On the inherant unreality of probability, and also everything else

A few months ago, before the American science universe started crumbling around our ears^[1] , this article was posted to Nature News.

And yet, any numerical probability, I will argue... is not an objective property of the world, but a construction based on personal or collective judgements and (often doubtful) assumptions.

With a minor modification, this unintentionally articulates a central tenant of Buddhism - one which also has important implications for science and beyond. In fact, any concept at all, not just numerical probabilities, are not objective properties of the world.

Folks that know me casually are often surprised to learn that I'm a Zen Buddhist. After all, I'm a scientist, skeptic, biologist, and atheist. Add this to my personality (extroverted, opinionated, loud), and it seems odds with what most people associate with Zen - quite, calm, woo-woo spirituality. But the truth is that my embrace of Buddhism is deeply informed by my science.

The central argument in this Nature News article, is that probability is in some sense "made up". A useful fiction to be sure, but it doesn't represent "an objective property of the world". This is essentially a narrow instance of a core Buddhist concept - śūnyatā (the first 's' is pronounced with a 'sh', and the emphasis is on the last syllable - [shoon-yah-TAH]).

This word is often translated as "emptiness", and the basic teaching, at least in the school of Buddhism that Zen is part of (and at least as far as I understand it^[2] ), is that all concepts are empty (śūnya). That is, nothing has intrinsic substance or reality.

From the inside, it is tempting for many practitioners to go off the nihilist cliff that nothing matters - that the world is empty of existential meaning. From the outside, it can be tempting to reach for the closest coffee mug or pencil, hold it up, and say, "What, you're saying this doesn't exist? That's absurd!" But both of these reactions, it seems to me, miss the point entirely. For those who, like me, are of a more materialist than spiritualist bent, there's a related idea that comes up frequently in science which to me points the way - it's summed up by the phrase:

All models are wrong, but some are useful.

According to wikipedia, the specific phrase can be attributed to the statistician George Box, but the underlying meaning is much older, and it is clearly not restricted to the narrow case of statistical models. Models are everywhere in science; even the most foundational theories with grand names like "The Central Dogma" of molecular biology^[3] and "The Modern Synthesis"^[4] are really just models.

Every individual experiment also relies on models - ideas about how some biological process works. And every experiment is itself a model - a process-in-miniature that we hope will represent a larger whole. We use that model to interpret what an experiment means, and try to fit the results into our larger models of pathways, cells, organisms, and behaviors.

All models are wrong...

Where we get into trouble is when we grasp our scientific models too tightly - when we forget that the image of reality that we've constructed is not reality itself. By necessity, a model is a simplification, a blurring of detail. Another familiar phrase that captures a similar idea is "the map is not the territory." A map is a model of the territory, a shrinking and simplification of the relationship between spaces.

A model or a map is not "wrong" in the sense of being false, it's wrong only inasmuch as we forget its fundamental nature as a simplification. If we're on a trail in the woods staring only at a map, we miss the trees and the birds, we may stumble over a boulder or fallen tree, we miss the gathering clouds telling us that a storm is coming. In science, getting attached to models also leads to stumbles. If we believe our models too strongly, we may continue down rabbit holes, or miss key insights. Even the Central Dogma was wrong, in the sense that it didn't account for the reverse transcriptase of retroviruses, which synthesize DNA from an RNA template instead of the other way around. I for one am real glad David Baltimore wasn't so enraptured by the existing model that he missed that crucial discovery.

The insight of śūnyatā is that it's not just maps or statistical and scientific models that are wrong (again, in the sense of being abstractions or simplifications), it's that all of our concepts are wrong. The coffee mug you reached for to prove that things have substance certainly has a hard surface and perhaps a warmth that you can feel with your hands, or an aroma that you smell when you bring it towards you. But the concept "coffee mug" is a simplification and abstraction that cannot capture or convey the actual experience of the thing itself.

...but some are useful

And yet, we do need models, in science and in life. Hiking a new trail without a map can be dangerous, and experiments that aren't guided by models cannot mean anything.

In a sense, every word of spoken language is a model, and our ability to communicate at all relies on us having similar models represented by the same words. If you ask me for a cup of coffee, I may brew it slightly differently than you would, and I may bring it in a mug that's different than the one you pictured, but I'm unlikely to bring you tea.

Recognizing śūnyatā, that all models are wrong, is not reason to despair or to embrace nihilism. It's not even support for relativism; after all, the flip side of the clause "... some are useful" is that many models of the world are not useful. Instead, it's an invitation to humility, to grasp your ideas a little less tightly. Being more consciously aware of our models helps us to recognize when they are useful, and when we should instead look up and see the sky.

h/t Bob Hayes for pointing me to the Nature News article.