The difficulty of predicting the future
On April 11, 2001, Defense Secretary Donald Rumsfeld sent a memo to President George W. Bush and Vice President Dick Cheney that might never have seen the light of day but for its timing.
“I ran across this piece on the difficulty of predicting the future,” wrote Rumsfeld. “I thought you might find it interesting”. The “piece” was a short article written by Linton Wells that examined the political situation at the start of each decade in the 1900s. In every case, there was a startling change from ten years earlier.
Exactly five months after Rumsfeld wrote this memo, two planes flew into the Twin Towers in New York. This shaped the following decade in ways few could have imagined at the turn of the millennium. Even fewer could have foreseen the onset of the Global Financial Crisis or the smartphone revolution ignited by Steve Jobs in 2007.
It is quite clear that economic or geopolitical forecasters cannot accurately predict events ten years before they occur. Humanity progresses in rapid step-changes rather than gradual linear growth. Events also do not occur in a vacuum, but instead influence the future through complex cause-effect relationships.
And yet, this sort of forecasting is common, even amongst institutions that should know better. Every four years, Congress requires the Department of Defense to produce a twenty-year forecast of the natural security environment- the Quadrennial Defense Review (now the National Defense Strategy). It was this exercise that prompted Lin Wells to write the above article as a form of intellectual rebellion.
However, Wells’ last remark points to a better way. If you have to plan far into the future, plan for surprise. Work from the assumption that reality will catch you completely off-guard and think about how to respond.
“Plans are useless. But planning is indispensable”
— President Eisenhower
Nassim Taleb has taken this idea one step further, calling for critical systems- such as the global banking system and nuclear weapons- to be “anti-fragile”. That is, they should not just be resilient to shocks but made even stronger by them.
To those of us living through 2020, Wells’ paper could not ring any truer. Who would have thought at the end of 2019 that we would see fires ravage Australia, a global pandemic take hundreds of thousands of lives and the constantly simmering racial tensions in the US reach a violent boiling point?
Despite the unpredictability of these events, it is clear that our systems were not designed to withstand these crises. At the onset of the COVID-19 pandemic, hospitals could not access vital PPE when they needed it most and supermarkets ran out of everything from toilet paper to pasta.
Michael Pollan wrote an excellent piece on how increasing specialisation in the meat industry has decimated the resilience of the food supply chain in the USA. In the name of efficiency, farmers and agricultural companies have opted for scale rather than adaptability. When huge meat-packing plants run by giants like Tyson and Cargill were shut down, there was a severe meat shortage in supermarkets. However, at the same time, farmers were euthanizing millions of chickens as they only specialised in niche products such as liquid egg whites, and could not rise to fill the gap in demand. The supply chain was simply not ready for such a shock.
Does this mean that all predictions are futile, and that the only thing we should do is hunker down in preparation for bad things to happen?
Not at all. Preparing for surprises is costly. In light of this, we have to set priorities, which brings us back to the business of forecasting. Consider building codes in Tokyo. Tokyo’s skyscrapers have to be constructed with advanced engineering to withstand massive earthquakes. This is a lot more expensive, but makes sense given the frequency of large earthquakes in Japan. Incurring this expense would make less sense in a relatively calmer city like Sydney or New York.
These kind of probability estimates lie at the heart of all long-term planning, but are rarely as explicit as in earthquake preparation. For decades, the US had a policy of maintaining capacity to fight two wars simultaneously. But why stop there? Why not three, four or five wars? It comes back to probability. The two-war doctrine was based on a judgement that the likelihood of fighting two wars justified the massive cost.
Judgements like these are unavoidable. And if it looks like we have avoided them, we have simply swept them under the rug. This is even more worrying. Probability judgements should be made explicit, so we can understand all our assumptions and best judge how accurate they will be. If we are just guessing, it is best to say so.
“Knowing what we don’t know is better than thinking we know what we don’t”
— Philip Tetlock
Although it may not feel like it, 2020 is only at its halfway point, with a US Presidential election still to come. Looking forward to 2030, the only thing we can be certain of is uncertainty.