The altimeter was stuttering at 14,004 feet, a nervous digital twitch that mirrored the drumming of my fingers against the console. We were caught in that thick, milky soup where the horizon ceases to exist and the world is reduced to a series of glowing emerald lines. I’d spent the last 14 minutes arguing with the lead engineer about the specific tension of the primary control threads-I was dead wrong, insisting on a 1.4-unit offset that didn’t exist, but I won the argument because I spoke with more conviction. Now, sitting in the humid silence of the deck, that victory felt like a lead weight in my stomach. It’s funny how we prioritize being right over being safe, yet in the margins of flight, safety isn’t found in the correctness of a single voice. It’s found in the static between two people.

I watched the researcher in the jumpseat, a quiet woman named Sarah who was currently obsessing over a 244-item checklist. She wasn’t looking at our hands or our eyes; she was listening to the gaps in our speech. Most safety audits are obsessed with what is said-the formal callouts, the standard operating procedures, the adherence to ICAO Level 4 or 5 English. They want to see a quantifiable individual proficiency. But Sarah was there for the ghosts in the machine. She was looking for the informal coordination that happens in the margins, the stuff that makes the difference when the 4th engine decides to cough its guts out over the Atlantic.

We focus so heavily on individual skill because it’s easy to measure. You can give a pilot a test, check their grammar, and slap a sticker on their license. It’s much harder to measure the way a First Officer anticipates the Captain’s need for the landing lights just by the way his shoulder tenses. This is the relational bias of safety. We test the parts and assume the engine will run, ignoring the fact that a collection of perfect parts can still result in a catastrophic failure if the friction between them isn’t managed. I once saw a crew with perfect language scores and impeccable individual flight records almost put a jet into a hillside because they didn’t know how to disagree with one another. They had the technical vocabulary, but they lacked the relational vocabulary.

The Tension Between Formal and Functional

There is a peculiar tension in the stickpit when the formal becomes the enemy of the functional. We are taught to speak in a specific way, to follow a script that has been bleached of all personality. Yet, when things go sideways, the script is usually the first thing to burn. In those moments, we fall back on the patterns we’ve built in the quiet hours of a 14-hour haul. The shared jokes about the terrible catering, the complaints about the 44-minute delay in Zurich, the tiny admissions of fatigue-these aren’t distractions. They are the glue. They build a mental model that is collective rather than individual.

[The silence between commands is where the real work gets done.]

I remember calibrating the thread tensions on a prototype 4 years ago. Drew D.R., they called me, the ‘Thread Tension Calibrator’ with a God complex. I was so focused on the individual metrics of the wire-the 104-pound pull-strength, the 4-millimeter oscillation-that I ignored the way the whole frame was vibrating. I was measuring the thread while the loom was falling apart. That’s exactly what we do with crew assessments. We look at the individual pilot’s ability to navigate the Level 6 Aviation requirements, ensuring they can handle complex linguistic structures, but we don’t look at whether they can hear the fear in their partner’s breath. We measure the thread, not the weave.

It’s a systemic failure of imagination. We believe that if we make the individuals better, the system becomes safer. But safety is an emergent property of the interaction, not a sum of the inputs. If you have two Level 6 pilots who hate each other or don’t trust each other’s intuition, you effectively have a Level 0 stickpit. The informal coordination-the ‘you got it?’ and the ‘I’m on it’ that happens outside the official radio frequency-is the actual safety net. And yet, because it doesn’t show up on a 4-point scale, it doesn’t exist to the regulators.

I’m thinking about that argument I won earlier. I was wrong, and the engineer knew I was wrong, but he yielded because of the hierarchy. That yield is a safety gap. In a healthy crew, he would have stepped into my ‘wrongness’ and corrected it without bruising my ego. He would have known that my insistence was a symptom of my 14 hours of wakefulness, not a reflection of the data. He would have read the margins of my behavior. This kind of deep, intuitive coordination isn’t something you can teach in a 4-day seminar. It’s something that grows in the spaces between formal training sessions.

