8 ms·

Given that inductive reasoning is possible and that that it is (in the case of Solomonoff induction) on very solid philosophical and mathematical footing, I'm s

Given that inductive reasoning is possible and that that it is (in the case of Solomonoff induction) on very solid philosophical and mathematical footing, I'm skeptical of any book which claims it's impossible.

pron 11y ago
> Given that inductive reasoning is possible The question is not whether or not it is possible (in the sense that it is useful) but whether it leads to knowledge. This, of course, depends on your definition of knowledge, which is the question of epistemology. In some circles these days there is a move back to a utilitarian definition of knowledge, namely, if you can interact with a system and predict its behavior then you can say you know how it works, because your model is useful. This is not, however, the commonly accepted view of what constitutes knowledge. Funny thing: the consequence of the Gettier problem (sort of like the "incompleteness conjecture" of epistemology) is that either we must accept a very unsatisfying definition to what constitutes knowledge, or use a good, rigorous and satisfying definition and then find out that nothing we think we know actually qualifies.
marklgr 11y ago
Even if it's true, Hume was a great thinker and keeps historical value. Besides, I'm sure he would agree with your being skeptical of any book, whatever its theses :)
graeme 11y ago
Hume meant inductive proof. I checked Solomonoff induction, and it seems to be talking about probabilistic inductive reasoning. It isn't logically possible to prove anything affirmatively. That's what Hume was referring to. https://en.wikipedia.org/wiki/Problem_of_induction https://en.wikipedia.org/wiki/Problem_of_induction
duaneb 11y ago
You can reason effectively, but proving is another thing entirely. You can never prove beyond the shadow of a doubt we are not about to be razed by gamma rays--there could be some arbitrary rules to the universe that only kick in right now.
- chroma 11y ago
  Sorry, but it's stuff like this that causes me to avoid most philosophy. Yes, there are extremely small likelihoods for all kinds of extremely unlikely things. That doesn't make tools like induction any less useful. Sure you can't prove X, but you can show that the likelihood of ~X is far, far lower than the likelihood that you will hallucinate ~X.
  - graeme 11y ago
    This one's at the foundation of our science though. Karl Popper's ideas on falsification have been very influential in how scientists frame their work. And errors of induction have caused some notable financial blowups because models missed tail risks. The example you replied to was farfetched, but not an abstract ivory tower notion – it's something deeply built into how we view the world. This doesn't mean there's no use of inductive reasoning, or that we shouldn't suspect that some things are far more likely true than others. But remembering the ultimate impossibility of inductive proof can help avoid real world errors.
    - chroma 11y ago
      Please notice how far afield you went to find a failure of inductive reasoning: Financial markets. Not a bad math proof. Not an incorrect physics theorem. Not even a civil engineering disaster. Markets! Markets aren't exactly known for regular behavior. If anything, they are anti-inductive.[1] If you find and exploit any regularity in them, the regularity eventually disappears. It's for these reasons that I remain unswayed by your claim. 1. http://lesswrong.com/lw/yv/markets_are_antiinductive/ http://lesswrong.com/lw/yv/markets_are_antiinductive/
      - graeme 11y ago
        Ok, Newtonian physics. That was accepted for a long time because it was utterly, utterly consistent with all our evidence. Until it wasn't, in some edge cases. And we adopted a new theory that explained things better. My point was that the impossibility of induction is at the core of science. Science generally does things well because it accounts for this. I cited financial crashes as an example of what can happen if you don't account for induction. What, exactly, are you disagreeing with?
pdabbadabba 11y ago
How it is that one can (seemingly) reason effectively using inductive logic despite its not being, according to Hume, formally valid is exactly the question Hume is interested in. You might think he's wrong to find this interesting, which is fine. But remember that the point of philosophy is to be grappled with--if you ignore the work just because you think the conclusion is incorrect, I submit that you're doing it wrong.
dpierce9 11y ago
Hume was deeply concerned to distinguish empirical reasoning from math and logic (matters of fact vs relations of ideas). Induction doesn't work in the sense that it doesn't guarantee its conclusion but of course even Hume thought it works whenever you are ok with something less than a guarantee (which he thought we are most of the time). Also, the problem of induction is only one part of this marvelous tome. Protip: get the new Fate-Norton Oxford edition and not the Selby-Bigge-Nidditch.
arstin 11y ago
Starting in philosophy is hard because it takes a lot of immersion to begin getting a sense for what people are even talking about. Most fields are like that, but philosophy is notable since it often uses normal words in (many different) technical ways. It can also be hard to stick with it long enough to feel even what the motivations for the problems are. The motivation for the "Humean problem of induction" along with a description of how it has evolved needs to be set up somewhat carefully. Marc Lange provides a nice intro if anyone is interested. http://www.stephanhartmann.org/HHL10_Lange.pdf http://www.stephanhartmann.org/HHL10_Lange.pdf But once you've got the basics from a "mere" phil of science perspective, you can go a lot further. For example, you can look at how "knowledge" is linked with human values and what it is to live a good life. Or at how the experience of art is connected to learning or human development. How does the "Humean problem of induction" bear on these themes? These far reaching connections---which typically involve an empirical or "scientific" component---are where philosophy really gets fun!

