The Fatality of Fluency

I know what it feels like to be the only person in the room carrying a signal nobody wants to receive.

You are not shouting and you are not being difficult, you are simply seeing something clearly that the room has collectively decided not to see, because seeing it would be inconvenient, disruptive, or threatening to the architecture that everyone in the room has staked their position on.

The room doesn't need to argue with you. It just needs to make you feel alone enough, long enough, that you stop.

My old boss at AKQA once gave me a reference that was almost a warning. Leo is a good guy to have on your team. He dares to defy convention. The implication was clear. If you want someone who toes the line, who fits in, who doesn't disturb the comfortable consensus, then he's not your man.

He was right, I'm not that man.

And neither were Sherron Watkins, Tyler Shultz, or Roger Boisjoly.

What happened to each of them is not primarily a story about fraud, or corporate governance, or regulatory failure. It is a story about what happens when an organisation systematically dials out the variance of the outlier, because of the unmeasured, unpredictable nature of their contribution, and substitutes fluency for truth.

The consequences, in each case, were catastrophic. And entirely preventable.

Enron. August 2001.

Sherron Watkins knew the architecture from the inside. She had worked for Arthur Andersen before joining Enron. She understood the structures, what the Raptor SPEs were concealing and why. When Jeffrey Skilling resigned without explanation in August 2001 she did what the outlier always does and named the thing nobody else was willing to.

Her memo to Kenneth Lay opened with a single sentence: I am incredibly nervous that we will implode in a wave of accounting scandals.

She wasn't nervous about her position, in fact she was certain. The nervousness was for Ken Lay, a CEO being told that the empire was built on fiction.

He received the signal and said that he would look into it. Then he handed the memo to Vinson & Elkins, the same law firm that had helped structure the Raptor deals she was questioning. He asked them to investigate and Arthur Andersen to assist. The very same auditors whose sign-off had enabled the accounting structures she'd identified.

The system audited itself and declared itself healthy.

Watkins described it later as the Titanic analogy. She was a crew member telling the captain they had hit an iceberg. The captain sent for the people who had told him the ship was unsinkable, who then confirmed it was unsinkable as the ship went down.

Enron filed for bankruptcy four months later. Twenty thousand jobs, billions in pension savings lost and Arthur Andersen, a ninety year old accounting firm, gone within twelve months.

Watkins was alone in that room, but she was right. The system chose the smooth answer over the true one.

Theranos. 2014.

Tyler Shultz was twenty-two. Eight months into his first job. Working on the Edison device, a revolutionary blood-testing machine that had attracted hundreds of millions in investment and the personal credibility of some of the most distinguished figures in American public life, including his own grandfather, former Secretary of State George Shultz.

He had tested himself and his colleagues for syphilis. More than twenty percent received false positives. He had watched the company discard outlying statistical values from validation reports and had seen quality control audits run not on the Edison but on commercially available equipment.

He then went to Elizabeth Holmes directly and she referred him upward. The response, from Theranos president Sunny Balwani, arrived in bold red font across eight pages, calling Shultz arrogant, accusing him of understanding nothing, suggesting that if he had a different last name he would have been fired long before he chose to leave.

Then he went to his grandfather.

George Shultz had watched Holmes build her company in his living room. He had introduced her to the board. He had staked a portion of his considerable personal credibility on the promise of her vision. When his grandson told him the technology didn't work, he told Tyler he was wrong. That Elizabeth had assured him they went above and beyond all regulatory standards.

The fluency of Holmes, the turtleneck, the unblinking eyes, the story of a nineteen-year-old in a Stanford dorm room who was going to change medicine, she had been processed by the board as truth. The signal from a twenty-two-year-old who had spent eight months with the actual machine was processed as complete arrogance.

Theranos hired private investigators to follow Shultz. Lawyers later appeared in his grandfather's house to pressure him into silence. His family spent over four hundred thousand dollars in legal fees and he was not invited to his grandfather's ninety-fifth birthday, but Elizabeth Holmes was.

