Prediction, forecasting, projection; these words represent an interesting concept. That of deciding what the future will hold. Prediction has been part of human life for, well probably as long as there have been modern humans (thousands of years). It is quite easy to state this because there are records and indicators which show that this statement is true. The concept itself however, prediction, cannot be used in the same way as statements about the past.
In our current society, we seem to have opted for a collective misunderstanding about prediction. We seem to have neglected to understand that any and all prediction is based on probability. Some things are more likely to happen than others but they are not certain. Some things are so probable that we might even think of them as certainties, but they are not. If I hold a coin in my hand and drop it, it will certainly hit the floor. Well, there is an exceptionally high probability that it will hit the floor, but it is not certain.
It seems that even the word certain is associated with prediction. The word certain is quite frequently used in association with predicting that things will happen. Yet the word, certain, used in this context, is incorrect and really should not be used. Nothing about the future is certain even though some things are very, very probable.
We don’t have to look too hard to see examples of our misunderstanding of prediction. Politics is full of it. Politicians claim that particular things will certainly happen at some stage in the future. The politicians are encouraged to make such predictions of certainty by journalists who want to be able to hold someone to account if things don’t actually turn out as predicted. We can all blame someone when things do not work out as predicted.
Prediction is used in this context in an even richer way. Take any major incident which is unfolding; my current example has to be the pandemic because I write this in January 2021. Politicians are encouraged to make predictions about future states and events based on their current actions and proposed actions. When a set of proposed actions is made public, and since there are many millions of people who are involved (the population including the population of experts), there are likely to be almost as many predictions as possible outcomes, good and bad. We just need to ask enough people (experts maybe) for their opinion and it is quite likely we will uncover wide disagreement and of course some agreement concerning any current proposal for activity.
This now sets the scene for an interesting outcome. If the politician’s prediction was correct, meaning that many predictions were incorrect, we simply ignore this, there is no story in it. If the politician’s prediction turns out to be incorrect then we can now assemble a rather good story. We can certainly point out the error. This is the mismatch between the prediction and the actual outcome which unfolded over time. We can also find experts who disagreed with the previously proposed actions from the population of those who disagreed. Furthermore, we can even find experts who predicted the actual outcome, there will almost certainly have been some given there were so many differing predictions at the outset. Even if we can’t find an expert who predicted the actual outcome, we will be able to find more than one who got closer to the actual outcome than the politician and the politician’s set of expert advisors. This now makes a rather good story and is also a surprisingly appropriate example of something else, our culture of blame.
The culture of blame is an important thing to note here because this offers quite a good explanation about why we collectively choose to ignore the actual meaning of prediction.
Nothing about the future is certain. Yet we do need to make predictions and we do need to plan actions which if carried out as planned, are highly likely to bring about a predicted event or state. To understand why a proposed set of actions is likely to lead to a future state, we need to understand the basis of the prediction and associated actions. We need to understand that the prediction has a probability associated with it, not a certainty. We should also understand that there are quite likely to be very many alternative predictions and associated actions and part of our choice is to select what we think is the best one.
It is also true that when a set of actions is undertaken over time in a highly complex environment, there will be unintended results of the actions and quite possibly unpredicted results of those actions. We can try to mitigate against this by thinking carefully what these unintended outcomes might be at the start but we might not think of them all. Sometimes, as with the pandemic, we might plan actions which are known to lead to rather poor outcomes because we believe that all outcomes are likely to be unfavourable but the proposed actions lead to an intended outcome which is less bad than some other potential outcomes. In such cases of course, you really can’t win.
Of course, we hope our leaders take advice (and act upon it) from the very best sources but even this is a choice of a few from many. Planning outcomes in multi variable environments can be very complex. Our actions might fix one problem, only to cause another which might turn out to be just as bad or worse.
This discussion has not proposed any solution to prediction. It has only suggested that we all should better understand what prediction actually is. If we were all clearer about this, we might even learn how to make better predictions and plan better actions. Yet we will never predict with certainty.
John L. Gordon – January 2021