You may enjoy a story very similar in spirit, about magnetic monopoles, and also very similar to JerL’s story about FTL neutrinos. I think Hume would approve.
Magnetic monopoles don’t exist, but if they *did* electromagnetism would be a beautifully symmetric theory, and it would actually make certain derivations that posit fictitious monopoles and then set their charge to zero easier.
You can build a monopole detector pretty easily, and many people have! None have ever clicked… except for one time in 1982, on Valentine’s day, when they did.
FTL neutrinos is a great example to bring up when people make the counterargument that Hume's argument against miracles would also apply against discovering new laws of physics. The truth is that we actually do require extremely high standards of evidence for new laws of physics, and we regularly reject even the results of seemingly well-done scientific experiments performed by trusted scientific institutions if they violate well-established laws of physics. When FTL neutrinos were reportedly detected, virtually every physicist said, "No, this result has to be wrong. There must be some flaw in the experiment," and they ended up being right, as you will be 99% of the time when an experiment seemingly shows a violation of the established laws of physics.
A more recent example is the Muon g-2 result that seemed to violate the Standard Model, but it turned out we just incorrectly calculated what the Standard Model predicts. (Though in this case, it was less obvious that the result is wrong, since we know the Standard Model must break down eventually, whereas "neutrinos can't go faster than light" is something that would completely shatter our understanding of physics if we found out it was wrong).
I think you're right. But also I think you've posted the wrong link haha
I think what's missed about Hume is that he doesn't say "an unlikely event doesn't occur" or that "all observations must conform to our experience" but that "to the extent observations can invalidate our experiences, they better be pretty well attested!"
This is very well argued, and I think I’ll need to revisit it to digest it fully.
Hume’s “legalistic probabilities” are a much weaker structure than Bayesianism, at least at first glance. But I think a rigorous mathematical treatment ought to be possible, reframing it as a decision theory around probabilities.
Having not yet done so, I’ll say that I think the Bayesian framework is basically correct, but that BB et al. do it a disservice by a) not accounting for notions of convergence, the quality of evidence, and modeling assumptions, and b) just being mathematically sloppy.
However, once you account for those factors, you reach the limits of tractable calculations, and this Humean approach provides a useful set of heuristics for dealing with these limitations.
I’ll have to dig deeper on this topic, but if the measure of an epistemology is how easy it is to fool its practitioners, then I think I’d have a harder time tricking Hume than BB.
I think I agree with you here. The thing is, I'm not smart enough on probability theory to sort of parse out bayesianism vs Hume's method, but otherwise come to the same conclusion you do.
All I can say is that I think Bayesian probability outside of highly measured, demonstrative (in the Humean sense) contexts, it's very limited and perhaps unhelpful. Like it's super helpful for election forecasting, sports gambling, weather forecasting, etc, but it's a lot less helpful for the big questions of life. In fact, thinking like a Bayesian instead of a Humean IMO will have you at a higher chance of being scammed. (Because if you assign say a 20% value that someone is being sincere in giving you a crazy high ROI on an investment [but 80% chance of being a con!], it could feasibly be rationally justified in Bayesian but obviously a scam in Humean means)
If you're curious about how done Hume scholars reconcile him with Bayes, there's a few good pieces by Millican on Hume dot org.
Yes, what's great about Millican is that he (unlike many academics) basically posts everything he's written on that website. I would highly recommend his writings on John Earman on Hume and subsequently similar concepts about how Hume is slightly wrong in Bayesian terms (he needs to reframe the probability balance to be about false positives).
But anyway, thanks for reading and thanks for your positive feedback!
I don't have anything meaningful to add, but I wanted to thank you for putting together such an ambitious piece. This is outstanding work, and digestible for laypeople like me!
The article is really good, but the thumbnail is really bad. What in the unholy AI body horror caused Hume's fingers to be fused together? I made you a new thumbnail (pure photoshop, no AI) if you want it: https://drive.google.com/file/d/17gnRBYIQJIz7iacmZMbqB7R_3Nb7pG03/view?usp=sharing
holy shit this is so much better lmao you’re the real MVP
thank you i will use it! I literally just used AI because I had spent like a week writing it and had no idea what to do for the thumbnail
You may enjoy a story very similar in spirit, about magnetic monopoles, and also very similar to JerL’s story about FTL neutrinos. I think Hume would approve.
