The central problem is that the practical value of “real facts” is often small, compared to the noise.
The central problem is that the practical value of “real facts” is often small, compared to the…
Henry Kim

I don’t necessarily agree with the Wired article, but even if the scientific method is obsolete, the human brain still relies on mental models for decision-making. Unless we’re prepared to allow computer algorithms/block boxes to make all of our decisions for us (decisions we couldn’t analyze, verify, or make sense of ourselves), we’d still need some way to convert machine learning into human-accessible models/theories.

But to your larger point: I agree that, under present conditions, the practical value of “real facts” or “good models/theories” is often small. Under present conditions, the benefit of switching from “coalitional facts” to “real facts” is so small, it’s typically more advantageous to stick with “coalitional facts” and get the security of being in a coalition.

But that’s only true under present conditions, where most of our models are isolated and relatively naive. Once we start integrating those models (drilling down and building up), and reconstructing those models so they can become robust and sophisticated, the equation changes. This is what occurred in science. At some point, as scientists began integrating scientific theories into more robust frameworks, the power and utility of science became obvious to everyone. I believe the same thing happens in vertical learning. Once we start to drill down and build up, the practical value of “real facts” and “good models/theories” is suddenly massive—and we see that engaging in scientific inquiry is more useful and advantageous than engaging in coalitional politics.

If I’m a scientifically-literate creationist, I can constantly tweak my version of creationism to include any scientific theories that are useful to me. In some ways, I get the best of both worlds: (1) I get to use the science I need, and (2) I get the security of my coalition. But that only works as long as I’m adding each scientific theory separately. If I try to integrate those scientific theories, and those theories are more powerful and useful when integrated, I quickly reach a point where the creationist stuff is largely incompatible. In the end, I’ll have science, and I’ll leave open the possibility that some creator just set everything in motion.

The way we make “real facts” and “good models/theories” significantly more useful is by learning vertically.