Ethan Carter: What Technical Coaching Used to Mean

← The Last Batch

The trail started with a court filing. Whitaker Payroll's year-end collapse generated enough IRS penalties and client lawsuits that the documents were public. One name kept appearing in depositions — a technical coach who had been hired, fired, briefly rehired, and fired again before the system he warned about finally broke.

Ethan Carter had left Ohio. He was running a small SaaS company registered in Wyoming, building on the extraction methodology he developed during his time at Whitaker. A cold email got a reply within a day — brief, direct, no pleasantries. He said he would talk, but not over video. "Come out here if you want the real version."

The drive from Sheridan took twenty minutes on a county road that turned to gravel after the last mailbox. The ranch sat on a bench of land above Piney Creek — a plain house, a barn with the doors open, and a corral where two horses stood nose-to-tail against the early flies. Ethan came out in a canvas jacket and work boots, carrying two mugs of black coffee. He set them on the porch rail and sat down in a chair that had clearly been his thinking chair for a while. The Bighorns filled the western sky, still white at the peaks.

Interviewer

You were introduced to Whitaker Payroll as — and I'm quoting from a court filing here — "a guy who is good at coding." How did you actually get there?

Ethan Carter

Preston Hale made the introduction. He's a management author — three bestsellers, keynotes, the whole circuit. Preston and Graham Whitaker knew each other from fundraiser committees. Graham had inherited the company after his parents died and never really understood the technical side. Nathan Cole, the lead developer, was holding the entire system together alone and had been asking for help for years. Preston connected us.

I sent retainer documents that described the role, the approach, the scope. They went unread. Graham said "work with Nathan" and went back to his galleries. I never met him until the day they fired me the second time.

That sounds like a detail. It's the whole story. If the owner defines you as hands, every act of judgment becomes trespassing. You're welcome to type. You're not welcome to think.

Interviewer

What did you actually do there?

Ethan Carter

I introduced test-driven development and continuous integration to a system running on COBOL nightly batches and VB6 desktop clients. Nathan and I built a test harness, then used AI-assisted extraction to pull business rules out of the COBOL and turn them into thousands of executable C# tests. For the first time in decades, someone could prove what the batches were actually doing versus what people believed they were doing.

The mismatches were significant. Withholding logic that had been producing wrong numbers for years across multiple jurisdictions. Not theory. Actual reruns against historical output. Verifiable.

And that's when the trouble started. Because proving the system was wrong meant proving the people whose identities depended on the system being right had been wrong too. And that was not a technical problem.

Interviewer

You keep saying "technical coaching." Most people hear "coach" and think someone running stand-ups and retrospectives. What did the role used to mean?

Ethan Carter

It meant helping a team build better software through better feedback, stronger technical habits, and a tighter connection between judgment and execution. Pairing. Test-first. Continuous integration. Refactoring. Work small enough that misunderstanding shows up on Tuesday instead of in production six weeks later.

The center of gravity was quality. Can the team tell whether the system still works? Can they change it safely? Can they get evidence before damage compounds?

By the mid-2000s, a lot of enterprises discovered you could sell "coaching" without touching code. Scrum Alliance started certifying scrum masters in 2002 — two-day course, no engineering requirement. Within a few years the pipeline was flooded. The engineering spine got removed. What remained was process moderation, emotional shock absorption, and executive reassurance at a discount. By 2009, people like Robert Martin were so alarmed they wrote a separate manifesto — the Software Craftsmanship Manifesto — trying to put the technical practices back into the conversation. That version without engineering is easier to buy because it never forces leadership to fund the uncomfortable parts — testing, CI, refactoring, pair work, or the political cost of giving developers real authority.

Interviewer

There's a set of values people in your field point back to. Individual interactions over process. Working software over documentation. What went wrong between those values and what happened at Whitaker?

Ethan Carter

The manifesto said: if your process starts defeating the production of working software, the process is no longer the grown-up in the room. People hear "over" and read "instead of." That's the wrong reading. It's a priority statement, not a rejection of discipline.

What Graham and Derek did after they fired me was the exact opposite. They imposed a formal workflow — SMEs define requirements in Word documents, developers implement against the documents, validation happens through sign-offs. Five boxes on a chart. Clean arrows. No blood on the page. It looked mature. It was paralysis with better manners.

Because the actual rules of the system were not sitting in anyone's head waiting to be written down. They existed as decades of accumulated behavior inside running code. You only discovered them by touching the code, rerunning the output, and comparing. That's what the test harness did. The formal process replaced that feedback loop with paperwork and called it governance.

Interviewer

Where does Extreme Programming fit? The practices you were actually using at Whitaker.

Ethan Carter

XP was the engineering answer. The manifesto gave values. XP gave mechanics. Write the test first. Integrate constantly. Pair. Refactor while the tests are green. Keep releases small. Keep feedback fast enough that fantasy cannot outrun the code.

Every executive loves the idea of adaptability until you show them the cost. Adaptability is not a mindset poster. It's a codebase that can take change without lying, plus a team allowed to learn in public. That's what I was building with Nathan at Whitaker. And it worked. The tests were green. The CI pipeline ran. The extraction method scaled. The problem was never the engineering. The problem was that working engineering made the hierarchy visible.

Interviewer

People still talk about the Chrysler payroll project as proof that these practices don't survive real organizations. You brought XP-style practices into another payroll company and it ended badly too. Is there a pattern?

