“I want to live one more day. Tomorrow I will still want to live one more day. Therefore I want to live forever, proof by induction on the positive integers.”
This immediately caught my attention, given that Harry talks in earlier chapters about his worldview relying on Bayesian inference. Yet, for induction over an infinite sequence of unknown, informative experiences to hold, he has to have assigned an integral prior. Hijinks!
My first thought was that this was a clue dropped by the author to Harry’s blind spot and potentially tragic flaw—after all, “will you notice your confusion” is right there on the tagline! The subsequent emergence of Harry’s Patronus spell as the True Patronus, based on this very conviction, undermined that theory somewhat, though if you elide the “True” part of it (ascribing the name to Harry’s preteen self-admiration, maybe), the special magnitude of Harry’s Patronus can be handwaved by the enormity of his happy thought and not necessarily the sound logic or underlying reality of it.
This has been on my mind for a while, but the most recent Author’s Notes (“nothing is there to deliberately fool the readers”) seem to confirm that this analysis is far too subtle: that if the narration calls Harry’s Patronus the True Patronus, it is; if Harry’s absolute faith in universal immortality has been given plot support, then it should be taken as axiomatic to the fiction.
Any ideas? Has Harry accidentally arrived at a true conclusion by non-inferential means?
Eliezer has used that line in nonfictiontoo; I’m very confident that Harry’s pro-immortality stance is endorsed by the author, but that the “induction proof” is meant rhetorically and should not be construed to imply infinite certainty.
Ah! Thanks for that background. Can you explain, though, why you think that statement is meant to be Harry taking the piss? (Within the text, that is—Eliezer does rightly frame it as a joke in those links). Harry’s surrounding statements are sincerely put, and the next paragraph suggests to me that Harry believes that the induction argument should have refuted Dumbledore:
The two cultures stared at each other across a vast gap of incommensurability.
In any case, I do hope that, at some point, Harry has to face down the taboo tradeoffs (to be topical to the current arc) implied by universal immortality.
Oh, dear. I suddenly seem to have acquired the desire to write a HPMOR/Do the Math crossover.
I don’t think Harry is actually taking the piss, and nor does he see it as a literal proof. It helps to remember who he’s talking to. He’s trying to get Dumbledore to consider not just death, sometime in the far abstract future, but a thing that you might actually welcome even just one day after a day when you didn’t welcome it. Not a proof but a rhetorical device.
Well, and here’s where it gets interesting: are there any other places where we see Harry use logic that he knows (or should know) to be unsound in an instrumental fashion? That is, where he makes a tactical choice to argue nonsense, believing it to have a better chance of convincing someone who disagrees with him?
Harry should consider the possibility that he “might actually welcome [death] even just one day after a day when [he] didn’t welcome it”—if he can’t anticipate the possibility of his utility function changing based on an infinity of new evidence, he should stop pretending to be solely rationalist. Which, interestingly, Chapter 82 seems to be hinting at.
Honestly, that comes across as a flaw in Eliezer’s worldview more so than Harry’s. I’ve seen him make the same argument in his own name, and it’s pretty transparently false(cf. anyone committing suicide, ever). Being forced to die is evil and ought to be opposed, but I have a feeling that literal immortality would appeal to many fewer people than might be expected.
Are you actually trying to suggest that literally nobody who has ever committed suicide has genuinely wanted to not be alive? I mean sure, there’s the “cry for help” gone wrong, and similar, but there’s also the ones who actually want to die.
Are you actually trying to suggest that literally nobody who has ever committed suicide has genuinely wanted to not be alive? [...] but there’s also the ones who actually want to die.
True, suicide by mistake is probably rare, but I don’t think that’s what Vladimir meant.
If I’m tortured and I can’t find a way of making it stop, I might “want to die”. But that’s not because I don’t want to live, it’s because I don’t want to live in torture.
I don’t really have much knowledge of the subject, but my impression was that most suicides are of the “I can’t live like this anymore” kind, not “dying would be kinda cool”, albeit with the “like this”=“torture” being a YMMV issue in many cases.
