Categories
Leadership Uncategorized

The Sawed-Off Chair: Hyman Rickover’s Brutal Lesson in Accountability

It sounds like a legend, but it’s true.

If you wanted to command a nuclear submarine in the Cold War U.S. Navy, you first had to survive a personal interview with Admiral Hyman G. Rickover—the uncompromising “Father of the Nuclear Navy.”

In his office sat a notorious wooden chair. The front legs had been deliberately sawed short—several inches in some accounts—causing anyone who sat in it to slide inexorably forward. The seat was often polished slick as glass. While candidates fought to stay upright, Rickover unleashed a barrage of rapid-fire questions on engineering, history, philosophy, and their deepest personal failures. A weak or evasive answer might earn you banishment to a broom closet for hours “to think about it.” Other times, he’d deliberately provoke you just to see how you’d react under pressure.

Why would the man responsible for the most advanced, unforgiving technology of the era—nuclear reactors that could never be allowed to fail—rely on such seemingly petty tactics?

Because Rickover understood a hard truth: technology doesn’t prevent disasters. People do.

A nuclear reactor doesn’t care about your rank, your procedures, or your consensus. It obeys physics.

In an environment where a single mistake could mean catastrophe, Rickover demanded officers who took absolute, personal ownership of every outcome.

He put it best himself:

“Responsibility is a unique concept. It can only reside and inhere in a single individual. You may share it with others, but your portion is not diminished. You may delegate it, but it is still with you. You may disclaim it, but you cannot divest yourself of it… If responsibility is rightfully yours, no evasion, no ignorance, no passing the blame can shift the burden to someone else. Unless you can point your finger at the man who is responsible when something goes wrong, then you have never had anyone really responsible.”

That philosophy is why the sawed-off chair existed. It wasn’t hazing. It was a deliberate test: When your environment is uncomfortable, unfair, and literally working against you, do you complain? Do you slide off and give up? Or do you dig in, brace yourself, and maintain control while thinking clearly under stress?

Rickover wasn’t building bureaucrats. He was building leaders who could be trusted with the most dangerous machines ever created—men who wouldn’t hide behind systems, committees, or “shared accountability” when things went wrong.

Today, in our matrixed organizations, endless committees, and culture of diffused blame, this feels almost radical. We’ve grown comfortable with collective responsibility that conveniently means no one is truly responsible. Rickover called this kind of bureaucratic diffusion “systematic strangulation.”

We may not run nuclear reactors, but the principle applies everywhere that matters: in engineering, in business, in life.

True leadership isn’t about comfort or consensus. It’s about character forged in discomfort. It’s the lonely recognition that the buck doesn’t just stop with you—it starts with you, lives with you, and cannot be outsourced.

Categories
AI

Reading Tea Leaves: Human-AI Agents

Ben Thompson shared what feels like a very important insight in yesterday’s Stratechery newsletter:

I think there is a massive AI-enabled opportunity that is currently being missed by all of the major model-makers, or at least their product teams: human-AI agents.

Right now all of the consumer-focused AI interfaces — i.e. the chatbots — are built for a single user… There is a huge productivity unlock, however, that happens when you make them multiuser.

He goes on to describe an interaction he had in the last few days with his assistant Daman:

Last week Daman did some initial research for a complex decision using ChatGPT; he then shared a link to his chat, which meant I could go back to the beginning and trace his assumptions, the back-and-forth of his conversation, and then continue the conversation on my side. After diving deeper into various options — and correcting a few errant assumptions from the beginning — I came up with a promising course of action and, instead of having to explain it all to Daman to follow up on, I simply shared a link to my version of the chat back to him for reference.

What Thompson is describing is team interaction with the participation of a chatbot — like having it as part of the team. Sort of like having a water cooler conversation with a group of colleagues, one of whom (the chatbot) has done a lot of work and is sharing it with the others who then “embrace and extend” its findings.

Ben concludes:

I was absolutely blown away by how well this worked. Instead of replacing Daman with an AI agent, … I accidentally stumbled into a way to supercharge Daman’s value to me: he’s my human AI agent!

Seems like this is such a great idea that it will be quickly embraced by the various team centric platforms. Maybe it already has been somewhere? Reminds me a bit of the NotebookLM podcast capability where you can interrupt the conversation going back and forth between the two hosts.

Fascinating stuff! Look forward to seeing how this evolves as it involves treating AI as an adjunct to human productivity rather than just a white collar job replacement.

Categories
AI Leadership

The Power of Two

I recently watched and thoroughly enjoyed Harry Stebbings’ interview with OpenAI’s Sam Altman (CEO) and Brad Lightcap (COO). In addition to gaining new insights into OpenAI’s evolution, their conversation covered a wide range of topics regarding the future of AI and its implications for society and new ventures.

One of the most fascinating aspects was the dynamic between Altman and Lightcap — hearing them discuss their respective strengths, weaknesses, and how those translate into their roles at OpenAI. It’s uncommon to witness a dual interview like this, with two colleagues who have clearly worked together for years and have complete confidence and trust in each other’s judgment and insights.

Throughout my involvement with various small companies, I wish I could have experienced such a powerful duo! In my experience, it’s not uncommon for the CEO to dominate the senior management team’s dynamics. While this sometimes works well, I’ve also seen it lead to reduced performance or frustration among senior managers due to the CEO’s actions.

Altman and Lightcap (and OpenAI by extension) appear to have a much more synergistic working relationship — effectively amounting to a co-equal division of responsibilities. I highly recommend watching this conversation for anyone involved in a startup aiming to scale quickly and effectively! Congratulations to Harry Stebbings for his hosting this excellent conversation with two key individuals leading the evolution of AI!