The Better Aircraft Lost

What a fighter pilot who never fired his guns can teach a board about AI

James A Lang · 26 October 2026 · Orientation debt

In March 1953 a young American lieutenant arrived in Korea to fly the F-86 Sabre. The war had four months left to run. He flew twenty-two missions over the Yalu River, and on every one of them he was a wingman. His job was to stay on his leader's shoulder and watch. He never fired his guns. He never shot anyone down.

His name was John Boyd, and what he watched that spring would occupy him for the next forty years.

I want to start with what he saw, because it is a small mystery.

The Sabre's opponent was the Soviet-built MiG-15. On paper, the MiG was the better aircraft. It climbed faster. It turned tighter. If you had laid the two performance charts side by side in 1953 and asked an engineer to pick the winner, the engineer would have picked the MiG.

The Sabre pilots won anyway. They kept winning.

How is it possible that the better aircraft lost?

The usual suspects

There are three answers people usually give, and each deserves a fair hearing.

The first is training. American pilots had more flying hours and better instruction. That is true, and it matters, but it does not explain why the advantage showed up so consistently in the exact situations where the MiG should have been strongest.

The second is courage, or aggression, or some quality of the pilot that is hard to measure. Every air force believes this about itself. It explains everything and therefore nothing.

The third is luck, or the numbers being wrong. Historians still argue about the exact scores, and they should. But nobody seriously argues that the MiG, the better aircraft on paper, dominated the fight.

Boyd's answer was different, and it was hiding in two details that never appeared on a performance chart.

The Sabre had a bubble canopy, so its pilot could see more of the sky. And it had hydraulic flight controls, so its pilot could move from one manoeuvre to the next faster than a MiG pilot heaving on manual controls. The Sabre pilot saw more, and changed faster. He could take in the fight, make sense of it, choose and act, and then do it all again, before his opponent had finished his first turn.

Hold on to that. It took Boyd twenty-three years to turn it into a sentence, and another nineteen to turn it into a drawing.

The story we tell

You almost certainly know the drawing, or think you do.

Observe. Orient. Decide. Act. Four circles in a ring, arrows running clockwise. The OODA loop. It is in every MBA course, most consultancy decks and a great many military briefings, and I have used it myself. When I was building the Army's first Cyber Protection Team, Boyd was one of the thinkers I went back to. The lesson everyone takes from it is speed. Whoever goes round the loop fastest wins. Decide faster than the competition. Ship faster. Loop faster.

It is a tidy lesson, and Boyd never drew it.

This is the part of the story I find remarkable. Boyd never wrote a book. He wrote one nine-page essay, in 1976, and spent the next twenty years giving briefings, some of them six hours long, to anyone in the Pentagon, Congress or the Marine Corps who would sit still. His slides are free online, in the archive at coljohnboyd.com. And in all of those slides there is exactly one drawing of the loop. It is in a five-slide briefing called The Essence of Winning and Losing, dated June 1995, two years before he died.

When you go back and look at it, three things jump out.

The first is the title. Boyd called it the OODA "Loop" Sketch, and he put the word Loop in quotation marks. He was telling you it is not really a circle.

The second is the shape. Orientation is not one step in four. It is the largest thing on the page, sitting in the middle, wired to everything else. Inside it Boyd wrote the things that shape how each of us sees the world: genetic heritage, cultural traditions, previous experience, new information. Two lines run straight from orientation to action, skipping the decision altogether. He labelled them implicit guidance and control. That is what an expert does most of the time. They recognise the situation and act, without stopping to decide.

The third is the labels. Beside the decision Boyd wrote, in brackets, hypothesis. Beside the action he wrote test.

We have been telling this story backwards.

What Boyd actually found

Boyd was not, at heart, an aircraft man. He was interested in how a mind keeps its picture of the world true.

His 1976 essay, Destruction and Creation, makes the argument in nine dense pages. Any model we hold of the world, any plan, doctrine or strategy, drifts away from the reality it describes. You cannot repair it from the inside. The only remedy is to break it apart and rebuild it with fresh observation. Then you do it again, for ever.

Notice what drives that cycle. Error. The gap between the picture and the world is the signal. A mind that is never allowed to be wrong has no signal, and its picture quietly rots.

