When Presence Changes the System

What a controversial goal can teach us about patient safety and board decision-making

Why measurable evidence is never the whole picture, and why strong governance depends on triangulating hard data with weak signals, professional judgement and lived experience.

Introduction

I should begin with a disclaimer: I am not really a football supporter.

I enjoy watching football, particularly with family members who are far more invested in it than I am, but I do not follow a specific team. This gives me the luxury of watching a Manchester derby without having to spend the following week defending either the result or my emotional response to it.

It may also explain why, while everyone else was debating an offside decision, my mind wandered towards human factors, patient safety, situational awareness and board decision-making.

I appreciate that this might not be the traditional route into an article about healthcare governance. Bear with me.

A toe, a flag and a much more interesting question

During the Manchester derby on 13 September 2026, Erling Haaland scored what proved to be the winning goal for Manchester City. The goal was initially disallowed because Haaland was thought to be offside. VAR subsequently showed that he was onside, apparently by little more than the width of a defender’s foot.

That part of the decision was relatively easy to understand. Lines could be drawn, body positions compared and a categorical answer produced.

The more interesting question concerned Enzo Fernández.

Fernández was in an offside position in the middle of the penalty area. As the cross arrived, he attempted to reach it but did not make contact. Behind him was Lisandro Martinez, with Haaland positioned beyond them at the far post.

The initial VAR explanation focused on what could be clearly established. Haaland was onside and Fernández had not touched the ball. The goal was therefore awarded.

Later that evening, however, the refereeing body Pro Ref acknowledged an error of judgement. It said that VAR had not recognised the likely impact of Fernández’s position and should have recommended an on-field review. Martinez subsequently explained that he had moved towards Fernández because Fernández was in front of him, which contributed to Haaland being left free behind them.

For me, the interesting issue is not simply whether the final decision was right or wrong.

It is what the initial process appeared to count as evidence of influence.

Touching the ball is not the only way to influence play

A narrow assessment might ask:

·   Did Fernández touch the ball?

·   Did he make physical contact with Martinez?

·   Did he completely obstruct Martinez’s view?

·   Can we draw a line to prove his position?

A systems-based assessment asks something different:

·   How did Fernández’s presence shape Martinez’s perception of the situation?

·   Did it influence where Martinez positioned himself?

·   Did Martinez treat Fernández as an immediate threat?

·   Would Martinez have moved differently if Fernández had not been there?

·   Did Fernández’s presence contribute to additional space becoming available for Haaland?

Influence does not have to involve direct physical contact. A player can change the behaviour of a defender by moving into a dangerous space, beginning a run or appearing likely to receive the ball.

In fact, football teams deliberately exploit this. Players move to draw defenders out of position, overload one area or create space for a teammate. Their movement may be tactically valuable even when they never touch the ball.

It would be rather difficult to explain much of football coaching if we assumed that only the person who touches the ball influences play.

Presence changes the system

This reminded me of a principle from police conflict management: sometimes presence alone changes a situation.

A police officer does not have to speak, touch someone or use force to influence what happens next. Their presence changes the environment in which other people perceive the situation, assess their options and decide what to do.

Presence does not guarantee the desired outcome. It does not necessarily resolve the situation. Nevertheless, it becomes part of the system within which decisions are made.

The same principle applies to Fernández. He did not need to win the header to influence Martinez. His presence in a dangerous position created information that Martinez had to notice, interpret and respond to.

The question is therefore not confined to whether Fernández touched the ball. It is whether his presence changed what another person did.

Seeing the incident through situational awareness

Mica Endsley’s model of situational awareness gives us a useful way to examine this. The model describes three connected levels:

1. Perception of relevant information

2. Comprehension of what that information means

3. Projection of what is likely to happen next

At the level of perception, Martinez was aware of Fernández’s position. We can see that awareness through his own location and movement.

At the level of comprehension, Martinez appears to have understood Fernández as a credible threat. Fernández was an attacking player in a dangerous central position, close to goal and moving towards the path of the cross.

At the level of projection, Martinez anticipated where the ball might travel and attempted to intercept it. He did not reach the ball. Neither did Fernández. Haaland, however, was left with additional space behind them.

The final image tells us who touched the ball. The unfolding sequence tells us much more about who influenced the play. That distinction matters well beyond football.

A useful distinction

The final image tells us who touched the ball. The unfolding sequence tells us much more about who influenced the play.

From the football pitch to the patient’s bedside

In healthcare, influence is rarely confined to the person who performs the final visible action.

