How far are you in the book? If you’re stuck in Part II, I would recommend skipping to its last section. In my opinion, for somebody just starting out with his philosophy, the rest of that part simply isn’t insightful enough to justify how difficult it is to read. Save it for later, if at all.
Remember that he wrote it over 200 years ago. You’ll have to spend a lot of time getting fluent in his idiosyncratic 18th century English to really get what he’s saying. I find that sort of thing interesting, so it was actually a bonus for me. But if it would only be an obstacle for you, and you have a sufficiently high time preference for this kind of thing, you might be better off sticking to something that uses more familiar language.
Now, I want to say something about his philosophy.
He used a rare method that I call “first-person epistemology” (FPE). He didn’t start out from the usual premise: that he was but one mind in a physical universe. No, he began much deeper: from nothing but the immediately given. His world was simply a sequence of sensatons. For example, he didn’t directly apprehend 3D space. His senses conveyed only a sequence of 2D images on his visual field. If the term “3D space” is to mean anything, we must define it as referring to a particular kind of sequence of those 2D images. Our belief in 3D space pays rent by helping us predict what 2D image we’ll experience in what situation (perhaps among other things).
I think that this method (FPE) is extremely important, but nobody ever seems to employ it. I’ve only seen two people: him and Berkeley. Perhaps there have been others. I’m still looking. Based on some bio I read a while ago, Carnap seemed to fit the bill, but I don’t really know. I haven’t tried him yet. I can’t read German, and I hate reading translations. They usually suck. Anyway, I said that nobody seems to do Hume any justice. I’m not prepared to substantiate this, but I think that at least part of that is because they don’t understand his method (FPE). Nobody seems to get FPE, even though it seems totally obvious to me.
I think that the real progress to be made in AI is in understanding how our own consciousness works. If we can understand our own action, we can build an actor. Maybe even a better one. But how could we do that? I think that FPE is the way to go, and Hume did it best. Human Nature (or at least Book I and some of the parts of Books II and III) is an excellent monograph on how our consciousness works. But what do I know? I’m not well-read in AI. Perhaps they all employ FPE like it’s nothing, and they’re all well past Hume’s stuff. No idea. Maybe you could let me know? Any idea?
Anyway, a few more things:
If you’re having trouble with a section, I might be able to help. I generally know what he’s talking about, and I can usually translate his points into more modern wording.
I think that he’s extremely important, and I think that his treatise is his best work. He’s tied with Mises as my favorite writer, and his treatise is tied with Human Action as my favorite book. I don’t have any authority around here, but perhaps this means something to you.
If I wanted to be cocky, I would say that you probably wouldn’t get anything important from Hume that you wouldn’t get better and easier from my future posts anyway. I intend to try to convey a lot of important stuff to this community, and Hume is one of my two biggest influences.
But enough of all that stuff. Let’s get to the real question. What are your goals? Why do you think that you would be better off reading what you mentioned instead of Hume? What would they have that Hume wouldn’t? What exactly are you trying to accomplish by reading this stuff? After all, where to turn always comes down to where you’re trying to go. I can’t have an opinion on whether you’re wrong until you tell me what you’re attempting to do.
I find that I must retract my statement (in great great grandparent) that first-person epistemology (FPE) is not used in the writings in AI around here.
In particular, one of the most heavily referenced lines of AI research around here is AIXI, which is essentially a single equation, namely,
(along with a mathematical tome’s worth of exposition to explain the significance of the equation).
There are four kinds of “points” or bound variables in the equation: computer programs (represented by the bound variable q), rewards (r), actions (a) and observations (o).
If you examine the text surrounding the above equation, you find that the author gives “a camera image” as an example of an observation.
I haven’t read Hume, but given what you have said about FPE above, this AIXI formalism seems like an instance of FPE.
I realize that it is unlikely that you want to learn enough math to understand this AIXI formalism, but I felt I had to bring AIXI up to stop the propagation of the probably-false information I had introduced in great great grandparent. Note that I probably do not have time to learn anything new about philosophy or to explain how AIXI might relate to the philosophical traditions or lines you are interested in.
First, a couple general considerations:
How far are you in the book? If you’re stuck in Part II, I would recommend skipping to its last section. In my opinion, for somebody just starting out with his philosophy, the rest of that part simply isn’t insightful enough to justify how difficult it is to read. Save it for later, if at all.
