My LinkedIn post featuring a picture of a worker who went underwater to fix a broken pipe sparked some strong and sometimes conflicting reactions. While some celebrated his judgement and adaptability, others argued that his action proves the system failed long before he entered that pit.

Some warned against glorifying unsafe acts, while others pointed out how often real work diverges from the neat world of procedures. These reactions are symptoms of how complex organisations actually operate, and they reveal something important that safety management often overlooks. Formal rules and real work are different worlds, and both matter.

Rules don’t describe reality, but they stabilise expectations. Though it’s easy to say: “The person should have followed the procedure”, formalisation is about creating membership conditions: “This is how we do it here, and these are the expectations you must meet to be part of this organisation.” These expectations become stable over time, across tasks, and across people. That stability lets complex organisations coordinate work even when no one can oversee everything.

Stability has side-effects, since expectations that don’t match the real environment become rigid. People carry the conflict between formal duties and informal loyalties. Visible agreement hides disagreement, because disagreement threatens membership. So when someone like the pictured worker adapts in a muddy backyard, he's not breaking rules, but he's dealing with contradictory expectations that formalisation can’t fully resolve.

You see, workers work inside multiple overlapping systems. In practice they face what the role requires, what the team expects, what the customer needs, what the environment demands, what the system imagines the task to be, and what their embodied skill tells them is possible. These expectations often clash.

Safety professionals frequently ask: “Was this right or wrong?” A better question is: “Which systems and expectations was the worker trying to satisfy in that moment?” This reframes the debate entirely. Actions at the sharp end are often adaptations at the boundary where formal expectations and real conditions don’t match.

 

Differentiated organisations need formalisation and adaptation

Modern organisations rely on specialised subsystems such as operations, safety, HR, engineering, procurement, and compliance. Each has its own priorities, constraints and identities. Conflicts between them are how the system distributes complexity. Though formalisation holds these subsystems together so we can work safely at scale, no structure can capture every function the organisation must fulfil. Some work such as improvisation, emotional labour, and early detection of risk always stays informal and invisible. This is why procedures will never perfectly describe work, safety can’t be reduced to compliance, and adaptation will always be part of the job. The controversy around my post wasn’t about Jimmy, that was just my illustration. It was about this structural tension.

 

Why improvisation is a system property

Commenters said, “We shouldn’t rely on heroics.” They’re right, but adaptation is not heroism. It’s a built-in function of work in any environment that changes faster than procedures can be updated. Even in a well-designed system, workers will face ambiguity, plans won't fit reality, expectations will clash, and people will use embodied skill and practical wisdom to bridge the gap.

Of course this doesn’t excuse poor design, but it does explain why “just follow the rules” has never been an adequate description of how safety is produced.

 

Why the safety debate keeps repeating itself

People react differently to a photo of a worker because they are operating from different systems of expectation. While compliance thinkers see a breach, engineers see a lack of preparation. While practitioners see competence and judgement, leaders see reputational risk. And frontline workers see reality. None of these perspectives are wrong, but they belong to different subsystems with different priorities. The controversy is a feature of differentiated organisations.

 

So what does this mean for safety practice?

A few reflections from a systems perspective:

• Don’t confuse visible agreement with genuine alignment, since people agree outwardly because disagreement threatens membership.
• Expect role conflict, don’t moralise it, since being a good colleague and a good rule follower are not always the same.
• Build system trust, not dependent on individuals, since permits, competencies and escalation paths must survive personnel changes.
• Keep expectations elastic, since strong systems flex while brittle ones break.
• Don’t romanticise improvisation but don’t pretend it can be eliminated, since improvisation is what keeps the system working at its edges.
• Treat formalisation as a safety function instead of an administrative nuisance, since it creates shared expectations that allow distributed work. But bear in mind it will never remove the need for human judgement.

The point I tried to make

Jimmy’s photo is not an argument against formal systems, but an argument against imagining they are the whole system. Working safely emerges when formal expectations provide coherence, informal adaptations provide workability, and people move between them with dignity and skill.

If we reduce safety to compliance, we misunderstand both the system and the person. If we reduce it to heroics, we ignore the structure that makes work possible at all. A human-centred safety approach embraces both the formal system that stabilises expectations, and the human person who makes those expectations workable in the real world. That, I believe, is modern safety management.