All knowledge is the capacity to predict
February 12th, 2025
Post coming, eventually.
Nothing new here other than the phrase itself. Think Popper, Shannon, Bayes.
We can have beliefs that don’t confer predictive ability. They may even be true. But not knowledge.
Note that mathematical knowledge can be interpreted as a set of predictions about the results of counting or measuring.
And knowing Caesar crossed the Rubicon predicts consistent records and artifacts.
We can have knowledge about probabilities as well as binary (true/false) facts.
Update, 2025-10-30: I’ve seen a claim that this phrase is a F.A. Hayek quotation. The claim is re Alan Ebenstein’s Hayek (2001), or his Hayek’s Journey (2003), page 96; of which I don’t yet know.
I haven’t checked either yet, but I find this plausible – for years I thought it was a Hayek quote but couldn’t find any evidence of him ever using the phrase; eventually I decided I must have come up with it. (I said “ability” in my version; I’ve updated the post.) Still, this deserves a longer exposition…someday.
Update 2025-12-01: It’s indeed Hayek. From “Economics and Knowledge,” Economica, 1937; reprinted in Individualism and Economic Order, note 17.
With a little more context:
“All knowledge is the capacity to predict…The concept of equilibrium merely means that the foresight of the different members of society is correct. Correct foresight is the defining characteristic of a state of equilibrium.”
Per ChatGPT 5.1:
Hayek is not saying “all knowledge whatsoever, in every philosophical sense, just is prediction.” The full sentence is more careful:
He’s defining “knowledge, in the sense in which the term is here used” – namely, the kind of knowledge relevant for equilibrium theory and coordination in the economy.
In that sense, he says, knowledge is identical with foresight only in the sense that all knowledge (so understood) is capacity to predict. Online Library of Liberty+1
So as an aphorism it’s fine, but philosophically it’s rooted in a very specific context: the role of expectations and foresight in economic equilibrium and in the division of knowledge.
But I think the phrase is more literally and broadly correct than Hayek claimed. I think it’s a useful definition of “knowledge”. It also fits with Bayesian epistemology (all knowledge = posterior predictive distributions), Popper’s ideas about science (falsifiability = prediction test), Shannon information theory (information as reduction of uncertainty = narrowing the distribution of possible futures) , and modern machine learning.
Someday I should do a proper writeup. But these are the core ideas.
February 12th, 2025 at 10:07 pm
I don’t get the Caesar crossing the Rubicon thing. Doesn’t that assume consistent records and artifacts, rather than predict them?
Oh… CAPTCHA works. 😉
February 13th, 2025 at 1:05 am
If you know Caesar crossed the Rubicon, you can predict that as-yet unexamined records and artifacts will be consistent with that. The knowledge gives you the ability to make that prediction, and is therefore real knowledge.
February 14th, 2025 at 2:51 am
But the only way we know Caesar crossed the Rubicon is from records.
There are probably few, or no, as-yet unexamined records. But even if we found them, that would just mean that the existence of some records predicted that future records from that time will say similar things.
Imagine, for example, that a new gospel was discovered, and it said pretty much the same stuff that the Christian Bible already says. You wouldn’t say “we know Jesus walked on water, so that predicts as-yet undiscovered gospels will tell the story”. Instead, you’d say “we know there are writings that make claims about Jesus, so any as-yet undiscovered writings from that time will say the same”.
But … y’know … we probably wouldn’t even say that. It’s not a very interesting prediction.
February 14th, 2025 at 5:44 pm
There are plenty of records as yet unexamined by you or by me. “Knowing” Caesar crossed the Rubicon lets us make predictions about the content of those as-yet unexamined records, and of any so far undiscovered records.
Similarly with Jesus, “knowing” he walked on water (a surprising thing, likely to be commented upon) predicts that records from the time will be consistent with Jesus doing so.
Of course sometimes we think we have correct knowledge, and it turns out we were mistaken. That doesn’t change the definition of knowledge.