winterkoninkje: shadowcrane (clean) (Default)
Only recently, thanks to the computer, has it become feasible to solve real, nontrivial problems of reasoning from incomplete information, in which we use probability theory as a form of logic in situations where both intuition and "random variable" probability theory would be helpless. This has brought out the facts in a way that can no longer be obscured by arguments over philosophy. One can always argue with a philosophy; it is not so easy to argue with a computer printout, which says to us: "Independently of all your philosophy, here are the facts about what this method actually gives when applied."

Daaamn. That's some gettin' told right there.

The above quote (emphasis added) is from the eminently readable Probability in Quantum Theory by E.T. Jaynes, which presents a critique and alternative perspective on the role of probability within quantum mechanics. If you've any interest in philosophy of science or the philosophical disputes between frequentism and Bayesianism, even if you've no real knowledge of physics, then I highly recommend reading it. While the frequentist vs Bayesianist argument is well-known of, the details of what is actually at stake are less well-known and often quite subtle. I think the author does a good job of bringing out and highlighting what the argument is about, and why it is relevant to the future of science (especially physics).

For my part, I've been well indoctrinated into the Bayesian philosophy. This semester I'm taking a course on frequentism, or rather on "experimental methods" as they call it. A professor here has been pushing hard for Bayesian methods in behavioral sciences, and the professor of my class delights in teasing him about it (though he admits to no investment in the philosophical debate). It's been a very long time since I've seen the frequentist perspective, and I'm always of the opinion that it's good to keep an eye on one's philosophical enemies. I've known that frequentism has long dominated the behavioral sciences, but I must shamefully admit that I've attributed this to them being "soft" (even for as much as I identify with my undergrad training in anthropology and humanities). However, coming from machine learning where Bayesianism is de rigueur, one thing I found startling about the article is that, apparently, in physics too it's the frequentists who've dominated the conversation for decades. Indeed, as Jaynes portrays it, it's the frequentists who ousted Laplace, rather than the other way around as is portrayed in AI/ML circles.

Date: 2012-01-27 04:18 pm (UTC)From: [personal profile] silmaril
silmaril: (Default)
Saved into my ebook directory; hopefully my reader can show the equation (it does read pdf). This sounds very interesting. Thank you.

June 2017

18192021 222324


Page generated 22 Oct 2017 01:03 am
Powered by Dreamwidth Studios