mcgann_letterhead

#60 On scientific confidence

Dear Jamie

However many times I’ve been truly guilty of dastardliness in past letters, my last one was a genuine effort to understand. My brain was having serious trouble remapping this noun “knowledge” to a verb. I was having trouble quite generally, and so this is an unacceptably late letter.

Since our call the other day I think I know what you mean by knowledge is “what happens”. You’re saying knowledge is nothing magic, it’s the algorithmic unfolding of the physical world. Well, sure, maybe, I mean it’s hard to disagree that the unfolding of the physical world is all there is. It’s clear that in some special cases things unfold algorithmically, i.e. in useful, entropy-reducing or otherwise productive ways. It seems valid for me for you to define it like this, but it doesn’t seem very useful.

However we define knowledge, I’m at least happy to join you in casting off the word truth as an obtainable or even approachable thing. But I’m not willing to say that as we make objective scientific progress we aren’t touching some objective reality. There really does seem to be a convergence and compatibility of our best theories and this suggests a real regulatory force at work on our theories. But I’m not precious about truth. Perhaps the best way to put it is as a conjecture, “truth” itself is a concept invented as a place holder to explain the convergence of theories; analogous to inventing dark matter to explain the distant convergences of matter. But you’re right bad conjectures can lead people down the wrong paths and hold back progress. I’m happy to move on without it in case truth is a bad one.

So perhaps we have different notions of what knowledge is at a very fundamental, physical level, but we think it is doing the same job in practice? That’s philosophy I guess.

We do have different notions but it’s because I don’t pretend to have any theory of knowledge at a fundamental physical level. We don’t need to rectify the independent existence of atoms and knowledge in the same sense that we don’t need to rectify germ theory and atomic theory, or economics and atomic theory. You might argue that even though we don’t need to explain it, it’s a valid intellectual exercise to explain it. It might be a bit of fun but theories are only as valuable as their problem situation is well defined.

In my opinion we don’t know enough about physics and knowledge in the domain they might interact to define a problem. So I wonder invent the theory? My accusation is that you’re proposing one because you’re trying to build a Dennettian crane. And you’re building one because of human, political concerns.

The idea of naturalistic reductionism is a strong claim, but I believe it actually comes from a place of weakness in the scientific community. If one sees the story of science as a plight where science is under threat from superstition and bias, then it makes sense to invent Dennettian cranes to mop up any mysteries that might let people’s imaginations run wild. But this war-like frame of mind forms an underlying belief structure that cuts off creativity and slows the progress of knowledge.*

Whatever threats might generally exist we must indeed defend ourselves practically, but insisting on hole-plugging theories purely because holes make us look bad only spoils the hard-won web of knowledge. We don’t need confidence in our theories, we need confidence in our methods and traditions of science.

Cheers,
Mat

 

* It may well be true that science is continually under siege, but that does not mean theories proposed in its defence have any reason to be true. In fact they’re not solving real, physical problem situation so there’s almost no reason to believe them.


Also published on Medium.