Engineering for a World That Won't Sit Still: Notes on Guru Madhavan's Wicked Problems

What stayed with me first from Guru Madhavan's Wicked Problems wasn't the engineering, even though the book is mostly about engineering. It was a painting.

Madhavan is a Senior Scholar at the National Academy of Engineering, and he opens the book by sorting problems into four kinds. Hard problems are bounded, which means you can put a fence around them and engineer them closed. The smartphone is a hard problem, and the lunar landing was a hard problem. Soft problems are about people, which makes them political and psychological, and the variables you're trying to measure tend to move the moment you start measuring them. Messy problems are the ones where the stakeholders don't even agree on what the problem is, let alone what should be done about it. And wicked problems are what happens when all three of those — hard, soft, and messy — show up in the same room at the same time, each one making its own demands.

Then Madhavan hands you a painting.

Hard problems are photographs — clarity and directness. Soft problems are the blurry brushstrokes of impressionism. Messy problems are spilled and splattered abstractions. A wicked problem is the cubist collage where the truth is simultaneously sharp, shaky, and squiggly.

That one paragraph changed how I look at every project I'm working on right now. AI deployment in healthcare turns out to be a cubist collage. Generative-AI policy inside a company turns out to be a cubist collage. Even my own role at The Thinking Spot is a cubist collage when I look at it honestly — partly the hard work of point-of-sale systems and inventory, partly the soft work of knowing the regulars and feeling the rhythm of an independent bookstore, and partly the messy work of figuring out what a STEM bookstore is even for in 2026. Once Madhavan put the painting in my head, I couldn't stop seeing it in everything I was already trying to build.

The synergistic six

Madhavan's engineering response to wicked problems is built around six attributes that have to be tuned together rather than chosen between. The six are efficiency, vagueness, vulnerability, safety, maintenance, and resilience. He calls them synergistic because pushing too hard on any one of them tends to cost you in the others. If you maximize for efficiency, you end up brittle and lose your resilience. If you try to eliminate vagueness too early, you over-specify the system and miss what is actually going on around it. If you treat safety as something separate from maintenance, the system degrades silently between audits and you don't catch it until something has already broken.

The two attributes that puzzled me at first were vagueness and vulnerability, because it wasn't obvious how either of those could be on the engineer's side of the table. The honest answer is that they aren't on anyone's side. They are conditions of the real world that the engineer has to design with, not virtues to chase and not flaws to engineer away. Wicked problems arrive vague, and they arrive vulnerable, and the actual engineering skill is to name those conditions out loud instead of pretending the world is going to be cleaner than it is.

I had a hard time remembering all six attributes when I first read the book, so I made myself a mnemonic.

Efficient Safety Maintains Vague Vulnerability's Resilience.

The sentence is six words long, and each word maps directly onto one of Madhavan's six principles. By accident, it also happens to state something true about Madhavan's underlying argument, which is that the engineering disciplines of efficient safety and good maintenance are what give a system its resilience in the face of the vague vulnerability that wicked problems always carry with them.

The flight simulator that became the digital twin

The case study at the heart of the book is Edwin A. Link's flight simulator. In 1929, in his father's piano-and-organ factory in Binghamton, New York, Link built the first practical training simulator out of organ parts. The Link Trainer let pilots learn instrument flying on the ground, which made it cheaper than putting them in real airplanes, safer than letting them figure it out in the air, and repeatable in a way that real flying never could be. During the Second World War, the Link Trainer was mass-produced and ended up training more than half a million American pilots.

I read this case study and something clicked for me, because I work in applied AI for a living, and the Link Trainer is the direct conceptual ancestor of the modern digital twin. The lineage runs cleanly from Link's physical mockup in 1929, through the analog flight simulators of the 1930s and 1940s, into the computer simulators of the second half of the twentieth century, into digital twins in the 2000s, and now into AI-augmented digital twins in the 2020s. Each step in that chain preserves the same engineering move that Edwin Link made first, which is to build a safe place where someone can practice and fail before the stakes go up in the real world.

That move, carried across nearly a century of changing technology, turns out to be one of the most under-rated shapes of AI deployment available to us right now. The popular framing of AI is that the model makes the decision for you, but Madhavan's lineage suggests a much more useful framing. The right move with AI in a high-stakes setting is to let the model build a parallel-world rehearsal of the decision so that you can see how it plays out before you commit to it in the real one.

The deeper claim

The argument Madhavan is making, underneath the taxonomy and the synergistic six, is that engineering is the right way to engage civic problems. He is not saying that engineering can solve them, because he is explicit that solving is the wrong frame for a wicked problem in the first place. What he is saying is that engineering brings the right set of disciplines to wickedness. There is the discipline of decomposing a problem into parts that you can actually work on. There is the discipline of building simulators so that you can fail safely. There is the discipline of iterating instead of trying to get the whole thing right the first time. And there is the discipline of holding two opposite thoughts at the same time without collapsing into one of them.

What Madhavan explicitly rejects is the seductive intellectual move where you read about wicked problems, get the diagnosis right, and then conclude that "well, you can't really solve a wicked problem, can you." That conclusion sounds sophisticated when you say it out loud, but in practice it is a permission slip for disengaging from the problem entirely. Madhavan's third path is to engage the wickedness with the disciplines of systems engineering, to recognize that solving is the wrong word for what you are doing, and to instead work patiently to make the problem more workable over time. That third path is the most operator-useful thing in the book.

There is also a move in the book that I didn't expect, which is that Madhavan reframes maintenance — usually the least glamorous of the six attributes — as tikkun olam, the Jewish ethical concept of "repair of the world." He brings sanitation crews and sewer upkeep, road resurfacing and bridge inspection, all of the unglamorous and mostly invisible work that keeps a civilization standing, into the same conceptual space as engineering's most celebrated achievements. Most engineering education over-indexes on new design and treats maintenance as a footnote, and Madhavan's argument is that getting maintenance right is at least as important as anything that gets celebrated in the front of the book. That is the chapter that will stay with me the longest.

Recommendation

If you work on anything where the problem keeps shifting shape on you — AI deployment in your business, healthcare delivery, education, climate adaptation, the cultural side of your organization, or even something as personal as how you keep your own learning alive in the age of large language models — Wicked Problems will give you the engineer's vocabulary for that work.

Pair the reading with Edwin Link in mind. Picture him working out of his father's piano factory, sketching his first simulator with organ parts and a pneumatic motion base, ending up with a machine that brought more than half a million American pilots home alive because someone had taken the trouble to build them a safe place to fail. That is what engineering for a world that won't sit still actually looks like.

— Arun

---

Pick up Wicked Problems: How to Engineer a Better World at The Thinking Spot in Minnetonka, or order it from bookshop.org — a portion supports independent booksellers like ours.