·   the information available at the time;

·   the design of a digital system;

·   staffing and workload;

·   interruptions;

·   competing priorities;

·   local practices;

·   the availability of senior support;

·   the layout of the working environment;

·   previous experience;

·   perceived expectations;

·   what colleagues say or do not say;

·   and what the clinician expects will happen next.

The eventual outcome may be recorded in a patient record, incident report or clinical system. Many of the influences that shaped it may never appear there.

The record might tell us which medication was administered. It may not tell us about the interruption immediately before it happened.

It might show when a patient’s observations were entered. It may not explain that the member of staff was looking after several deteriorating patients at the same time.

It may show that a referral was made later than expected. It may not capture the repeated unanswered calls, the ambiguity about who held responsibility or the way the perceived urgency of the situation changed over time.

If we investigate only what was documented, we risk treating the record as a complete representation of reality. It is not. It is one source of evidence created within that reality.

The danger of confusing measurement with understanding

There is a phrase often repeated in boardrooms: “What gets measured gets managed.”

A related version is: “If you cannot measure it, you cannot manage it.”

Both are frequently attributed to W. Edwards Deming. This is particularly unfortunate because Deming argued against managing organisations through visible figures alone. He warned that some of the most important figures required for management are unknown or unknowable, but that management must nevertheless take account of them. The W. Edwards Deming Institute specifically identifies “you cannot manage what you cannot measure” as a misattribution that conflicts with Deming’s thinking.

This does not mean that measurement is unimportant. Deming was a statistician. He was hardly opposed to data. The point is that data needs context, theory and an understanding of the system that produces it. A figure does not explain itself. Nor does the absence of an unfavourable figure demonstrate that there is no problem.

A board can receive a dashboard filled with green indicators and still preside over a deteriorating system.

The indicators might be accurate within their definitions. The problem could be that the definitions do not capture what people are experiencing, that the thresholds are insensitive to early deterioration, or that important concerns have not entered the formal reporting system at all.

The dashboard may contain correct answers to questions that are no longer sufficient.

What the past cannot tell us about the future

The work of Nassim Nicholas Taleb offers another important warning.

In The Black Swan, Taleb uses the example of a turkey that is fed every day. Each day provides additional evidence that the farmer is friendly and that tomorrow will resemble today. The turkey’s confidence grows with the amount of historical data available.

Unfortunately for the turkey, the historical pattern does not provide a complete understanding of the system.

The same evidence that appears to confirm safety can create increasing confidence immediately before the situation changes completely. Taleb’s broader point concerns the limits of induction, our vulnerability to rare events and our tendency to construct tidy explanations once an unexpected event has already occurred.

Healthcare boards can encounter their own version of the turkey problem:

·   “We have not had a serious incident of this type.”

·   “The risk has been on the register for years without materialising.”

·   “Complaints remain within tolerance.”

·   “Staff survey scores have not changed significantly.”

·   “Our mandatory training compliance is above target.”

·   “The dashboard has been green for six months.”

Each statement may be factually accurate.None proves that the system is safe.

Historical measures tell us about what became visible in the past, using the definitions and reporting arrangements that existed at the time. They cannot fully describe what has not yet happened, what has not been reported, or what the organisation does not yet know how to measure.

This is not an argument for abandoning quantitative evidence. It is an argument for humility about what quantitative evidence can tell us.

Weak signals and soft intelligence

Boards understandably want reliable evidence. They want trends, validated data and assurance that can be examined consistently.

However, emerging risks do not always arrive as validated datasets.

·   a concern raised informally by a clinician;

·   a change in the tone of staff conversations;

·   difficulty recruiting to one particular team;

·   repeated low-level complaints with no individual incident reaching an escalation threshold;

·   patients describing similar experiences in different words;

·   workarounds becoming normal;

·   growing unease among middle managers;

·   increasing dependence on a small number of experienced people;

·   or someone saying, “I cannot prove it, but something does not feel right.”

These are sometimes described as weak signals, soft signals or soft intelligence. The Care Quality Commission’s review of soft signals found widespread recognition of their value, while also identifying practical difficulties in collecting, sharing and triangulating them. Soft signals may arise through formal reporting, conversations, observation and other informal channels.

Research into employee voice in healthcare has also shown the danger of requiring concerns to be converted into hard, formal and tractable evidence before they are taken seriously. Some staff may hesitate to raise an intuition or emerging concern if they believe they must first assemble a complete evidential case. Attempts to make all soft intelligence “hard” can therefore suppress the very signals an organisation most needs to hear.

The fact that a concern is incomplete does not make it irrelevant.

A weak signal is not necessarily weak because it lacks importance. It may be weak because the issue is at an early stage, distributed across several parts of the organisation or not yet visible through established measures.