Magnetic monopoles don’t exist, but if they *did* electromagnetism would be a beautifully symmetric theory, and it would actually make certain derivations that posit fictitious monopoles and then set their charge to zero easier.
You can build a monopole detector pretty easily, and many people have! None have ever clicked… except for one time in 1982, on Valentine’s day, when they did.
https://www.forbes.com/sites/exl/2025/09/23/how-healthcare-leaders-are-moving-ai-from-pilots-to-enterprise-impact/
FTL neutrinos is a great example to bring up when people make the counterargument that Hume's argument against miracles would also apply against discovering new laws of physics. The truth is that we actually do require extremely high standards of evidence for new laws of physics, and we regularly reject even the results of seemingly well-done scientific experiments performed by trusted scientific institutions if they violate well-established laws of physics. When FTL neutrinos were reportedly detected, virtually every physicist said, "No, this result has to be wrong. There must be some flaw in the experiment," and they ended up being right, as you will be 99% of the time when an experiment seemingly shows a violation of the established laws of physics.
A more recent example is the Muon g-2 result that seemed to violate the Standard Model, but it turned out we just incorrectly calculated what the Standard Model predicts. (Though in this case, it was less obvious that the result is wrong, since we know the Standard Model must break down eventually, whereas "neutrinos can't go faster than light" is something that would completely shatter our understanding of physics if we found out it was wrong).
I think you're right. But also I think you've posted the wrong link haha
I think what's missed about Hume is that he doesn't say "an unlikely event doesn't occur" or that "all observations must conform to our experience" but that "to the extent observations can invalidate our experiences, they better be pretty well attested!"
Whoops! Wrong forbes article hahaha
https://www.forbes.com/sites/startswithabang/2019/02/07/the-enduring-mystery-of-detecting-the-universes-only-magnetic-monopole/
Great article!! Enjoyed it.. thanks for doing it. I also really found the Fogelin book helpful! Well done!
This is very well argued, and I think I’ll need to revisit it to digest it fully.
Hume’s “legalistic probabilities” are a much weaker structure than Bayesianism, at least at first glance. But I think a rigorous mathematical treatment ought to be possible, reframing it as a decision theory around probabilities.
Having not yet done so, I’ll say that I think the Bayesian framework is basically correct, but that BB et al. do it a disservice by a) not accounting for notions of convergence, the quality of evidence, and modeling assumptions, and b) just being mathematically sloppy.
However, once you account for those factors, you reach the limits of tractable calculations, and this Humean approach provides a useful set of heuristics for dealing with these limitations.
I’ll have to dig deeper on this topic, but if the measure of an epistemology is how easy it is to fool its practitioners, then I think I’d have a harder time tricking Hume than BB.
Thanks for a great read.
I think I agree with you here. The thing is, I'm not smart enough on probability theory to sort of parse out bayesianism vs Hume's method, but otherwise come to the same conclusion you do.
All I can say is that I think Bayesian probability outside of highly measured, demonstrative (in the Humean sense) contexts, it's very limited and perhaps unhelpful. Like it's super helpful for election forecasting, sports gambling, weather forecasting, etc, but it's a lot less helpful for the big questions of life. In fact, thinking like a Bayesian instead of a Humean IMO will have you at a higher chance of being scammed. (Because if you assign say a 20% value that someone is being sincere in giving you a crazy high ROI on an investment [but 80% chance of being a con!], it could feasibly be rationally justified in Bayesian but obviously a scam in Humean means)
If you're curious about how done Hume scholars reconcile him with Bayes, there's a few good pieces by Millican on Hume dot org.
I’ll have to check that out, thanks!
Yes, what's great about Millican is that he (unlike many academics) basically posts everything he's written on that website. I would highly recommend his writings on John Earman on Hume and subsequently similar concepts about how Hume is slightly wrong in Bayesian terms (he needs to reframe the probability balance to be about false positives).
But anyway, thanks for reading and thanks for your positive feedback!
I don't have anything meaningful to add, but I wanted to thank you for putting together such an ambitious piece. This is outstanding work, and digestible for laypeople like me!
So once again, thank you!
The secret is that I am also a layperson lol
Well, let's say there are levels to being a layperson haha