Prediction paradoxes
On any given future event of significance, be it economic, political, or social, it is certain that a number of prophets will foretell any of the N possible outcomes -- and with equal certainty, after the event comes to completion, there will be T of them who will claim the often unrewarding "I told you so".
That brings about the two paradoxes that I had in mind for this reflection. The first one is the Cassandra paradox, where, for some rather occult reason, people repeatedly fail to heed some prophecy or another (and why should they, after all, prophecies are somewhat occult themselves, are they not?) The second paradox is what I would call the "minority report" paradox, in honor of Philip K Dick's story that intimately relates to it: if I believe a prediction to be true and I dislike the outcome, I will take actions to prevent it from happening, which will invalidate the prediction. Then a second, divergent prediction might be correct, because it accounts for me knowing of the first prediction. Unless I also know of the second prediction and act upon it as well. And so on.
The question is not if one can escape either paradox, it's rather how one should relate to predictions in order to benefit from them.
After all, most of our lives we are acting upon some sort of prediction or another, most of them being correct. We watch the weather report, we plan our trips based on scheduled flights, we write budgets for the next quarter. Most of these events are out of our control, yet we often make, or read an educated guess about them, which trains our decision.
But we sometimes fail. When economists like Nouriel Roubini predicted the economic crisis of 2008, or sociologists like Michael Moore predicted Donald Trump's victory of 2016, those who were directly interested that the said events do not happen, those for whom those events would have been an utter disaster -- have duly failed to listen. And in such an obstinate manner that makes those particular prophets true modern versions of Cassandra.
This paradox needs to be psychological, there is no other explanation. If we are thoroughly convinced that some future outcome is the only one possible, then there is no amount of reasoning that will sway our firm beliefs. It's probably this latency and inflexibility that has made our kind so oblivious to the major changes in history, even when predicted by any number of prophets, scholar or occult though they may have been.
The second kind of paradox is more interesting though. As initially mentioned, for any given future event, N outcomes will be prophesized, and T will be correct. Can T>1 be correct though? Indeed, not all predictions need be mutually exclusive. But even more, even though they are mutually exclusive, not all but one predictions need to be wrong. Let's suppose that 3 auditors are charged with independently assessing a company's forecast for the following year. Auditor 1 claims that the company will be severely fined by the fiscal authority for not declaring such and such assets. Auditor 2 claims that the company will go bankrupt, because its marketing strategy will fail to produce the increased sale forecasted. Auditor 3 claims that the company will be aggressively acquired by an competitor, because of its stock market vulnerabilities. Some of these claims are mutually exclusive, some are not. Since the reasoning is sound, the company CEO can well assume that all of them are correct however. It is therefore his or her duty to act upon all of them -- declare assets with the fiscal authority, change the marketing strategy and fix the stock market vulnerabilities. If the CEO is successful, all given predictions will be invalidated -- which is in fact a bad term, because the initial predictions were made on a certain course of events and most likely they were valid. Naturally, since the course of events has changed, nobody can be absolutely certain that the initial predictions were valid. But everybody can be certain that the outcome is based on those predictions.
With that in mind, we can basically divide our prediction-outcome spectrum intro 4 quadrants: prediction for outcome A, act upon it, outcome is not A (succesful use of a prediction); prediction for outcome A, no action upon it, outcome is not A (quack prediction); prediction for outcome A, act upon it, outcome is still A (final destination); prediction for outcome A, no action upon it, outcome is A (Cassandra).
Naturally, the ideal quadrant is the first one. It is also the most commonly encountered in our daily lives, as previously mentioned.
The problem is at the (sometimes) blurry line between the 4 quadrants. For example, we often dismiss predictions as quack before the outcome. Sometimes it's the boy who cried wolf too many times. But more often than not it's the steadfast conviction that we know better, which leads to Cassandra effects.
Another example is dismissing a prediction as quack after the outcome was different than predicted, even though we (more or less consciously) acted upon the prediction. A prediction which is altogether reasonable and correct may also fail without any action on behalf of the interested party, because it predicted only a higher probability outcome, not a certain outcome. Auditor 1 in the previous example may be proven wrong, with no action from the CEO, if the fiscal authority simply fails not notice the wrongdoing. Then the CEO may fire the Auditor for being wrong, or hire them again through the next year for being correct in their analysis.
The point is -- there's no real paradox with predictions. We use them all the time. They are valid as long their reasoning is valid and we are always right to pursue actions based on predictions. They are not validated or invalidated by the outcome, but by the success or failure of the correction course recommended.
To an extent, we can say that a good prophet is the one who provides "closed prophecies": the future outcome is A -- however if you do this and that to correct the future, than it can change to B. Most scholar prophets use this form of prediction (the global warming panels e.g.). While gray areas can still exist between A and B, there is less ambiguity and, if the spectrum is narrow enough, "between A and B" is still a much better prediction than "anywhere".