The American Bill Bartley took up Popper's ideas on rationality with both hands and made it into a major project. He fell out with Popper for 12 years until they reconciled, and Bartley edited the 3 volumes of the Postscript to the Logic of Scientific Discovery. Tragically, he died in his prime, aged 55, in 1990. He was working on Popper's biography and also a biography of Hayek.
I have written a great deal about Popper and related matters over the years.
Thanks for the information and the kind words, Rafe. I’ve been a long-time fan of your website ever since I first came across it in 2017, and I still find it refreshing to visit.
I hadn’t heard of Bill Bartley before, but I’ll look him up. It doesn’t sound as though he and I would agree on this, though.
Interesting thoughts. However, strictly speaking markets do not exemplify rationality in the sense you are talking about. Rationality is the use of your mind as a means to achieve knowledge. And it is the attribute of the individual, while a "market" is the collection of trading individuals (some can choose to be irrational).
And as you said "Rationality is not a matter of possessing truth" it is the tool which allows you to seek and find the truth. High IQ doesn't mean rationality. Many good scientists, for example, are very religious. And belief is not knowledge (belief is based on mysticism, knowledge on reality) but they compartimentalise the two things allowing themselves to hold the contradiction in their mind.
I used the term ‘market’ a bit loosely to mean something like ‘business’ or ‘the private sector’. The main point is that outside of science, people engage in the problem-solving cycle too.
Yes, I know what you mean, and that’s certainly true. My point was that only a human can be rational. Institutions, groups, collectives, committees, sectors do not possess reason but can only display the outcome of the rationality of the individuals involved.
Hey Sam, we met at RatFest 2025. Enjoyed the article.
One question. The Bayesians you’re arguing against could say updating just is error-correction, so I’m curious how you’d separate the two. My instinct: updating can only re-weight explanations you already have, it can never invent a new one. Kepler could describe how planets move; only Newton’s creative leap to gravity explained why. That leap is the thing updating structurally can’t produce. Would you locate the real dividing line there, in the power to generate new good explanations rather than just re-weight old ones?
Bayesian updating looks a bit like error-correction, but it is not how people think. For an LLM, updating might genuinely be the error-correcting part of their programme. But, as you note, people can go beyond that and produce new ideas, which updating doesn’t allow. Also, people don’t think like an LLM in another way: namely, they assess the explanatory power of a theory, which is mult-dimensional and hard to capture with a single number.
The American Bill Bartley took up Popper's ideas on rationality with both hands and made it into a major project. He fell out with Popper for 12 years until they reconciled, and Bartley edited the 3 volumes of the Postscript to the Logic of Scientific Discovery. Tragically, he died in his prime, aged 55, in 1990. He was working on Popper's biography and also a biography of Hayek.
I have written a great deal about Popper and related matters over the years.
http://www.the-rathouse.com/writingsonpopper.html
http://www.the-rathouse.com/writingsonbartley.html
https://www.amazon.com.au/s?k=Rafe+Champion&i=stripbooks&crid=GZ66NWUYZ193&sprefix=rafe+champion%2Cstripbooks%2C262&ref=nb_sb_noss
It is great to find your site!
Thanks for the information and the kind words, Rafe. I’ve been a long-time fan of your website ever since I first came across it in 2017, and I still find it refreshing to visit.
I hadn’t heard of Bill Bartley before, but I’ll look him up. It doesn’t sound as though he and I would agree on this, though.
Interesting thoughts. However, strictly speaking markets do not exemplify rationality in the sense you are talking about. Rationality is the use of your mind as a means to achieve knowledge. And it is the attribute of the individual, while a "market" is the collection of trading individuals (some can choose to be irrational).
And as you said "Rationality is not a matter of possessing truth" it is the tool which allows you to seek and find the truth. High IQ doesn't mean rationality. Many good scientists, for example, are very religious. And belief is not knowledge (belief is based on mysticism, knowledge on reality) but they compartimentalise the two things allowing themselves to hold the contradiction in their mind.
I used the term ‘market’ a bit loosely to mean something like ‘business’ or ‘the private sector’. The main point is that outside of science, people engage in the problem-solving cycle too.
Yes, I know what you mean, and that’s certainly true. My point was that only a human can be rational. Institutions, groups, collectives, committees, sectors do not possess reason but can only display the outcome of the rationality of the individuals involved.
Hey Sam, we met at RatFest 2025. Enjoyed the article.
One question. The Bayesians you’re arguing against could say updating just is error-correction, so I’m curious how you’d separate the two. My instinct: updating can only re-weight explanations you already have, it can never invent a new one. Kepler could describe how planets move; only Newton’s creative leap to gravity explained why. That leap is the thing updating structurally can’t produce. Would you locate the real dividing line there, in the power to generate new good explanations rather than just re-weight old ones?
Bayesian updating looks a bit like error-correction, but it is not how people think. For an LLM, updating might genuinely be the error-correcting part of their programme. But, as you note, people can go beyond that and produce new ideas, which updating doesn’t allow. Also, people don’t think like an LLM in another way: namely, they assess the explanatory power of a theory, which is mult-dimensional and hard to capture with a single number.