Triangulation, not a contest between hard and soft evidence

The answer is not to replace quantitative evidence with anecdotes. Nor should every uneasy feeling automatically become proof of systemic failure.

The answer is triangulation.

A board should bring together:

·   quantitative performance and safety indicators;

·   incident and near-miss information;

·   complaints and patient experience;

·   workforce measures;

·   Freedom to Speak Up information;

·   staff narratives;

·   observations of work as done;

·   internal audit and clinical audit findings;

·   regulatory intelligence;

·   operational pressures;

·   and direct conversations with people delivering and receiving care.

No single source gives a complete account.

Hard evidence can identify patterns, scale and variation. Soft intelligence can provide meaning, context and earlier indications of change. When several different sources begin pointing in the same direction, the combined picture may be more valuable than any single indicator considered alone.

This is not about giving experience and observation unquestioned authority. Experience without critical examination can reinforce poor assumptions. Deming argued that experience by itself is insufficient without theory, because theory enables us to ask questions and interpret what we observe.

Good governance therefore requires data, experience, observation, theory and challenge. It requires boards to ask not only, “What do the figures show?” but also:

·   What might these figures conceal?

·   How was this information produced?

·   Whose experience is absent?

·   What are staff and patients telling us?

·   Which concerns have not reached a formal threshold?

·   What assumptions are we making about the continuation of past trends?

·   What would we expect to see if the system were beginning to deteriorate?

·   Where could we look for disconfirming evidence?

·   What are we currently unable to know?

Board assurance

Board assurance should reduce uncertainty. It should not conceal uncertainty behind a green rating.

The seductive comfort of the outcome

Football supporters are excellent “couch coaches”. From the sofa, particularly with the benefit of slow-motion replays, multiple camera angles and knowledge of the outcome, the correct decision can appear obvious.

“Why did he go with that player?”

“Why didn’t he mark Haaland?”

“How could he miss that?”

“He should have known where the ball was going.”

These judgments focus on the output. They largely ignore the information available to the player, the speed of the situation and the thought behind the action.

Annie Duke describes a related decision-making trap as resulting. This is the tendency to judge the quality of a decision by the quality of its outcome. A good outcome does not prove that the original decision was good, just as a poor outcome does not demonstrate that the decision was unreasonable.

A defender can make a well-informed decision and still fail to intercept the ball.

A clinician can choose an appropriate treatment and still encounter an adverse outcome.

A board can make a defensible decision under uncertainty and later discover that events developed differently.

Conversely, a risky or poorly informed decision may produce no harm at all. If we judge only the outcome, we may inadvertently reward unsafe decision-making because the organisation happened to be lucky.

That is why near misses matter. They allow us to examine the quality of the system without waiting for harm to provide the motivation.

Our mistakes and their brilliance

Attribution theory helps explain how people assign causes to success and failure. We may locate causes internally, such as ability, effort or strategy, or externally, such as luck, officials, environmental conditions or the actions of an opponent. These explanations also vary according to whether we see the cause as stable or changeable, and controllable or uncontrollable.

Sports discussions often contain familiar examples:

·   We lost because the referee made a poor decision.

·   They won because they were lucky.

·   We won because we showed character.

·   They lost because they were not good enough.

But there can also be an interesting inversion.

Sometimes supporters blame their own team because doing so preserves a sense of control.

“They did not beat us. We gave the game away.”

“They did not create the goal. We made a mistake.”

“They were not brilliant. We were poor.”

If our failure caused the outcome, then correcting our failure appears to restore control over the future. Acknowledging the opposition’s skill is less comfortable because it means accepting that another team can influence events in ways we cannot fully control.

The same can happen in organisations after an incident. It may feel more manageable to conclude that one member of staff failed to follow a procedure than to recognise that a complex system produced interacting conditions that no individual could completely control.

Individual blame offers a clear cause, a visible action and a reassuring remedy.

Retrain the person. Remind the team. Rewrite the procedure. The organisation can then announce that the problem has been addressed.

Unfortunately, a simple explanation is not necessarily an accurate one.

Slips, mistakes and the expectation of perfection

Don Norman distinguishes between slips and mistakes. A slip occurs when the intention is appropriate but the action is not executed as intended. A mistake occurs when the goal, plan or understanding of the situation is wrong. The distinction matters because the two require different responses. Slips call for design that makes the correct action easier and errors easier to detect or reverse. Mistakes require better information, clearer mental models and feedback that supports more accurate understanding.

Blame culture often fails to distinguish between them. It observes that the action was wrong and concludes that the person was careless. The proposed remedy is usually to “be more careful”, despite highly trained and committed people already trying to do exactly that.

