repairs, technical projects. The sort of challenges that look intimidating until you understand the problem or become clear about the goal.
The first time you change a spark plug, it feels complicated. There are tools to buy, specifications to learn, parts that can be damaged if you are careless. Once you’ve done it a few times, the mystery disappears. The procedure has not changed. Your understanding has.
The same thing happens with software. People often assume programming requires years of formal education. Certainly, becoming a professional software engineer requires deep knowledge of large systems. But building something useful often requires something simpler: learn the language necessary for the problem in front of you, learn the relevant tools, connect the pieces together. The system may still be vast, but it stops feeling unattainable.
That pattern appears everywhere. For centuries, flight seemed impossible. Human beings watched birds cross the sky and had no framework for understanding how it was done. Today, we manage thousands of planes in the sky at once, balancing wind, turbulence, fuel, and fixed runways—all held together by schedules and systems that turned the sky into a manageable grid. The world did not become simpler. We became better at describing and navigating it.
Perhaps some forms of complexity are genuinely irreducible. If so, the more interesting question is how often we assume we have reached that boundary when we have not.
What makes me suspicious is how often people surrender to complexity before they have truly engaged with the problem. “That’s above my pay grade,” we say. Sometimes that reflects appropriate humility. Expertise matters. I would rather have a trained surgeon perform an operation than a confident amateur.
But we may have expanded the boundary of expert-only problems further than necessary. You don’t need to be a mechanic to change a tire. You don’t need a computer science degree to automate a repetitive task. You don’t need medical school to learn basic first aid. In many cases, you only need enough understanding to solve the problem you have right now. Yet somewhere along the way, we confused legitimate respect for expertise with a surrender of our own curiosity.
Carol Dweck’s work on growth mindset suggests that people who believe they can improve tend to accomplish more than those who believe their abilities are fixed. That seems right. But over time, I found myself asking a different question—not whether effort matters, but where to direct it.
It is one thing to say, “I can get better at this.” It is another thing to know where to place your hands. Telling yourself a problem is not “above your pay grade” is a start, but it does not tell you which bolt to loosen, which menu to open, which concept to learn first. At some point, mindset has to become method.
That is where leverage enters. I remember struggling with a stubborn bolt. I pulled with all my strength and nothing happened. Then I slipped a length of pipe, a cheater bar, over the wrench. The bolt turned almost effortlessly. Of course, a longer lever introduces new risks. Too much torque, and something breaks. The tool doesn’t eliminate the variables; it changes the leverage you can apply to them.
Science is leverage. Mathematics is leverage. Engineering is leverage. Language itself is a form of leverage. The right concept can dissolve confusion. The right equation can replace thousands of trials. The right model can make an unstructured system legible. Understanding does not eliminate complexity so much as compress it into something we can work with.
Seen this way, AI is just the newest extension of that leverage. It expands what can be searched, compared, represented, and recombined. It can hold more variables in play, trace more relationships, and surface patterns that would be difficult for an individual mind to see.
Which raises a related question: what if some of what we call complexity is actually a limitation of language?
In an earlier essay in this series, I developed the asymptotic theory of truth. In part it is what I call the π factor—the unavoidable gap between the elegant human definitions we invent and the slippery reality they try to describe. We build tools, measurements, and formal systems that work remarkably well, but they remain constructs—always a little short of the world they aim to capture.
But leverage can also be misapplied. In our search for answers, too much precision can break the problem—while too little leaves us without a solution.
The history of science may not just be the history of discovering the world. It may also be the history of translating mystery into vocabulary. Calculus did not create motion, but it gave us a language for describing continuous change with extraordinary precision. Electrical theory did not create electricity, but it gave us units, concepts, and conventions—volts, amps, current, resistance—that made a once-mysterious force available for design and control. Scientific progress often begins when something that was merely experienced becomes something that can be named, modeled, and taught.
That possibility extends even to the language we use to describe thinking itself. Terms such as linear thinking, systems thinking, sequential processing, and parallel processing are helpful today, but they are still models rather than reality. Like any vocabulary, they help us see certain patterns while obscuring others. These are also human constructs. We’ve taken an analog world and broken it into discrete, manageable pieces—a move that says more about our tools and thinking than about reality. Future generations may see our current language of thinking the same way—useful for their time, but not the final word.
That matters because language does more than describe reality. It changes what we are able to notice. It changes what can be compared, measured, and reasoned about. Once a field develops a working vocabulary, what once felt like an opaque whole often starts to reveal joints, patterns, and regularities. A complex challenge becomes discussable. Then teachable. Then, sometimes, routine.
And because these are human tools, they have limits. We refine our building codes, our specifications, and our models because reality keeps reminding us where the tolerances are. We do not treat those failures as proof that the world cannot be understood; we treat them as data for the next revision—refining the language, adjusting the models, and testing them against reality once again.
That makes me wonder whether some of today’s complex problems are not final mysteries, but places where our current vocabulary has not yet caught up with the world it is trying to describe.
I don't know which challenges, if any, are truly irreducible—resistant to simplification beyond a certain point. I am increasingly skeptical of claims that something is simply too complex, especially when those claims arrive early. If complexity is often a measure of the current gap between reality on one side and our understanding and tools on the other, then today's most urgent problems are likely more approachable than they first appear.
And history suggests that gap has a habit of shrinking in ways that once would have looked impossible even to describe, let alone solve.
More from The Unfinished Question:
Collaborations ✨
On Belonging with C Simone
On Enough with Marya Kazmi
On Learning with C Simone
On Time with Rebecca Mbaya
Solo Reflections
Wow! Great article! The question of whether something is actually complex, or whether we just don’t have the right language or tools for it yet, really got me thinking. I also loved what you said about not confusing respect for expertise with giving up our own curiosity. Some things absolutely need experts, of course, but I know I’ve talked myself out of trying things before just because they seemed too far over my head. Well done!
Now this article got me thinking. I think your point about language is right. What I have heard in response to the systems, thinking article you read, is that they have language for something they couldn’t articulate before. My own view is once we have language for something we can incorporate it into our self-concept. Once that happens, I am now wondering if there is less that feels “above our paygrade.” I had thought of getting language for something that we hadn’t yet expressed in developmental terms of, but hadn’t really thought about it in terms of increased capacity for relating to complexity. Thanks for this!