Now go back to Korea. The Sabre pilot's edge was not, in the end, speed. Speed was the by-product. His edge was that his picture of the fight stayed true for longer than his opponent's. He saw more, so his orientation was better fed. He could change faster, so every manoeuvre tested his picture against the world and corrected it.

Boyd proved the point himself at Nellis Air Force Base in Nevada, where he taught at the Fighter Weapons School. He made a standing bet with any pilot: start on my tail, the worst position in a dogfight, and within forty seconds I will be on yours. By his biographer's account he never paid out. He did not do it by being faster. He did it by doing something his opponent's picture of the fight did not include. The other pilot's orientation stopped matching reality, and in that gap Boyd was already behind him.

That is what "getting inside someone's loop" really means. It is not about out-running their clock. It is about making their picture of the world stop matching the world. In 1987 Boyd wrote the sentence that holds the whole idea: orientation is the Schwerpunkt, the point of main effort.

Judgement, in other words, is not made at the moment of decision. It is made in orientation, and orientation is built by being wrong, safely, often, and learning.

Then we handed it to a machine

Here is where the old story turns into a new one.

Every AI system your organisation has bought learned the same way. A guess. An error. A small correction. Repeated a billion times. Without the error there is no signal, and without the signal there is no learning. The machine was given the right to be wrong, at scale, and it earned its competence that way.

And what did we buy it to do? We bought it to orient. Modern AI no longer just fetches the data. It reads the situation, weighs it against everything it has seen and proposes what to do. In Boyd's terms, it does the sense-making. That is precisely the expensive, slow, human part, which is why boards buy it.

In 1983 a British psychologist called Lisanne Bainbridge wrote a short paper with a wonderful title: Ironies of Automation. Her point was simple and uncomfortable. The more you automate, the more you leave the human with only the hardest moments, the rare and fast and strange ones, and the fewer chances they get to practise them. Aviation has lived with that irony for forty years. We are now building it into the boardroom.

Every decision the system takes is one a person stops practising. Their orientation stops being corrected by the world, because they are no longer in contact with it. Nobody notices, because the numbers are good. The board remains accountable for every outcome.

I call the gap that opens up orientation debt: the distance between what your systems decide for you and what you could still decide without them.

Picture two lines starting at the same point on the day you switch the system on. One rises, because the system improves and you hand it more. The other falls, because your people stop practising. The wedge between them is the debt. It grows quietly. There is no line for it in the management accounts.

Compliance cannot see it either. ISO 42001 and the EU AI Act govern the model: its data, its risk class, its documentation. Nothing in either governs the judgement of the people answerable for it. You can be fully compliant and deeply in debt.

The day it hands one back

Like all debt, it falls due. It falls due on the day the system hands a decision back to a human, and that happens in one of three ways. It fails. It is attacked. Or it is simply, confidently wrong. Each arrives at machine speed, and each lands on a person who has not made that call unaided for months.

I spent twenty-two years in the British Army, and I watched a version of this long before AI. In barracks, mission command was routinely taken away from us. Every decision needed a signature. On operations it was demanded of us instantly, with lives attached. We practised judgement least where it was safest to practise it, and needed it most where it was not.

Boards are about to repeat that mistake with AI, at a scale and a speed the Army never managed.

None of this is an argument against the technology. In 2020 I became the Ministry of Defence's first cybersecurity AI product manager, and I would build that platform again. The lesson is not that AI is bad. The lesson is that we have misunderstood what it is for. AI can hold the context. Only people can hold the orientation, because orientation is the part that has to be answerable, and the part that has to be able to take the decision back.

So the work for a board is the work Boyd described. Keep the loop running in the people who are answerable, where being wrong is safe, so they can be right when it is not. Let them commit to their own view before they see the system's. Delegate the decisions you can take back, never the sense-making. And protect the person who overrules the machine for a stated reason, even when they turn out to be wrong.

Resilience, in the end, is what buys you the right to be wrong.

Back to the wingman

Which brings us back to the lieutenant over the Yalu in 1953.

He never fired his guns. The legend prefers an ace who learned it all in a dogfight. I think the real detail is the most important fact in the story. Boyd was not in the fight to win it. He was there to watch, and to understand. Twenty-two missions on someone else's shoulder built the orientation that changed how pilots fly, how fighters are designed and how armies think.

The better aircraft lost because the better loop won. Most boards are now buying better aircraft.