Elite footballers practise repeatedly and perform at extraordinary levels. They are still human. A foot can be placed slightly differently from intended. Attention can be drawn towards one threat rather than another. An interception can be missed by a small margin.

Healthcare professionals are also highly trained. They are still affected by fatigue, interruptions, workload, ambiguity, poor interfaces and the limitations of human attention and memory.

Expecting perfection from human beings is not a safety strategy.

Well-designed systems anticipate variability, make important information visible, support detection and recovery, and reduce the likelihood that a single slip can produce serious harm. Norman’s work challenges us to move from asking why a person failed to perform perfectly to asking why the system required perfect performance in the first place.

Seeing the situation from inside the tunnel

After an event, we stand outside the tunnel. We know the outcome. We can replay the sequence. We can pause at the important moment and identify the information that now appears significant.

The people involved did not experience the situation that way. Sidney Dekker’s work on local rationality encourages us to understand why actions made sense to people at the time, given their goals, knowledge, attention and constraints. His “new view” of human error moves away from the belief that otherwise safe systems are undermined by a few unreliable individuals. Instead, apparent human error becomes the beginning of an investigation into how the wider system shaped performance.

This does not mean that every decision was correct. It does not remove accountability or suggest that intention is the only thing that matters.

It means that we cannot improve a system until we understand why the action seemed reasonable from inside the situation.

There is a substantial difference between asking:

• Why did they do something so obviously wrong?

and asking:

• What did they see, understand and expect at the time, and what made this course of action appear reasonable?

The first question invites defence.

The second creates the possibility of learning.

Why PSIRF learning responses matter

The NHS England Patient Safety Incident Response Framework, or PSIRF, represents a move towards system-based learning, compassionate engagement, proportionate responses and supportive oversight. It asks organisations to understand how incidents happen, including the factors that contribute to them, rather than defaulting to a narrow search for a root cause or an individual failure.

This is why well-facilitated learning responses matter. A clinical record, policy or electronic audit trail may tell us what was documented. It cannot, by itself, recreate the working environment or explain how the situation appeared to those involved.

A learning response needs to bring together formal evidence with the insider perspective.

An After Action Review can provide a structured opportunity to explore:

·   What was expected to happen?

·   What actually happened?

·   Why was there a difference?

·   What can we learn?

When facilitated well, an AAR is not a search for the person who got it wrong. It is a way of reconstructing the event from the perspectives of those who experienced it and identifying opportunities to improve the system. PSIRF recognises AARs, swarm huddles, multidisciplinary team reviews and patient safety incident investigations as proportionate forms of system-based learning response.

The words “when facilitated well” are important.

An AAR is not psychologically safe simply because it is called an AAR. Poor facilitation can turn a learning response into a group interrogation or a retrospective assessment of who failed to meet an idealised procedure.

Good facilitation helps people describe:

·   what they noticed;

·   what they were trying to achieve;

·   what pressures were present;

·   what information was available;

·   what information was absent;

·   which assumptions seemed reasonable;

·   and how the system supported or hindered recovery.

It helps those outside the tunnel understand what it was like inside.

What boards can learn from VAR

VAR had something the defender did not: time.

It had multiple angles, slow-motion footage, technology and a team of officials able to review the event away from the immediate pressure of play.

Yet the initial decision appears to have concentrated on what was easiest to establish categorically:

·   Haaland was onside.

·   Fernández did not touch the ball.

Both facts were relevant. Neither was sufficient.

The wider question required judgement about influence, attention, movement and the interaction between players. Boards can fall into the same trap.

They have dashboards, reports, audits, risk registers and performance indicators. These are valuable tools. However, if they focus only on the evidence that can be reduced to a number, threshold or rating, they may overlook how the system is actually changing.

A board should not reject measurement. It should use measurement intelligently, while remaining curious about what the measures cannot show.

From comfort-seeking to problem-sensing

Green dashboards can be reassuring. Sometimes too reassuring. Healthcare research has distinguished between comfort-seeking and problem-sensing approaches. Comfort-seeking looks for evidence that confirms the organisation is performing adequately. Problem-sensing actively searches for vulnerabilities, fallibilities and information that challenges the preferred account. Conventional metrics remain important, but they may not reveal the full range of risks within an organisation.

A problem-sensing board might ask:

·   Where are we most dependent on staff compensating for poor system design?

·   Which services appear safe because experienced people are continually rescuing them?

·   What concerns are repeatedly described as isolated?

·   Where is demand growing faster than our formal measures reveal?

·   What are patients and families noticing before we see a measurable trend?

·   Which indicator would only turn red after harm has already occurred?

