The Man Who Loved Dragons

There is an old story about a man who loved dragons.

He loved them completely. He had dragon pictures on every wall. Dragon curtains, dragon carpet, dragon duvet. He slept in dragon pyjamas and dreamt of dragons. At work he wore a dragon tie, had a dragon screensaver next to his dragon mouse mat. In his spare time he painted dragons, read about dragons, watched films about dragons. His entire life was organised around the love of dragons.

The queen of the dragons heard about this man and thought she would pay him a visit. She flew to his village, landed outside his house, and knocked on the door. The man opened the door, saw a real dragon standing in front of him — fire, teeth, scales, the full reality of what he had been celebrating all his life — and he screamed and ran for the hills.

When the queen returned to her kingdom, confused and a little hurt, she told a wise friend what had happened. The friend explained: “This was a man who liked the idea of dragons. Not dragons themselves.”

I have been telling that story to leaders for years. It never fails to land. And the reason it lands is that every leader in the room recognises themselves in it — at least a little.

• • •

This is the fifth article in a series about AI and the systems perspective. In the first, I argued that AI applied to work you haven’t studied just does the wrong things faster. In the second, I went deeper into the system conditions that determine whether AI helps or harms. In the third, I asked you to go and look at your own system — and showed what it looks like when someone actually does. In the fourth, I said: the people in the system matter, and the organisation’s ability to notice what is happening to them is what determines whether AI improves the system or burns its people out.

That article ended with a question: if you have understood the problem, seen the system conditions, looked at the work, and recognised the human cost — what do you do now?

This article is about you. The leader. And what it actually requires to stop admiring the idea of change and start confronting the reality of it.

• • •

What most leaders do

Most leaders I work with are intelligent, experienced, and genuinely well-intentioned. They care about their organisations. They want to do the right thing. Many of them have read the books, attended the conferences, and used the right language — customer-centricity, continuous improvement, learning organisation, empowerment.

And yet, when confronted with the reality of what those words actually require — when the dragon shows up at the door — they run. Not always dramatically. Sometimes they run by commissioning another report. Sometimes by setting up a working group. Sometimes by saying “yes, but we need to be realistic.” Sometimes by simply changing the subject.

Chris Argyris spent decades studying this pattern. He called it “defensive routines” — habitual ways of thinking and behaving that protect individuals and organisations from embarrassment or threat, but simultaneously block learning. What makes defensive routines so powerful is that the people using them are usually unaware they are doing it. The smarter and more successful you are, the better you are at defending yourself from the discomfort of being wrong — and the less likely you are to notice.

Here is the part that stings: Argyris found that most leaders espouse one set of values — openness, inquiry, learning, collaboration — while their actual behaviour reveals another. He called this the gap between “espoused theory” and “theory-in-use.” In practice, the theory-in-use for most leaders looks like this: maintain control, win the argument, suppress negative feelings, and appear rational. Not because they are bad people. Because this is what success in most organisations has rewarded for their entire career.

The result is a kind of learned incompetence. You become an expert at advocating your position without testing it. At making inferences about others without checking whether those inferences are correct. At designing conversations that feel open but are actually controlled. And the real cost is not that you make bad decisions — it is that you become unable to learn, because learning requires exactly the vulnerability that your defensive routines are designed to prevent.

This is not about personality. It is about system conditions — the same theme that runs through every article in this series. The system that promoted you rewarded control, certainty, and decisiveness. Now that same system needs you to be uncertain, curious, and willing to discover that what you thought you knew is wrong. That is a structural contradiction, and pretending it is just a matter of “mindset” does not make it go away.

• • •

The assumptions you carry

If the first step is recognising that you are running, the second step is understanding what you are running from.

Every leader carries assumptions about how organisations work. Most of these assumptions were never explicitly chosen — they were absorbed through decades of experience, education, and promotion. They feel like common sense. They feel like reality.

Here are some of the most common ones I encounter:

People need targets to perform. Without targets, there is no accountability, no direction, no way to know whether things are working.

Specialisation drives efficiency. Break work into functional steps, assign each step to a specialist, and optimising the parts will optimise the whole.

Managers exist to control variation. The role of management is to standardise, monitor, and correct — to ensure that everyone does the work in the right way.

Technology is the lever. When you want to improve, automate. When you want to scale, digitise. When you want to transform, implement AI.

If you have been reading this series, you already know that each of these assumptions has a shadow side. Targets drive gaming and failure demand. Specialisation creates handoffs that destroy flow. Controlling variation removes the ability to absorb the natural variety in customer demand. And technology applied to work you do not understand amplifies whatever dysfunction is already there.

