Perspective on knowledge: The future of predictability
The ancient Egyptians knew that when Sirius disappeared for a few weeks and then became visible before dawn, a few weeks later the banks of the Nile would overflow. For three thousand years, this prediction held. Except it wasn’t really a prediction. The Egyptians didn’t have an idea of the future that enabled them to make predictions in the modern sense.
For example, when we talk about banks and note that they’re closed on Sundays, we’re not really making a prediction. We’re stating a fact of modern life in the United States. If we put on Karnack’s turban—mimicking the old Johnny Carson fortune-teller character—and were to say in a grave and mysterious voice, “I predict the banks will be closed on Sunday,” it would be either a joke or a misunderstanding of what it means to make a prediction. The same would be true if we were to “predict” that an apple will fall when dropped, that day will be followed by night, or that two plus two will turn out to equal four.
To make a prediction, the future has to be unpredictable. The ancient Egyptian future was cyclical, which is the opposite of unpredictable.
But having an unpredictable future is not enough to enable the concept and practice of prediction to arise. For the ancient Egyptians, that which was not cyclical, such as the day the pharaoh will die, was too unknowable to be predictable. The ancient Hebrews, on the other hand, had a non-cyclical future. They could rely on it because it came straight from the mouth of God, but it came as a promise that someday they would return to their land and the world would be redeemed. How, and even if, they got there was up to them. That’s why the words of their prophets are generally too conditional to sound like predictions: If our people continue in these wrong-headed ways, then we will face deprivation and punishment, but if we follow the word of God, then we will be blessed.
The ancient Greeks, on the other hand, believed in a future that enabled predictions of a certain sort. When they looked up, they saw the same wheeling stars that the Egyptians did and believed in their regularity just as firmly. But at eye level, below the heavens, there was no telling what would happen. The Greek framework for making sense of this didn’t entirely cohere. The Fates determined your lifespan, as well as some of the broad-brush themes, such as whether your marriage was going to be happy. The gods could not undo the Fates’ decrees, but they could intercede in other ways relevant to your life. Then there were the daimons who intervened in individual lives in unpredictable ways. So, quite a mix of super-human forces determined the turning points in your life, including just plain bad luck that might, for example, have you captured by an enemy army and turned into a slave.
The preordination of events and the Greeks’ awareness that they were not fully in control of their future created a space for predictions … predictions famously delivered by the Oracle in pronouncements that typically could not be understood until the events they predicted had come to pass. Just ask King Oedipus who knew the Oracle had said he would kill his father and marry his mother, and yet could not escape that fate.
Even so, we probably wouldn’t say that the Oracle was making predictions so much as pronouncements, for there is no way for the Oracle’s statements not to become true. For predictions in the modern sense we need a future that is determinate yet not fully knowable. We need a future that we can be right about, but that we can also be wrong about.
Knowable but not too knowable
That’s the idea of the future that we have grown up with. Predictions in our culture are always probabilistic, even if we don’t explicitly state the probability. “The Democrats aren’t going to win any new Congressional seats in 2018” is a prediction because it’s understood that no matter how confidently I pronounce it, I recognize that the future is uncertain. For us to have the form of speech we call “predictions” we need a future that can be known only probabilistically.
But that’s not enough for predictions. We also have to be able to say why we think the future will turn out this way instead of that. If you ask me why I’m so pessimistic about the Democrat’s chances, if I say, “I dunno. I just think that,” then it’s such a weak form of prediction that it’s really just a guess, like my guessing that the next throw of the dice will be a four.
So, for predictions to be a form of thought, we need a future that is knowable but not too knowable. And that’s what we have now. We believe the future is determined by a set of scientific rules—what we used to call Laws of Nature—operating on a set of data too vast to be perfectly comprehended.
We are now, however, on the cusp of changes in both those beliefs. First, we are well along in accepting that simple rules can quickly yield highly complex results. Second, our machines are now far better able to manage vast quantities of data, and doing so without much reducing the complexity of their interrelationships. Some of those machines’ predictions are already being made without the possibility of human brains understanding how the machines came up with them: too many variables, too many contingent relationships. Yet those predictions are showing themselves to be highly accurate.
We thus have a type of future within which predictions make sense. It’s just a future far more complex and interdependent than the Egyptians, Hebrews, Greeks, or we about 10 years ago could ever have predicted.