·   What evidence would challenge our current assurance?

·   Are staff able to raise concerns before they can prove them?

·   What are we not hearing because people believe the concern is not yet serious enough?

These questions do not replace the dashboard. They make the dashboard more meaningful.

The real lesson from the goal

It would be easy to finish by saying that VAR got the decision wrong.

That would be true according to the subsequent acknowledgement from Pro Ref, but it would also risk repeating the same outcome-focused thinking this article has questioned.

The more useful lesson is about how decisions are made.

VAR initially favoured evidence that was categorical, visible and measurable. It established whether a player was onside and whether another touched the ball. In doing so, it overlooked information about influence that required a broader interpretation of the system.

Healthcare organisations and their boards face the same risk.

We can measure incidents, waiting times, complaints, vacancies, mortality, training completion and dozens of other indicators. We should measure them, understand variation and use them to support improvement. But we should not confuse the availability of a measure with the importance of the thing being measured.

Some of the earliest indications of risk will not arrive as statistically significant trends. They will arrive as stories, observations, workarounds, discomfort, professional intuition and disconnected pieces of information that have not yet become a recognised pattern.

We need hard evidence.

We also need soft intelligence.

Most importantly, we need the curiosity, capability and governance arrangements to triangulate them. Sometimes the most important influence is not a touch, a collision or a measurable outcome. Sometimes it is the presence of something that changes what everybody else does next.

The challenge for boards is whether they are sufficiently connected to notice it.

And, apparently, the challenge for my family is accepting that I cannot watch a football match without turning it into a discussion about patient safety governance.

Questions for boards and senior leaders

1. Does our assurance process encourage problem-sensing, or primarily provide comfort?

2. How do soft signals reach the board before they become serious incidents?

3. Do we require staff to prove concerns before we are prepared to hear them?

4. How do we triangulate performance data with staff, patient and family experience?

5. Which of our measures are leading indicators, and which only describe outcomes that have already occurred?

6. When reviewing an incident, do we reconstruct the insider’s perspective or judge actions from outside the tunnel?

7. Can we distinguish between a slip, a mistake, a poorly designed system and genuinely unacceptable conduct?

8. How do we assess the quality of a decision separately from the quality of its outcome?

9. What currently appears safe simply because people are successfully compensating for system weaknesses?

10. What should we be curious about that is not included on the dashboard?

References

College of Policing (2020) Conflict management using de-escalation, communication and negotiation: Guidelines. Ryton-on-Dunsmore: College of Policing. Available online

Deming, W.E. (1986) Out of the Crisis. Cambridge, MA: Massachusetts Institute of Technology, Center for Advanced Engineering Study.

Deming, W.E. (1994) The New Economics for Industry, Government, Education. 2nd edn. Cambridge, MA: MIT Press.

Dekker, S. (2014) The Field Guide to Understanding “Human Error”. 3rd edn. Farnham: Ashgate. Available online

Duke, A. (2018) Thinking in Bets: Making Smarter Decisions When You Don’t Have All the Facts. New York: Portfolio.

Duke, A. (2020) How to Decide: Simple Tools for Making Better Choices. New York: Portfolio.

Endsley, M.R. (1995) “Toward a theory of situation awareness in dynamic systems”, Human Factors, 37(1), pp. 32–64. Available online

International Football Association Board (2026) Laws of the Game 2026/27: Law 11, Offside. Zurich: IFAB. Available online

Klein, G.A. (1993) “A recognition-primed decision model of rapid decision making”, in Klein, G.A. et al. (eds.) Decision Making in Action: Models and Methods. Norwood, NJ: Ablex, pp. 138–147.

Martin, G.P. et al. (2018) “Making soft intelligence hard: a multi-site qualitative study of challenges relating to voice about safety concerns”, BMJ Quality & Safety, 27(9), pp. 710–717. Available online

NHS England (2022) Patient Safety Incident Response Framework. London: NHS England. Available online

NHS England (2022) Patient Safety Learning Response Toolkit. London: NHS England. Available online

Norman, D.A. (2013) The Design of Everyday Things. Revised and expanded edn. New York: Basic Books.

Taleb, N.N. (2007) The Black Swan: The Impact of the Highly Improbable. London: Allen Lane.

Weiner, B. (1985) “An attributional theory of achievement motivation and emotion”, Psychological Review, 92(4), pp. 548–573. Available online

Care Quality Commission (2024) Identifying, using and acting upon soft signals: a rapid literature review. Available online

Professional Game Referees / match reporting (2026) Clarification regarding Erling Haaland’s awarded goal in the Manchester derby, reported 13 September 2026. Available online