Sometimes you run into something that you wish you had discovered a bit earlier in your journey, something that could have helped you avoid doing stupid things or, let’s say, optimize your life.
For me, the latest examples were Kahneman’s “Thinking fast, Thinking Slow” and Taleb’s “ Black Swan” which I discovered a few years ago.
Both books deal with cognitive biases which have a huge impact on the way we make decisions, act or make predictions. The problem here is that we are not always aware of their existence or their importance. As human beings, we “have” to believe that free will exists and guides all our choices and actions while, in reality, our behaviors are largely impacted by external factors merely brought to our consciousness and leaving us only one choice: to do or not do.
The list of cognitive biases would be too long to detail. And you can find them easily on the web or by reading Kahneman’s book.
I wanted to spend some time on one aspect which is prediction (or forecasting).
Human beings have the ability with the consciousness to project themselves in the future, to imagine scenarios and act so as to reach a future state which they desire. Predicting comes with the need to plan and to gain control over one’s life because uncertainty is not something mankind loves very much. But forecasting just gives the illusion of control: it’s not because you write something down that it will happen. In fact, it’s quite the opposite: as soon as something is put on paper, life will choose another way. Therefore, we come back to the famous idea that the one who survives is the one who adapts to change.
In his book, Nassim Nicholas Taleb describes the example of oil price forecasting (professional distortion). He writes:
« One anonymous person (he is employed by a governmental agency) explained to me privately after the talk that in January 2004 his department was forecasting the price of oil for twenty-five years later at $27 a barrel, slightly higher than what it was at the time. Six months later, around June 2004, after oil doubled in price, they had to revise their estimate to $54 (the price of oil is currently, as I am writing these lines, close to $79 a barrel). It did not dawn on them that it was ludicrous to forecast a second time given that their forecast was off so early and so markedly, that this business of forecasting had to be somehow questioned. And they were looking twenty-five years ahead! »
Among the problems of forecasting, one of the biggest problems is that we don’t know what we don’t know.
This is especially true with “experts” who tend to be trusted more over simple passers-by: the self-delusion of knowledge and the abundance of information doesn’t always help to make more accurate or better decisions.
More information builds confidence, that’s for sure, but also complexity. As the number of variables grows, it becomes more difficult to assess which parameter has a real impact on the results you’re observing, or if they are intertwined or useless. And we even assume here that all parameters and their relations are known, which is not the case: ignoramus et ignoramibus: we are ignorant and will remain so.
Taleb talks about examples of predictions made by CFOs, economists or bookmakers which were less accurate than the ones made by taxi drivers. So, it would be good to question the error rate of said experts and their predictions. That’s something we don’t see often enough: an error rate. From his perspective experts are mathematicians, accountants, astronomers, physicists or livestock judges. As for non-experts: stockbrokers, economists, financial forecasters, finance professors, political scientists, “risk experts”.
That’s a topical issue today in a world where the word “data” has stormed the stage and where everyone seems to spend a lot of energy and money on all sorts of predictions: from epidemics to economic growth, influencing greatly all decision-makers with a degree of confidence and certainty which sometimes defies common sense. While mathematics is a great tool when used and understood properly, it also contains enough base for caution. For example, Taleb summarizes how Poincaré showed in his Three-Body Problem that you need an increasing amount of precision about the dynamics of the process you are modeling since the error rate grows very rapidly. Small prediction errors can result in a large divergence from the initial scenario. Be careful when you round numbers in your Excel table.
Talking about tables, anyone attempting to make predictions should be aware of the anchoring effect. It basically tells you that the first number you will see or hear will be the base for any future discussion on the topic. Imagine you want to buy a pair of shoes and you didn’t fix a maximum price beforehand. You walk into a market, see a pair you like and ask the vendor “how much for this one?”. He replies “95 dollars”. This number becomes anchored in your mind and if you want to negotiate, you might have a hard time starting, say, from 10 dollars (which might be their real worth). Instead, you are more likely to start around 65 or 70 dollars. To avoid this, you can rephrase the question to “how much am I ready to pay for this pair?”. This might save you a few coins.
The same applies for every prediction based on numbers. Assume you want to forecast your sales for the next 3 years; how do you start? Well, usually by looking at this year, and then making assumptions. That’s a starting point. But is it the right one? Would it be better to look at the world now and its uncertainties and start fresh? And what is your error rate? Did you consider uncertainties like, for example, a worldwide epidemic?
So, as Taleb wrote, forecasting seems to be used more as anxiety relief rather than efficient decision-making. The question is then what brings more anxiety: anticipating change or having to deal suddenly with it?
Another interesting point made by Karl Popper is that to predict historical events you have to predict technological innovation. And it’s not possible. To paraphrase Taleb, if you expect to expect something in the future, then you already expect something at present or, « to understand the future to the point of being able to predict it, you need to incorporate elements from this future itself ».
The problem with technological innovation and breakthrough is that they are not predictable: the person who invented the wheel or discovered Gravity or the Cosmic Microwave Background were just not sitting there and planning to invent or discover exactly it. It was serendipity. It just happened.
(I could tell you what I am going to invent tomorrow, but I don’t know myself. I’ll let you know when it happens, assuming I realize I created something worthy to be called an invention.)
Prediction or forecasting are very risky businesses. Even if planning is a part of what being human means, we should always bear in mind that, when we plan something, the probability of an event to happen exactly as predicted is much lower than the one of it to never happen. That’s called life.
What to do then? Here, with an example: you can’t predict the occurrence of Earthquakes, but you can try to protect the community from possible damages.
And as holiday season is approaching, I will finish this post with a personal forecast: the weather at your destination will be sunny during your stay with 50% chance of rain in the afternoon and a temperature averaging 21 degrees Celsius ± 10 degrees.
Good luck picking your clothes.