Ethan Carter

C3 at Chrysler is one of the worst-retold stories in software. Kent Beck and Ron Jeffries helped introduce the practices that became XP. The system went live for a real payroll population. It paid actual people. Then the customer role burned out, the Daimler merger changed sponsorship, and management appetite shifted. The project was stopped. People retold that as "XP failed" because that version protects management culture better than the truth.

The truth is harsher. Disciplined engineering got a failing payroll effort into production, and the organization still chose a different political future. That's what happened at Whitaker too. The engineering worked. The tests proved it worked. Graham watched me show him branches producing wrong withholding numbers for years. He saw the evidence. And he decided it was more important to protect the people whose identities depended on the old numbers being right.

The pattern is not that disciplined practices fail in real organizations. The pattern is that real organizations sometimes prefer failure to the status disruption that success would require.

Interviewer

Preston Hale. You mentioned him. He keynotes conferences about self-organizing teams and autonomy.

Ethan Carter

On stage, Preston talked about unlocking human potential. Trust your people. Command and control is dead. Self-organizing teams outperform managed teams. The audience wrote it down like scripture.

Behind closed doors, he told Graham to keep programmers on short reins. His words, according to Nathan: "Short reins, Graham. People do their best work when they know exactly where the fence is. Don't give them room to improvise."

Same man. Same month. He sold autonomy to people who buy books and sold control to the person who bought his consulting. Danny Kovacs — a colleague of mine — called it livestock management with a TED Talk vocabulary. I haven't found a better description.

Interviewer

You got fired twice. Walk through the mechanism.

Ethan Carter

The first time, I was in a meeting with Derek Lawson — the VP — alone because Nathan got pulled away by a support emergency. I showed Derek evidence of a Detroit tax defect. Real mismatch data. Derek processed it as insubordination from someone without standing and wrote up a budget memo that killed my access. The language was something like "fit, scope, and governance around external support." Triple abstraction. No verb anyone could take to HR.

Nathan got me back in on temporary terms. Over two weeks I used AI-assisted extraction to turn the COBOL into thousands of testable rules. The system became visible for the first time. That scared everyone whose survival depended on opacity — and I do not say that with contempt. Linda, Sharon, Donna, the subject-matter experts — they were not villains. They were women in their late forties whose professional life had narrowed until institutional memory was both the asset and the trap. When the extraction method made their moat drain, fear became institutional resistance.

The second firing came from Graham directly. He watched the evidence, watched the SMEs freeze, watched Derek defer to the freeze, and concluded that maturity required removing the source of disruption. "Graham would like Ethan's system access concluded effective immediately." Concluded. As if my access had been writing a novel and simply reached its natural ending.

Interviewer

There's a pattern in the filings where nobody actually decides anything. Everybody uses everybody else's hesitation as permission to wait.

Ethan Carter

The SMEs wouldn't say the numbers were wrong because they wanted Derek to make the call. Derek wouldn't override the SMEs because their uncertainty made him feel like the evidence wasn't settled. Graham watched them both hesitate and concluded the matter required more time. Nobody had to lie. Everyone just kept deferring.

And the person who broke the loop — the one who said the plain sentence out loud — became the threat. Not because I was wrong. Because I had made a judgment nobody else was willing to make, and that judgment made every other person's delay visible.

A verifiable number from someone with standing is a finding. A verifiable number from a contractor is an opinion with delusions of authority. Same data. Radically different social weight. That's the mechanism. And it doesn't require malice. It only requires a hierarchy that treats the framing as more real than the evidence.

Interviewer

What would you say to someone running a company like Whitaker who hasn't hit the wall yet?

Ethan Carter

Read the engagement documents before the person walks in. That's not a metaphor. Graham's unread retainer was the framing. "I don't need to understand who you are because who you are doesn't matter to me." Once that's set, everything the person does that exceeds the silent frame gets treated as overreach.

If you hire someone for technical judgment, say so publicly. Not in the hallway. In the room where the SMEs and the VP and the developers are sitting. Make the standing explicit. Otherwise you've hired judgment and caged it inside a role defined as labor, and then you'll fire the person for acting like what you needed instead of what you named.

Keep the engineers and the business people in the same evidence loop. Don't build a chain where one group defines, another implements, and a third validates after the fact. That model preserves status at the price of truth.

And watch your own reaction when evidence arrives. The moment it starts feeling like insult, you're in the danger zone. Graham saw correct numbers on a screen and experienced them as a threat to his father's legacy. That's not business judgment. That's grief wearing a suit. And a company can't survive on grief.

He stood up and took both mugs inside. From the porch the Bighorn foothills ran north in a long unbroken line, white at the peaks and brown at the tree line — the kind of view that does not require your opinion to continue existing.

One of the horses walked to the fence and stood there. Ethan came back out, noticed, and went over. He rubbed the animal's neck without saying anything.

He said Whitaker was not unusual. He had seen the same reflex in other states and other industries: people wanting modernization without status disruption, process language standing in for judgment, evidence treated as aggression when it arrived from the wrong title. Whitaker just took the pattern all the way to the wall.

At the edge of the porch he turned back. "If a company makes truth wait for hierarchy before it counts, the collapse has already started. Whitaker proved that. Chrysler proved that. The next one will prove it again."

Then he walked to the barn. There was a fence section down at the south end and he wanted to get it done before the afternoon wind.