pron 11y ago

> Given that inductive reasoning is possible The question is not whether or not it is possible (in the sense that it is useful) but whether it leads to knowledge. This, of course, depends on your definition of knowledge, which is the question of epistemology. In some circles these days there is a move back to a utilitarian definition of knowledge, namely, if you can interact with a system and predict its behavior then you can say you know how it works, because your model is useful. This is not, however, the commonly accepted view of what constitutes knowledge. Funny thing: the consequence of the Gettier problem (sort of like the "incompleteness conjecture" of epistemology) is that either we must accept a very unsatisfying definition to what constitutes knowledge, or use a good, rigorous and satisfying definition and then find out that nothing we think we know actually qualifies.

marklgr 11y ago

Even if it's true, Hume was a great thinker and keeps historical value. Besides, I'm sure he would agree with your being skeptical of any book, whatever its theses :)

graeme 11y ago

Hume meant inductive proof. I checked Solomonoff induction, and it seems to be talking about probabilistic inductive reasoning. It isn't logically possible to prove anything affirmatively. That's what Hume was referring to. https://en.wikipedia.org/wiki/Problem_of_induction https://en.wikipedia.org/wiki/Problem_of_induction

duaneb 11y ago

You can reason effectively, but proving is another thing entirely. You can never prove beyond the shadow of a doubt we are not about to be razed by gamma rays--there could be some arbitrary rules to the universe that only kick in right now.

chroma 11y ago
Sorry, but it's stuff like this that causes me to avoid most philosophy. Yes, there are extremely small likelihoods for all kinds of extremely unlikely things. That doesn't make tools like induction any less useful. Sure you can't prove X, but you can show that the likelihood of ~X is far, far lower than the likelihood that you will hallucinate ~X.
- graeme 11y ago
  This one's at the foundation of our science though. Karl Popper's ideas on falsification have been very influential in how scientists frame their work. And errors of induction have caused some notable financial blowups because models missed tail risks. The example you replied to was farfetched, but not an abstract ivory tower notion – it's something deeply built into how we view the world. This doesn't mean there's no use of inductive reasoning, or that we shouldn't suspect that some things are far more likely true than others. But remembering the ultimate impossibility of inductive proof can help avoid real world errors.
  - chroma 11y ago
    Please notice how far afield you went to find a failure of inductive reasoning: Financial markets. Not a bad math proof. Not an incorrect physics theorem. Not even a civil engineering disaster. Markets! Markets aren't exactly known for regular behavior. If anything, they are anti-inductive.[1] If you find and exploit any regularity in them, the regularity eventually disappears. It's for these reasons that I remain unswayed by your claim. 1. http://lesswrong.com/lw/yv/markets_are_antiinductive/ http://lesswrong.com/lw/yv/markets_are_antiinductive/
    - graeme 11y ago
      Ok, Newtonian physics. That was accepted for a long time because it was utterly, utterly consistent with all our evidence. Until it wasn't, in some edge cases. And we adopted a new theory that explained things better. My point was that the impossibility of induction is at the core of science. Science generally does things well because it accounts for this. I cited financial crashes as an example of what can happen if you don't account for induction. What, exactly, are you disagreeing with?

pdabbadabba 11y ago

How it is that one can (seemingly) reason effectively using inductive logic despite its not being, according to Hume, formally valid is exactly the question Hume is interested in. You might think he's wrong to find this interesting, which is fine. But remember that the point of philosophy is to be grappled with--if you ignore the work just because you think the conclusion is incorrect, I submit that you're doing it wrong.

dpierce9 11y ago

Hume was deeply concerned to distinguish empirical reasoning from math and logic (matters of fact vs relations of ideas). Induction doesn't work in the sense that it doesn't guarantee its conclusion but of course even Hume thought it works whenever you are ok with something less than a guarantee (which he thought we are most of the time). Also, the problem of induction is only one part of this marvelous tome. Protip: get the new Fate-Norton Oxford edition and not the Selby-Bigge-Nidditch.

arstin 11y ago

Starting in philosophy is hard because it takes a lot of immersion to begin getting a sense for what people are even talking about. Most fields are like that, but philosophy is notable since it often uses normal words in (many different) technical ways. It can also be hard to stick with it long enough to feel even what the motivations for the problems are. The motivation for the "Humean problem of induction" along with a description of how it has evolved needs to be set up somewhat carefully. Marc Lange provides a nice intro if anyone is interested. http://www.stephanhartmann.org/HHL10_Lange.pdf http://www.stephanhartmann.org/HHL10_Lange.pdf But once you've got the basics from a "mere" phil of science perspective, you can go a lot further. For example, you can look at how "knowledge" is linked with human values and what it is to live a good life. Or at how the experience of art is connected to learning or human development. How does the "Humean problem of induction" bear on these themes? These far reaching connections---which typically involve an empirical or "scientific" component---are where philosophy really gets fun!