Remember that he wrote it over 200 years ago. You’ll have to spend a lot of time getting fluent in his idiosyncratic 18th century English to really get what he’s saying. I find that sort of thing interesting, so it was actually a bonus for me. But if it would only be an obstacle for you, and you have a sufficiently high time preference for this kind of thing, you might be better off sticking to something that uses more familiar language.
Now, I want to say something about his philosophy.
He used a rare method that I call “first-person epistemology” (FPE). He didn’t start out from the usual premise: that he was but one mind in a physical universe. No, he began much deeper: from nothing but the immediately given. His world was simply a sequence of sensatons. For example, he didn’t directly apprehend 3D space. His senses conveyed only a sequence of 2D images on his visual field. If the term “3D space” is to mean anything, we must define it as referring to a particular kind of sequence of those 2D images. Our belief in 3D space pays rent by helping us predict what 2D image we’ll experience in what situation (perhaps among other things).
I think that this method (FPE) is extremely important, but nobody ever seems to employ it. I’ve only seen two people: him and Berkeley. Perhaps there have been others. I’m still looking. Based on some bio I read a while ago, Carnap seemed to fit the bill, but I don’t really know. I haven’t tried him yet. I can’t read German, and I hate reading translations. They usually suck. Anyway, I said that nobody seems to do Hume any justice. I’m not prepared to substantiate this, but I think that at least part of that is because they don’t understand his method (FPE). Nobody seems to get FPE, even though it seems totally obvious to me.
I think that the real progress to be made in AI is in understanding how our own consciousness works. If we can understand our own action, we can build an actor. Maybe even a better one. But how could we do that? I think that FPE is the way to go, and Hume did it best. Human Nature (or at least Book I and some of the parts of Books II and III) is an excellent monograph on how our consciousness works. But what do I know? I’m not well-read in AI. Perhaps they all employ FPE like it’s nothing, and they’re all well past Hume’s stuff. No idea. Maybe you could let me know? Any idea?
Anyway, a few more things:
If you’re having trouble with a section, I might be able to help. I generally know what he’s talking about, and I can usually translate his points into more modern wording.
I think that he’s extremely important, and I think that his treatise is his best work. He’s tied with Mises as my favorite writer, and his treatise is tied with Human Action as my favorite book. I don’t have any authority around here, but perhaps this means something to you.
If I wanted to be cocky, I would say that you probably wouldn’t get anything important from Hume that you wouldn’t get better and easier from my future posts anyway. I intend to try to convey a lot of important stuff to this community, and Hume is one of my two biggest influences.
But enough of all that stuff. Let’s get to the real question. What are your goals? Why do you think that you would be better off reading what you mentioned instead of Hume? What would they have that Hume wouldn’t? What exactly are you trying to accomplish by reading this stuff? After all, where to turn always comes down to where you’re trying to go. I can’t have an opinion on whether you’re wrong until you tell me what you’re attempting to do.
Certainly not the writings in AI discussed on LW. Probably not any other writings either.
Isn’t that what I said? I don’t get what you’re trying to say here.
ETA: Oh, are you responding to “perhaps they all employ FPE like it’s nothing”? At first, I thought you were responding to “I’m not well-read in AI”.
Yes. Sorry for the unintended ambiguity.
I see. No problem.
By the way, do you have an opinion on whether it’s good or bad that nobody in the AI community seems to employ FPE?
I find that I must retract my statement (in great great grandparent) that first-person epistemology (FPE) is not used in the writings in AI around here.
In particular, one of the most heavily referenced lines of AI research around here is AIXI, which is essentially a single equation, namely,
http://www.hutter1.net/ai/aixi1linel.gif
(along with a mathematical tome’s worth of exposition to explain the significance of the equation).
There are four kinds of “points” or bound variables in the equation: computer programs (represented by the bound variable q), rewards (r), actions (a) and observations (o).
If you examine the text surrounding the above equation, you find that the author gives “a camera image” as an example of an observation.
I haven’t read Hume, but given what you have said about FPE above, this AIXI formalism seems like an instance of FPE.
I realize that it is unlikely that you want to learn enough math to understand this AIXI formalism, but I felt I had to bring AIXI up to stop the propagation of the probably-false information I had introduced in great great grandparent. Note that I probably do not have time to learn anything new about philosophy or to explain how AIXI might relate to the philosophical traditions or lines you are interested in.