But knowing this intellectually is not the same as being willing to act on it. That is the dragon at the door.

Ibby Hussain, a colleague I have worked with closely, uses a framework I find remarkably useful for leaders at this point. He asks you to take your assumptions and sort them into three columns:

What do you need to give up? Which beliefs, once examined against the evidence of your own system, are no longer defensible? This is the hardest column. Giving up a belief that has guided your career for twenty years is not an intellectual exercise — it is a loss. And losses need to be acknowledged, not glossed over.

What do you need to replace? Some assumptions are not wrong in themselves — they are incomplete. “Measure performance” is not wrong. But “measure performance against targets to control” might need to become “measure performance against purpose to understand.” The replacement is often subtle but consequential.

What do you need to adopt? What new assumptions does the evidence demand? If 95% of performance is determined by system conditions rather than individual effort — as Deming argued and as every study I have done confirms — then the leader’s job is not to manage people. It is to understand and change the system conditions that determine how people are able to work. That is a fundamentally different assumption about your role.

Let me make this concrete. Remember the insurance company from Article 3 — a million calls a year, 70% failure demand, and a sales incentive system that rewarded the people who gamed it and penalised the ones who actually helped customers. Hayley, the leader who listened to fifty sales calls and found that in only two could she honestly say it was the salesperson who made the difference.

Now imagine Hayley sitting down with these three columns.

Give up. The belief that individual sales targets drive sales performance. The evidence from fifty calls says otherwise — the system drives the outcome, not the individual. The top performers were not better salespeople. They were better at exploiting a broken incentive structure. That is not a small thing to give up. It means the entire performance management framework — the league tables, the bonuses, the coaching conversations built around individual numbers — was not just ineffective but actively harmful. It rewarded the wrong behaviour and punished the right one.

Replace. “Measure sales performance” becomes “understand what matters when customers buy.” What is the customer actually trying to achieve when they contact you? What information do they need? What is missing from the product that creates confusion, complaints, and calls that should never have happened? Replace the obsession with individual output with curiosity about the system that produces the output. Replace “how do we sell more?” with “why are 70% of our calls from people who already bought something and need help we failed to provide?”

Adopt. Use the knowledge you now have — about claims, about failure demand, about what customers actually need — and build it back into the system. How can feedback from claims flow into product development so customers get the right cover in the first place? How can advisers be supported to understand the customer’s real situation rather than cross-sell against a script? And here is where AI becomes genuinely interesting: not to automate the call centre, not to route failure demand faster, but to help people do the value work better. AI that helps underwriters see patterns across thousands of claims. AI that helps advisers understand what this customer actually needs based on their situation. AI that closes the gap between what is sold and what is needed — because someone finally understood what the gap was.

None of that is possible if Hayley runs from the dragon. All of it becomes possible the moment she stays.

This is not a one-time exercise. As you study your system and learn more about how the work actually flows, assumptions you thought you had resolved will resurface. The columns evolve. That is the point — it is a practice, not a checklist.

• • •

What studying your system actually looks like

If the assumptions work tells you what to question, the study of the system tells you what to look at. This is where the series delivers on what Article 2 promised: a way of understanding your organisation that starts with the work, not with the strategy. The strategy is hidden in the work.

The insurance example above already shows much of the method in action. But let me make the steps explicit, because they apply far beyond insurance.

Understand demand. Before anything else, understand why people contact your organisation — from their perspective, not yours. Hayley’s insurance company had a million calls a year. Seventy percent were failure demand — caused by something that had not been done or not been done right, from the customer’s perspective. That is the starting point, and most leaders have never done it.

Study the flow end-to-end. Follow a piece of work — a claim, a case, a customer request — from the moment it enters the system to the moment it is resolved. Not the process map on the wall. The actual journey, with all the handoffs, queues, rework, and waiting that the process map does not show. Measure the time. Count the touchpoints. Note where value is created and where waste is clocking the system.

Measure against purpose. Stop measuring what is easy to count and start measuring what matters. The purpose of an insurance claim process is not to process claims quickly — it is to return the customer to their pre-loss position. The purpose of an adult social care service is not to deliver hours of care — it is to help people live as independently as possible. When you measure against purpose, you see things that activity metrics hide.

Understand the system conditions. What is determining the performance you see? Not who — what. What targets, measures, roles, procedures, IT systems, and management practices are shaping how people do the work? This is Deming’s 95/5 in action: 95% of performance is determined by the system, not the individual. If you want to change performance, change the conditions.

