Why Smart People Join Stupid Companies

On Competence as a Liability

In a meeting during my second month, the CEO was told by two of his executives, including me, that what he was proposing was a HIPAA violation. He sat with that for a beat. Then he said: “I don’t think that’s a blanket statement.”

Federal privacy law, apparently, did not apply to him categorically. This was clarifying. What I want to examine is not what that environment was (I have written about it at length across this series) but why I was in that room in the first place. I had seen, before I accepted the offer, what I was walking into. I went anyway.

What I Found in Due Diligence

I researched the organization. I talked to people. I asked for documents. What I received should have been disqualifying.

The financial structures enriched the founders regardless of how the business performed. The governance architecture provided no real oversight. The track record, when I examined it carefully, didn’t match the story I was being told about the company’s trajectory. None of this was hidden particularly well. It was visible to anyone who looked with the right framework.

I also talked to the engineers who would report to me. They were technically strong. They were frustrated with the current leadership. They were, in my assessment, exactly the kind of people I had built good organizations around before.

That assessment was not wrong. The category error came next.

What I Told Myself

Good team. Bad leader. I can be a buffer.

Twenty years of fixing organizations had produced a specific kind of confidence. I had walked into dysfunction before and built something real out of it. Career frameworks. Compensation models. Recruiting infrastructure. I had reduced attrition from 40% to 18% across a 350-person organization. I had taken on-time delivery from under 40% to over 90% at multiple companies. My track record was not imaginary. The belief that my frameworks were the relevant variable was not vanity. It was evidence-based.

The problem was that I was applying evidence from one category of problem to a different category of problem.

The red flags I had seen (the financial structures, the governance gaps, the narrative that didn’t hold up) I had processed all of them through a framework designed for systems that had drifted from their purpose. Systems where the dysfunction is entropic, where the architecture was built in good faith and has degraded through neglect, bad hires, compounding exceptions. That kind of dysfunction is fixable. I had fixed it. The frameworks are the right tool for it.

This was a different kind of system.

The Two Categories

The most important diagnostic distinction in organizational work is one most people learn only by getting it wrong.

There are systems that have drifted from their stated purpose, and there are systems that are serving their actual purpose. The second kind is serving it precisely. The corruption is precisely the design.

In the first kind, the career framework is broken because nobody built it correctly. In the second kind, the career framework was never intended to constrain authority. It existed to provide the appearance of constraint. The performance review process, in the first kind, produces bad outcomes because the criteria are poorly designed. In the second kind, the criteria are performing exactly as intended: measuring proximity to the founder and calling it impact.

From inside both systems, they look similar. The org chart doesn’t tell you which you’re in. The offer letter doesn’t. Even the people inside often can’t tell you, because the second kind is designed to make the gap between stated and actual purpose invisible.

What tells you is the answer to one question: are the correction mechanisms built in good faith?

Not whether they exist. Every organization I have ever seen has a handbook, a reporting mechanism, a stated commitment to X. The question is whether those mechanisms were ever intended to do what they claim to do. When the answer is no, you are in the second kind of system, and none of the standard tools for organizational repair have any purchase. There is no ground for them to stand on.

I knew enough to ask the question. I looked at the governance structure and concluded it was poorly designed. What I did not fully reckon with was the possibility that it had been designed to fail.

The Competence Trap

Corrupted systems don’t fail to recruit competent people. They recruit them preferentially, because a competent person who believes they can be a buffer is more valuable to the extraction architecture than a compliant one. A compliant insider doesn’t need to be convinced. A competent believer provides cover, organizational legitimacy, and a credible buffer that the system can operate behind until it no longer needs one.

The “I can be a buffer” reasoning is coherent. I had been a buffer before, successfully. My frameworks had protected engineers from arbitrary authority. My structural interventions had created conditions where competence was recognized and rewarded. The reasoning was not naive. It was a pattern match to the wrong prior experience.

The system processes people who believe they can fix it. That is the specific output it is designed to produce from the specific input of a competent person with good frameworks. You go in believing architecture is the variable. You are correct that architecture is the variable. You are wrong about which architecture.

By the time this becomes visible, you are already inside. The retaliation architecture has had time to observe you, catalog where you push back, map your relationships, and identify the people you would warn if you saw what was coming. The system is running its actual operation on you while you are still trying to run your repair frameworks on it.

What the Frameworks Can Actually Do

After “I don’t think that’s a blanket statement,” I stayed for several more months. Not because I didn’t know what I was looking at. Because I had not yet accepted that knowing what I was looking at was insufficient to change it.

The frameworks this series has built (the distinction between drift and design, the three-layer coherence model, the pattern recognition for capture) are diagnostic instruments. They tell you what you’re looking at. They don’t give you leverage over it. Inside a corrupted kernel, the only thing they can do is make the exit decision legible before the system makes it for you.

Where they could have protected me is before I walked in. Applied honestly to what I found in due diligence, they would have produced the right conclusion. I had the vocabulary. I discounted the answer because I had twenty years of evidence that my competence was the relevant variable.

Architecture is always stronger than the person. I have written that. I believed it about other people’s situations. I made a private exception for my own.

Category errors don’t feel like errors while you’re making them. They feel like confidence.