Unnecessary complexity is just a hallucination with extra steps.
When a system becomes too tangled, it almost always means one thing: there was no clarity at the beginning. Instead of thinking, people start adding. A little “just in case,” a bit “to make it clearer”… and suddenly the process turns into Frankenstein — parts don’t match, logic is lost, and layers are stacked on top of each other.
From the outside, everything looks solid. Inside — chaos.
And then the “second shift” begins. Cleaning up the mess. Removing the unnecessary. Fixing the misalignments. Straightening things out. This isn’t development. It’s treating hallucinations.
You see it instantly in real processes
Three levels of approval for a $500 expense.
After the CFO signs off, five more people jump into the thread asking “why?” — turning what should be a decision into a circle of clarifications. Responsibility is replaced by ceremony.
Research confirms it: perceived complexity almost always appears when goals are vague and management is reactive. The University of Groningen found that excessive procedures without context lead to overspending and frustration. Not innovation.
And if you don’t clean your processes regularly, the complexity always comes back.
Amazon’s Invent and Simplify
This reminds me of one of Amazon’s leadership principles: Invent and Simplify.
Most people focus on the “Invent” part. But Simplify is what makes innovation real.
Inventing something new is not that hard. Ideas are cheap. The hard part — the art — is simplification. Taking something complex and making it feel obvious. Making it so clear that anyone can operate it without a manual.
Here’s a test I use: if you can’t explain a process in three steps — a → b → c → result — then what you have isn’t innovation. It’s a corporate hallucination.
Why this matters more with AI
AI is an accelerant. It makes it faster to build things. But it also makes it faster to build the wrong things with unnecessary complexity.
I’ve seen AI-generated architectures that look impressive on a diagram. Six microservices, three queues, a state machine, and a saga pattern — for what turned out to be a CRUD app with two users.
The AI wasn’t wrong, technically. It was hallucinating complexity — exactly like an organization that adds process without purpose.
The antidote is the same in both cases: start with clarity about what the system needs to do, then remove everything that doesn’t serve that purpose.
The bottom line
Complexity doesn’t make a system mature. Simplicity makes it resilient.
The next time you’re looking at a process, a codebase, or an organizational structure that feels “too heavy,” ask yourself: was there clarity at the start? Or did people start adding because they didn’t know what to do?
If you can’t explain it in three steps, you have a hallucination — not a system.
Simplify is the underrated half of innovation. I’ve built my career on taking complex systems and making them feel inevitable. That’s the real engineering work.