5/ History, the Fickle Teacher
Science requires the repeatability of outcomes should the starting conditions, the environment and the action/actor are repeated. In certain cases, this is not possible to bring the system into Stage 0 again, so the experiment must be run multiple times on multiple subjects simultaneously. But … the starting conditions and the environment are very rarely the same as before, so the learnings from the previous situations / simulations can’t be directly applied to the new situations.
[MK: in human speak, this means that our past experiences have more limited predictive powers than we think, despite our overconfidence.]
The hardest cause-effect assumptions are found in history when there’s no way on Earth to re-run the past events. (And as we know, simulations can’t factor all relevant variables.) The absence of “counterfactual” versions of history (i.e., we don’t know what might’ve happened) makes us perceive the current state of affairs to be have been inevitable. This is called creeping determinism (treating past events as having been more predictable than they actually were).
Sampling bias makes researchers try looking for common attributes among winners, while ignoring the fact that losers display the same attributes. Say, in air crash investigations a set of necessary conditions for the crash is also present in non-crash situations, but when a crash occurs, they are thought to explain the entire accident. Same works for all rare events (terrorist attacks, school shootings, stock market crashes).
Focusing on reducing/eliminating the occurrence of the necessary conditions for a crash is a good first step in avoiding catastrophes. [MK: but their presence doesn’t offer a lot of predictive power, either, as the thinking goes like “it was bound to happen given the circumstances”.]
Post-hoc fallacy (B followed A, so A caused B) can be more or less reliably caught and addressed when dealing with physical objects and the laws of nature, but things get complex in social settings when humans explain the cause and effect only based on their (fragmented) observations. [MK: and the fewer facts are available, the higher the perceived importance of each observed fact is.]
Common sense explanations pretend to tell us why something happened when in fact all they’re doing is describing what happened.
History doesn’t make much sense to those participating in it as no one sees the whole picture with enough depths to find the patterns and determine if they’re supporting or rejecting the narrative. What’s even more complex is that one can’t build the narrative (i.e., foresee what the current events will lead to) in real-time. The narrative is the very essence of historical explanations. And there’s inevitably a certain time period (rather long) that is needed to make sense of the current events – in the future.
Ahh, of course, the narrative will change based on the future date the facts are evaluated, as the outcomes unfold. So what may seem like a good idea in one year may look like a terrible idea in two years’ time and so on.
The very notion of the “outcome”, at which point we can evaluate, once and for all, the consequences of an action, is a convenient fiction. The events we think of as outcomes are never really endpoints – just milestones. Something always happens afterwards. Stock price performance is the best example of this rule.
Any time is no better than the other time to explain the milestones.
Storytelling (especially in history) is constrained by facts and observable evidence, not helped by them. Certain events are given more significance for the purpose of a drama. And the story itself unfolds in the usual pattern: a beginning, a middle and an end, when everything starts making sense.
This is the reason why simpler explanations are more effective (they’re just simpler!). Adding even irrelevant details to a simple explanation increases its believability. Or the mere fact that there’s some (even a badly wrong one) explanation of a statement makes it look more plausible than a statement with no explanation at all.
If the uninteresting or unfitting evidence is ignored and buried, there’s no way to even ask the “what if” question as the narrative becomes cemented.
Whenever we’re seeking to learn about the past, we’re invariably seeking to learn from it. It’s an implicit switch between storytelling and theory building, which we tend to ignore, despite the different objectives and standards of evidence. There’s only as much the history can help us with predicting the future.