> The important thing is that we both agree that finding facts that your model can't account for means you have to replace/augment your model,nk matter how useful it's conclusions had seemed so far
Ya I think we agree on this. Maybe I go one step further though, and I'm proposing that even if your model accounts for everything, it is still not the truth, just a good model, and that it works perfectly today does not guarantee it will work perfectly forever or that reality is exactly as your model explains it to be.
I can imagine both a reality which changes in time and space, thus a model could be perfect given some specific time and space, but no longer apply in another. And I can also imagine equal models that are both just as accurate, yet they differ in their explanation. So a model which would have perfect prediction still might not be the true reality, it could be an equal model with similar behavior, but how you've modeled things could still be very different from how reality works.
Which is where I'm kind of suggesting that the search for truth is irrelevant, and trying to argue for or against truth is futile. It be better to search for models with practical benefits instead, that the model is true to reality or not isn't something we should fight over, because it doesn't matter, and it's not objectively provable.
So when I say "search for truth", I don't mean the search for an understanding that enables better prediction or manipulation of outcomes. I take it to mean the search for what reality actually truly is like. And I'm saying the search for that, while it sounds noble, it really just ends up being dogma, and each of us can believe whatever we want. But the search for the former, the search for models of reality that have practical benefits is something we can all discuss and benefit from. And so I think our focus should move away from this "truth seeking" and towards this "practical model seeking".
Hope that made sense too. I guess in a way I'm saying that for example, if someone says that "this is what really happened in the past", and then they show you evidence as to why, there's this parchment, this old painting, this focile, well that's still just someone's belief of what happened, even if it constructs an explanation that fits the data. It might not be what actually happened at all, it's just a good model that fits the current data. So practically speaking, all you can do is nod and smile, or ask: ok what can I do with this? Can I use it to accurately predict what will happen in the future? Can I use it to make better decisions today so that the future ends up in that way instead of some other way? If no, well your explanation is crap, and all we can do with this explanation if you insist it is true is get mad at each other.
Oh, and the last thing is, with this frame of mind, it is perfectly acceptable for contradicting models to coexist. A model doesn't have to be "absolutely true to reality" to be practical. In fact, a not true to reality model could be more useful sometimes, an approximate model for example can sometimes be more easily computed, or put to use, it can make it that predicting things (while possibly less accurate), can be done more quickly or at higher scale, or with less data, etc. So again, I'm trying to show that if you discuss practical models, you kind of remove all sense of nobility and righteousness and pride and whatever that a belief in "truth" brings.
Edit: Oh and if anyone has read this far, to bring it back to Malcom Gladwell. What this means is that if you don't look at his revisionist history as a claim to the truth, but instead consider it a model of the past. The question isn't if he is right about the truth or not, but how practical is his framing of history for the reader. Can I leverage it to make better decisions today about political policies? About work policies? In my day to day life? Can I apply it to predict social trends? This is now what becomes important, how good a model is it, not how truthful is it.