The Data of Human Connection

The data supports this, though the data itself is a bit of a character in this story, always trying to simplify the messiness of human connection into a 44-column spreadsheet. Studies of emergency management consistently show that the most successful crews aren’t necessarily the ones with the highest individual test scores. They are the ones who have spent the most time together, or who come from cultures that emphasize collective responsibility over individual heroics. They are the crews who have developed a shorthand, a way of communicating that bypasses the need for full, grammatically correct sentences.

Sometimes, a grunt is more informative than a three-paragraph briefing.

If we really wanted to test for safety, we’d stop putting pilots in simulators by themselves or with random partners. We’d test the pair. We’d see how they handle the 4th hour of a boring flight, how they manage the subtle drift of attention, how they correct each other’s tiny mistakes before they become big ones. We’d look for the ‘micropatterns’ of trust. But that would require us to admit that we don’t have total control over the variables. It would require us to admit that safety is a bit of a mystery, a social dance that we are still trying to map.

[We are obsessed with the singer, but the song depends on the acoustics of the room.]

There was this one time, about 14 months ago, when I was working on a sensor array. The lead technician was a guy who barely spoke English, maybe a Level 4 on a good day. I was frustrated, trying to explain the 4-way split in the data stream. He just watched me, eyes moving between the screen and my face. Finally, he reached out, touched a single wire, and gave a tiny, almost imperceptible shake of his head. He didn’t have the words to tell me I was misinterpreting the 114-hertz spike, but he had the relational awareness to see my confusion and the technical intuition to point to the source. We fixed it in 4 seconds. If I had been judging him on his formal proficiency, I would have dismissed him. Instead, I relied on the margin.

This individualist bias is a comfort blanket. It tells us that if we can just find the ‘bad’ pilots and train them better, we can eliminate risk. It ignores the fact that ‘good’ pilots do ‘bad’ things when the relational dynamic is broken. It ignores the 234 different ways a team can fail even when everyone is doing their job perfectly on paper. We are terrified of the relational because we can’t automate it. We can’t put it in a manual. We can’t tell someone, ‘On page 44, you will find instructions on how to care about your co-pilot’s subtle hesitation.’

So we stick to the metrics. We focus on the ICAO levels and the simulator hours and the 4-year degree requirements. We build a fortress of individual competence and then wonder why the roof still leaks. The leaks are in the joints, the places where one person’s responsibility ends and another’s begins. That’s where the informal coordination lives. That’s where the 44-millisecond delay in a response becomes a warning sign of a mental overload.

From Level 6 Individuals to Level 1 Team

As we began our descent, the stuttering altimeter finally settled. I looked over at the controls, my hands still feeling the ghost of that wrong argument. I realized then that the most important thing I could do wasn’t to be right, but to be reachable. To make sure that the margin between me and the rest of the crew was wide enough for them to step in. We are so busy trying to be Level 6 individuals that we forget to be a Level 1 team-a single, unified organism moving through a space that doesn’t care about our certifications.

The researcher, Sarah, finally closed her notebook. She had 44 pages of notes, and I suspect most of them were about the things we didn’t say. She saw the way I checked the 4th gauge twice when I thought no one was looking, and the way the First Officer adjusted the cabin temp just as I was starting to sweat. She saw the coordination that isn’t in the manual. She saw the truth of the flight: that we aren’t just flying a machine; we are maintaining a connection.

In the end, the tension isn’t just in the threads of the plane. It’s in the threads of the conversation. If those are too tight, they snap under pressure. If they’re too loose, the whole system becomes unresponsive. Calibrating that tension is the hardest job in the stickpit, and it’s the one we spend the least amount of time testing. We are so focused on the words that we miss the music. And when the music stops, that’s when the ground starts coming up to meet you much faster than the 4-meter-per-second descent rate you were expecting.

I’m still thinking about that engineer. I should probably go find him and tell him he was right about the 1.4-unit offset. Not because it matters for the machine anymore-the plane is landed, the threads are holding-but because it matters for the next time we’re in the soup. It matters for the margin. If I want him to speak up when it really counts, I have to be willing to be wrong when it doesn’t. That’s the real crew resource management. It’s the humility to realize that 4 eyes are only better than 2 if they’re allowed to see different things.