In the early morning February first, 1953, a mighty storm in the North Sea, coupled with extremely high spring tides, led to a breach of the dikes protecting the southwest part of the Netherlands. The flooding led to the tragic loss of 1,835 lives and ~30.000 animals. Approximately 200.000 hectares of land were swamped, and more than 45,000 buildings and 300 farms were destroyed. Consequently, cattle died, and crops failed to yield. The total damage amounted to about 5% of the country’s GDP, a devastating blow to a nation still in the process of recovering from World War II. 

The disaster was the result of an unusually powerful storm and an inability to conceive and prepare for it. Both the Dutch Government and the people relied firmly on their infrastructure-based protection; a system of dikes widely perceived as infallible. The risk of failure was commonly perceived as trivial.  

However, with the collective awareness of the country’s vulnerability refreshed, the government gradually shifted its focus from disaster response and immediate dike repair to a plan of action to ensure that such a tragedy would never happen again. In 1957, the Delta Works project began, aimed to block and control the entry of seawater. The objective was clear and specific: To protect the country from flooding due to storm surges and high tides. 

By placing excessive emphasis on one specific risk factor — protection against coastal flooding — other potential risks were inadvertently given less attention.

Unfortunately, the possibility of flooding may also occur from rivers carrying meltwater from the Alps. Add heavy rainfall and we get a scenario that materialized in 1995. 10 years after the completion of the Delta Works, the water levels surged to a critical point, posing a significant risk of breaching the river dikes. To prevent this, the Dutch authorities decided to discharge excess water into the sea.

However, due to the dike design, built to block inflow of water, high tides and strong winds, the water from the rivers was unable to flow into the sea effectively. 

Hence, the escalating water levels in the rivers continued to rise, putting immense pressure on the dikes resulting once again in widespread flooding across multiple regions. 250.000 people and 1 million animals were evacuated. Although the flood protection system did not completely collapse, the event exposed the need for adjustments from the existing efforts to robustify parts of the infrastructure.  

The initial economic cost of the flood was estimated at 1 billion euros, but this figure continued — in tandem with the water — to rise. In response, the Dutch government launched the Room for the River program. This initiative aimed to adopt a more comprehensive and resilient approach to flood protection, by providing rivers with ample space, thereby mitigating the risk of future flooding. 

The evolution of the Dutch water projects illustrates one of the major changes in strategy in this century, in both corporate and government, the switch from robustness (try to prevent failure) to resilience (early detection and fast recovery). In 1953, when the storm loomed and the first indications of a potential flood emerged, people mistook these warnings for ordinary weather events that happened every winter. Once the extent of the danger was apparent, it was too late to act. Neither detection capabilities nor a system built for resilience was in place. 

Risk has traditionally been assessed based on a bell curve Gaussian distribution. This is a pragmatic approach to risk management which allows organizations to allocate resources efficiently by focusing on preparing for — and managing — the most likely outcomes, rather than trying to account for every possible scenario.

The bell curve helps visualize the distribution of potential outcomes or events. If we do probability assessments, we are expected to manage anything which falls within the expected probability distribution. Anything which falls outside, however, is called a black swan in popular language, or otherwise known as a low probability or an outlier. Something which is so unexpected or so unusual, it's not reasonable to expect planning processes to handle it. 

A problem with applying normal distribution is that so-called low probability events seem to be happening more frequently of late. For instance, since Finance Norway began statistics on weather and natural events in 1980, seven of the ten largest events have occurred after 2010.

In nature we more frequently see Pareto distributions, not Gaussian distributions. Thus, the so-called low probability events have turned into a medium probability, which dramatically changes the strategic agenda.  

If we take essentially any example from nature or a social system; The most frequent word occurs about twice as often as the second most frequent; There are a large number of fleas, but a small number of elephants; and there are a large number of villages, but a small number of cities. Zip’s law states that in a frequency-ranked list, an item's frequency is inversely proportional to its rank. The most frequent item appears twice as often as the second, three times as often as the third, and so on. Economically, we see that a small percentage of the population holds a large proportion of the total wealth. According to the Pareto principle, 20% of all people receive 80% of the income, and 20% of the most affluent 20% receive 80% of that, etcetera and so forth.

These outcomes arise from complex interactions and feedback mechanisms, leading to the emergence of patterns characterized by a small number of dominant elements and a large quantity of smaller ones. When examining statistical patterns using a double log scale to represent size against frequency, a notable observation emerges: Power laws tend to unveil across a wide range of phenomena. 

If we, as a thought experiment, take a statistical representation and overlay a Pareto distribution on the Gaussian model, the curve has a fatter tail, with a much higher frequency of rare events than a Gaussian curve would predict. The number of events we must plan for has grown so large that it is no longer feasible to manage them all effectively. Consequently, both business and risk managers face daily challenges in conducting fail-proof planning within an increasingly complex environment.

We must transition from systems focused solely on robustness to designing systems for resilience, recognizing that failure is inevitable. This shift fundamentally alters the strategic agenda. However, when approached thoughtfully, acknowledging failure becomes a valuable opportunity. By reframing failures as learning experiences, we can develop resilient strategies that prioritize early detection, rapid recovery, and the ability to capitalize on new opportunities. This approach is commonly seen in entrepreneurial environments, such as Silicon Valley, where experimentation and learning from failures are actively encouraged and valued.

A few start-ups and new product launches succeed, though the majority fail. The issue for large companies and governments is how to reap the ability and entrepreneurial capacity to exploit new opportunities early, capitalize on the failure of traditional industries, and figure out how to get that insight into their organizations. In essence, this is what resilient strategy is all about. Yet, institutional change tends to be a strenuous effort, and obsolescence can ultimately define the outcome of even a promising transformational process.

Traditional methods, such as the Gaussian statistical universe, work quite well when dealing with a large number of probable events, with a many-to-many relationship. We can say: If we do [this], it will produce [such] a result with a high probability of success. That's the world we all would like to live in. Unfortunately, a large part of our business environment isn’t like that. Thus, events in the long tails are often more important to monitor than worrying about the average

As we move towards the border between Gaussian and Pareto universes, the range of possible outcomes becomes difficult to assess in terms of probability. Traditionally, scenario planning and expert opinions have helped us navigate in this realm of possibilities, but they have limitations in predicting events that fall outside past experiences.

Further into the tail territory of uncertainty, events we used to treat as non-probable become plausible, and traditional methods will fail to provide reliable insights. 

I.e. in a Pareto world, strategies cannot be based on probability assessments. Hence, hypothesis-based research and deductive reasoning (“top-down”) become ineffective. It can even blind us. Remember what H.L. Mencken thought us some 100 years ago: For every complex problem, there is an answer that is clear, simple and wrong. Instead, we must: 

• Adopt a null hypothesis approach — a type of statistical hypothesis that proposes that no statistical significance exists in a set of given observations 
• Engage in abductive reasoning — a form of logical inference that seeks the simplest and most likely conclusion from a set of observations, and  
• Cultivate heightened awareness — stay curious, rather than judgmental 

In this endeavor, technology plays a crucial role in augmenting human intelligence, enabling us to capture and realize the significance of unusual happenings in vast amounts of data.