When I used to pick NFL games against the spread for RotoWire, I would joke after a bad week that my picks were perfect, but it was the games that got themselves wrong.
I’ve touched on this concept quite a bit here too — But My Process Was Good and Why Blake Snell Should Win The NL Cy Young Award This Year both highlight the category error of speciously applying tools of prediction toward the past. Tldr: the past is in no need of predicting because it already happened.
The error stems in part from mistaking the map for the territory — once the forecasts are substituted for reality itself, the future has already happened, so to speak — it’s right there on the spread sheet! — so it is no different than the past which has also already happened. So any tool that can be applied to the future can equally be applied to the past.
This is how we get Sam Harris arguing that mRNA shot mandates were correct because covid *could have been* more deadly (and implicitly that the shots could have been safe and effective.) It’s one thing to model a possible future (with the requisite humility your forecasts might be off base), and quite another to force your failed model onto past events as though they were still in any doubt.
If this is confusing, that’s because a lot of us have been so trained to think this way we don’t even notice we’re doing it. Examples are legion where we identify what we deem an optimal process to achieve a desired result, then substitute that process for that result.
If the local police department believes it needs to arrest 50 people per month to reduce violent crime and incentivizes its officers to make those arrests, pretty soon arrests will be the goal, and people will get arrested for more and more non-violent offenses to make the quota.
If mass-vaccination is believed to be the key to good public health policy, then authorities will incentivize uptake, disincentivize bodily autonomy and not care one bit whether a person had prior immunity from the virus already, if the injection actually stops the spread or has myriad serious side effects. If the result (public health) were still the goal, the idea of forcing someone with natural immunity to inject himself with an experimental technology would be ludicrous. But once the goal shifted to the process (mass vaccination), it made perfect sense.
Once we substitute the process by which we achieve something for the achievement itself (which might temporarily be sound, if you’re, say, training for a marathon), we create distortions in our thought processes and policies.
But I want to go back to where the bait and switch of process and result gets most absurd, and that’s when it comes to time. In this case, you’re not arresting jaywalkers to make your quota, you’re lamenting that had jaywalking correlated better with domestic violence, your policy would have been the correct one.
In other words, you’re not only substituting the process for the result in a forward-looking way, you are doing so even though you now have complete access to the result because it has already happened. And yet you are still grasping via your spread sheet for some other reality that might have been!
You do not need to forecast events that have already happened. The probability of the Chiefs winning the Super Bowl last year is 100 percent. We no longer need your tools to tell us how much to bet on them. The game is over, and whatever we believed the probability to have been before the fact is irrelevant.
That’s why I wrote about Blake Snell’s Cy Young Award case — despite Snell allowing 42 percent fewer runs on his watch than Spencer Strider — people are still arguing for Strider because the way Strider pitched *portended* fewer runs allowed than Snell. It is not enough for them that the results (runs allowed) are already in. They argue that because Strider’s profile forecasts a better result than Snell’s, Strider should win. Or, put differently, had we simulated 1000 seasons for each, Strider’s would on average be better. They are upset Snell got “lucky” as though reality must be stripped of all the effects for which we can’t attribute to particular causes and normalized in order to fit our model.
. . .
Mistaking the past for the future is one form of this category error, but it can also happen in reverse wherein one mistakes the future for the past. You see it when people treat future events at fait accompli and proceed as such. The notion of talking past the sale, e.g., if the car salesman says: “You’ll love this car and be thanking yourself for saving so much money on gas!” he’s treating your possible future agreement on a price as something that’s already happened.
People also do this to avoid responsibility, pretending as though there’s nothing one can do about some injustice or that it’s too late to make amends for wrongdoing. They think and talk about the future as though it’s the past.
When our dog Oscar got poisoned from a particular kind of caterpillar (Largata do Pinheiro) last winter, we brought him to the hospital with his tongue swelling so much it was stuck outside his mouth. The hospital treated him for a day, saying they could only wait and see, and he might lose his tongue (to necrosis) and have to get put down.
We could have taken their approach of it not being in our hands, but instead we convinced them to let us break him out and took him to our holistic vet who sedated him, drained his salivary gland, massaged the tongue extensively with ozone and anti-bacterial herbs and sent him home with us that evening.
Oscar, who is passed out on the sofa next to me as I type this, only lost about 10 percent of the tongue, but otherwise made a full recovery.
. . .
I’m not sure why people so easily mistake cause for effect, process for results, future for past and vice-versa. If I had to speculate I’d say when our emotions get involved, we go from logical thinking to association, and associations work in both directions — effects and causes are associated without regard for which is which.
The modus operandi of most advertising (and propaganda) is to seed associations in your mind between things it wants you to associate — it’s certainly not to convince you rationally which is a much more difficult task. Maybe the influencers, marketers and propagandists have simply found an angle they can exploit. Once those associations take hold, they can be powerful, and often no amount of rational argument will un-cement them.
But I think that’s only part of the story. The other part is how most of us have been educated, and that's to focus on process not results. To be “process-oriented” not stuck in “results-based thinking.” This is a mark of erudition, of belonging to a certain class of people who have mastered this lesson. The class who trusts in modelers, forecasters and data-scientists waiting at the ready to be of service to The Science, even if it means treating the future as though it’s already happened and the past as though it can be re-forecast according to their models, conjugating reality, as it were, in the wrong tense.