That is the precise description of what it means to carry the signal. Your reality and the room's reality do not overlap. The room will protect its own.

Theranos collapsed in 2018. Holmes was convicted of fraud and the patients who received incorrect results, some for cancer, some for HIV, were never fully counted.

Challenger. January 27th 1986.

Roger Boisjoly had been raising the O-ring concern for months.

He was a senior engineer at Morton Thiokol, the company that manufactured the solid rocket boosters for the Space Shuttle. He had written an internal memo in July 1985, six months before the launch, warning in explicit terms that the O-rings sealing the booster joints had shown signs of erosion at low temperatures. His language was not cautious at all. In fact, he wrote that if the problem was not addressed then the result could be a catastrophe of the highest order, a complete loss of mission, vehicle, and crew.

The memo existed, in writing and six months before the launch.

The night before Challenger lifted off, with the forecast showing launch day temperatures of 29 degrees Fahrenheit, which was far below the 53-degree threshold at which the O-rings had been tested, Boisjoly and his colleagues held an emergency teleconference with NASA management and argued against the launch. They presented the data and were explicit about the risk.

The schedule pressure was enormous and the launch had already been delayed, so NASA pushed back. There were people watching and commitments had been made.

Thiokol management asked for a few minutes off the line to discuss internally. What happened in that room is the heart of the story. One manager turned to Boisjoly and told him to take off his engineering hat and put on his management hat. That sentence is the entire essay. Take off the thing that sees clearly and put on the thing that makes the decision the room has already decided to make.

Thiokol approved the launch without its own engineers. Boisjoly watched it happen and he knew the disaster that was coming.

Challenger broke apart seventy-three seconds after liftoff on January 28th 1986. Seven crew members died. As a teenager, I remember that day clearly. A day of incredulity.

Boisjoly was right and had been right for six months. He suffered severe depression in the aftermath. He was ostracised by colleagues and he never worked in the aerospace industry again.

He too had been alone in that room. And he was right. And the system chose the launch window over the truth.

The Pattern

Three rooms, three outliers and three correct signals. Three systems that received the truth and dialled it out. The mechanism in each case was identical.

The signal was uncomfortable, it contradicted the consensus and it came from someone whose position was inconvenient, the vice president without board authority, a junior engineer and a senior engineer whose data disrupted an institutional schedule. The system's response in each case was to route the signal through its existing verification architecture, the same lawyers, the same auditors, the same management consensus, and then receive back the answer it was already inclined to believe.

This is the system functioning exactly as designed.

By continually dialling out the variance of the outlier, because of the unmeasured, unpredictable nature of their performance, you build in the fallacy of fluency as a substitute for the hard truth. The smooth answer travels faster and the frictionless answer feels safer. The consensus answer protects everyone in the room who helped build the consensus.

And you expose the system to consequences that the outlier could see coming from six months away.

The outlier is not a management problem and the outlier is not a cultural fit issue. The outlier is not the person who needs to learn how to communicate better, influence more effectively, or take off their engineering hat. The outlier is the early warning system with an asymmetric advantage.

And right now, in most organisations, that system is being systematically disconnected in favour of a large language model that will tell you, fluently and confidently, exactly what the consensus already believes.

What This Means Now

Executive coaching exists and clean rooms exist. Multimodal AI exists, red teams exist and white hat penetration testing exists.

All of them are recognised as necessary investments in the health of complex systems.

However, the development of sovereign human judgment, the deliberate cultivation of the outlier's capacity to see clearly, carry the signal, and remain standing long enough to be heard, is not yet on that list.

It categorically should be.

Not because it is philosophically interesting or because Enron, Theranos or the Challenger disaster happened but instead because in each case the answer was already in the room, and the room chose not to listen.

The question is not whether your organisation has outliers. I'm sure it does.

The question is what are you doing with them and how will they be nourished?

Written beyond the air-gap.