Black Swans, Hindsight Bias and the Limits of Risk Assessment

Importantly, several scholars and institutions have questioned whether so‑called Black Swan accidents are truly unforeseeable. Studies of natural‑hazard‑triggered technological accidents (Natech events) suggest that many could have been anticipated using existing scientific knowledge but were excluded from risk assessments because they lay outside design assumptions or regulatory scope. A recent review by the European Commission’s Joint Research Centre concluded that most recorded Natech accidents could not meaningfully be described as Black Swans, because relevant hazard information and experience already existed. From this perspective, Black Swans in industrial safety are less about inherent unpredictability and more about organisational blind spots and risk blindness. Disasters occur not simply because information is unavailable, but because it is discounted, normalised or rendered invisible by prevailing assumptions, commercial pressures or fragmented governance.

Overcoming the trap of ‘that’s never happened before’ therefore requires a fundamental shift in how risk assessment is conceived and practiced. One important development is the explicit consideration of uncertainty, rather than its concealment within numerical likelihoods that appear more precise than the underlying knowledge justifies. The latest thinking in risk management argues for distinguishing more clearly between risk and uncertainty, emphasising the strength and limitations of the knowledge that underpin risk judgements and making ignorance visible rather than implicit. By acknowledging what is not known, risk assessments can better inform decision‑makers about the limits of confidence, rather than presenting a false sense of completeness or control.

Closely related to this is the contribution of High Reliability Organisation (HRO) theory. Research into organisations such as air traffic control centres, nuclear power operations and aircraft carriers highlights the importance of ‘collective mindfulness’, a sustained preoccupation with failure and reluctance to simplify interpretations. Weick and Sutcliffe argue that HROs treat success as provisional and actively search for weak signals that may indicate emerging risk, even when everything appears to be going well. Rather than assuming that absence of accidents implies safety, HROs question routine assumptions, remain sensitive to operations and defer to expertise closest to the work. This mindset directly challenges the complacency embedded in the phrase ‘it’s never happened before’, replacing it with the more uncomfortable but more useful question: ‘what would surprise us most?’

From this angle, resilient systems are characterised by their ability to anticipate, monitor, respond and learn across a wide range of conditions, rather than by their adherence to a fixed set of controls. The practical implication is that the goal of risk assessment is not to identify every conceivable accident scenario, but to understand system behaviour under stress and to design margins, buffers and adaptive capacity that help the organisation absorb and recover from surprise.

Practical techniques can support this broader conceptual shift. Scenario‑based approaches that deliberately explore implausible or extreme conditions can help counteract the normalisation of risk and the tendency to anchor on the familiar. Techniques such as ‘what‑if’ analysis, red‑teaming and imaginative stress testing encourage assessors to step outside routine assumptions and challenge the boundaries of credible failure. These methods are particularly powerful when they draw on diverse expertise, including frontline workers who possess tacit knowledge of how work is actually done, rather than how it is imagined in procedures. Risk rarely exists only where the procedure says it is. It is where workers must make trade‑offs under pressure, in situations that are overlooked or only partially understood.

At Finch, where we routinely test the adequacy of our clients’ risk assessment programmes in the context of major loss events and regulatory scrutiny, these ideas have practical consequences. They point towards methods and conversations that interrogate uncertainty, probe assumptions about ‘credible’ scenarios, and pay closer attention to the dynamic adaptations that keep systems functioning on a day‑to‑day basis. They also suggest a shift in emphasis when engaging boards and senior leaders, moving from assumed assurance based on historical operating frequencies, to a more candid discussion about knowledge gaps, black spots in organisational awareness and the capacity to cope when things do not behave as expected.

Many catastrophic industrial accidents retrospectively described as Black Swans are therefore not the product of pure randomness, but of constrained imagination, cognitive bias and methodological limitations in risk assessment. Hindsight bias creates the illusion that such events were obvious, while traditional probabilistic models can obscure uncertainty and reinforce reliance on historical performance. By putting uncertainty at the centre of our thinking, and by cultivating organisational mindfulness and resilience, safety practitioners can move beyond the comforting but hazardous assumption that the absence of past failure equates to future safety.