Now — and only now — ask where AI fits. Once you understand demand, flow, purpose, and conditions, you can see clearly where AI might help and where it would simply automate the dysfunction. AI is powerful. But its power is entirely dependent on the quality of understanding it is applied to.

This is not a twelve-month programme. Many of these questions can be answered within days if you actually go and look — as Article 3 invited you to do. The point is not to make it a separate project. The point is to make it normal.

• • •

The deeper requirement

There is something harder than studying your system, and it is this: you cannot do any of the above honestly while operating in control mode.

Argyris identified three conditions for genuine organisational learning: valid information — everyone sees the real picture, not the curated version. Free and informed choice — people participate because they understand and choose to, not because they have been told to. And internal commitment — people believe in the direction because they arrived at it through their own understanding, not because it was mandated.

These sound reasonable. Most leaders would say they agree with them. And this is precisely the trap: the gap between what you say you believe and what your behaviour actually produces.

Do you share valid information — or do you filter it? Do you create the conditions for informed choice — or do you present a conclusion and ask people to get on board? Do you allow genuine commitment to emerge — or do you design programmes that produce compliance dressed up as buy-in?

If you are honest — and the whole point of this article is that honesty is the prerequisite — you will probably recognise moments where your behaviour has not matched your values. Not because you were acting in bad faith. Because the system you operate in, and the habits you developed to succeed within it, make it almost automatic.

The dragon is not just the truth about your system. The dragon is the truth about yourself.

• • •

The door is open

The man who loved dragons had every opportunity to stay. The dragon came to him. She was not there to threaten — she was there to say hello. The opportunity was standing right in front of him. He ran because the distance between the idea and the reality was too great.

You are at that door now. The previous four articles have brought you here — through understanding the problem, seeing the system conditions, looking at the work, and recognising the human cost. The dragon is no longer a picture on the wall. It is standing in front of you.

Staying in the doorway means accepting that some of what you believe is wrong. It means studying your own system with honest curiosity rather than defending what you built. It means creating the conditions for others to tell you the truth — and being able to hear it. It means measuring against purpose rather than activity, and redesigning the work based on what you find rather than what you assumed.

None of this requires AI. All of it makes AI genuinely useful if you choose to use it.

The question that remains — and it is the question the final article in this series will address — is this: once you have stayed in the doorway, once you have done this work yourself, how do you bring others on the same journey? Because the cruellest paradox of all is that the clearer you see the potential, the more tempting it is to do the one thing that prevents others from seeing it too — to tell them, rather than creating the conditions for them to discover it themselves.

But that is for next time. For now, the question is simpler and harder: are you ready to meet the dragon?

• • •

How serious are you about dragons?

If this article has made you uncomfortable, that is not a side effect — it is the point. The discomfort is information. It is telling you something about the distance between where you are and where the work needs you to be.

Here is one thing you can do this week. Pick one assumption from your leadership practice — just one — and test it against reality. Not by reading about it. Not by discussing it. By going and looking.

If you believe your targets are driving the right behaviour, go and listen to five customer conversations and ask yourself: is the person I am listening to serving the customer, or serving the target?

If you believe your process is efficient, follow one case from start to finish and count every handoff, every queue, every moment of waiting.

If you believe your people know what matters to customers, ask three of them. Then look at whether the system actually enables them to deliver on that understanding. Listen to the gap between what they know and what they are able to do.

The dragon is not the evidence. The dragon is what the evidence means for you. What it requires you to give up, to replace, to adopt. What it demands of you as a leader.

How serious are you about dragons?

---

---

SOURCES & ACKNOWLEDGEMENTS

• The story of the man who loved dragons is adapted from “When Rhetoric Meets Reality” by Can Scorpions Smoke (canscorpionssmoke.com, 2017). I first encountered it as a teaching story in the Vanguard Method community.
• Ibby Hussain’s assumptions framework (give up / replace / adopt) is used here with his kind permission. Ibby is a Vanguard Method consultant and the originator of the UCER model (Understand, Conceive, Execute, Review).
• Chris Argyris’s work on defensive routines, espoused theory vs. theory-in-use, and the conditions for organisational learning draws primarily from Organizational Learning II (with Donald Schön, 1996) and On Organizational Learning (1999).
• W. Edwards Deming’s 95/5 principle is discussed in detail in Article 3 of this series.

---

If this resonates — if you are a leader trying to make sense of what your organisation actually needs, with or without AI — I would love to hear from you. You can reach me at jonas@vanguard-consult.dk or connect with me on LinkedIn.

This article was written with AI assistance — which felt appropriate, given the subject matter.