I’m not saying there aren’t other possibilities, just that simply pointing to suicides without more careful analysis is not clear proof of Harry’s induction being wrong.
Harry’s induction is only correct if either wanting to live one more day implies wanting to live two more days as a mathematical law(which is not true) and that it’s impossible to change your mind partway through a two-day period(which is also not true), or if he knows what he will desire at every possible future date. Since neither of those conditions holds, mathematical induction fails.
I’m drawing attention to the reasoning steps, not the conclusion. My point is that it’s generally incorrect that people’s behavior contrary to some goal implies that goal not being held, that it’s only weak evidence.
Killing yourself is weak evidence of wanting to be dead?
Strangely, yes. The trouble is, for a human it’s not that simple, there are many senses of having a goal that may well disagree with each other. The post I linked distinguished “urges” (more immediate tendencies that control one’s behavior) and abstract goals, but urges could be further subdivided into wants and likes, what one tends to seek vs. what one enjoys having been done, and the structure of abstract goals is potentially much more complicated. Any single reading, however dramatic, doesn’t reveal the details of this picture.
I say again, you seem to have ruled out any possibility of strong evidence being presented for the thesis. What’s an observation that you would consider strong evidence, if actual suicide is weak? And if there isn’t one, then why should I assume that you’re arguing in good faith?
You perceive this as an argument, but I have no interest in arguing about anything in particular, except to point out that technical inaccuracy about the meaning of “wanting” in your original comment (which you should judge on object level, based on what you conclude from reading the linked posts, if you choose to do so, not from how you perceive the context of me linking to those posts). That is all I meant to do, but arguing about what I meant to do is similarly beside the point, so I’m bowing out.
The point is, I can conceive of no stronger evidence that a person wants to die than to observe that they committed suicide. Mocking the strongest evidence available for being weak is nonsensical.
(I don’t think that a single action, however dramatic or final, is the strongest available evidence about goals. This comment is then a second out-of-context technical remark that shouldn’t be taken as a relevant argument in the preceding context.)
Mocking the strongest evidence available for being weak is nonsensical.
“Death is bad” is a true conclusion that Harry has arrived at through legitimate means. “I will always want to live forever” is utter nonsense, but it’s not necessary for the True Patronus.
And my problem, here, is that “death is bad” cannot be an unqualified truth. “Human death is bad” can be aspirationally true, and I am willing to believe that the unprecedented depth of Harry’s aspiration might be the key that unlocks the power of his Patronus—but it does look like EY means it literally, and that means that Harry should at some point need to distinguish between his ideology and that of, as Edward Abbey puts it, “the ideology of the cancer cell.”
I have a confusion!
Way back in Chapter 39, Harry says:
This immediately caught my attention, given that Harry talks in earlier chapters about his worldview relying on Bayesian inference. Yet, for induction over an infinite sequence of unknown, informative experiences to hold, he has to have assigned an integral prior. Hijinks!
My first thought was that this was a clue dropped by the author to Harry’s blind spot and potentially tragic flaw—after all, “will you notice your confusion” is right there on the tagline! The subsequent emergence of Harry’s Patronus spell as the True Patronus, based on this very conviction, undermined that theory somewhat, though if you elide the “True” part of it (ascribing the name to Harry’s preteen self-admiration, maybe), the special magnitude of Harry’s Patronus can be handwaved by the enormity of his happy thought and not necessarily the sound logic or underlying reality of it.
This has been on my mind for a while, but the most recent Author’s Notes (“nothing is there to deliberately fool the readers”) seem to confirm that this analysis is far too subtle: that if the narration calls Harry’s Patronus the True Patronus, it is; if Harry’s absolute faith in universal immortality has been given plot support, then it should be taken as axiomatic to the fiction.
Any ideas? Has Harry accidentally arrived at a true conclusion by non-inferential means?
Eliezer has used that line in nonfiction too; I’m very confident that Harry’s pro-immortality stance is endorsed by the author, but that the “induction proof” is meant rhetorically and should not be construed to imply infinite certainty.
Ah! Thanks for that background. Can you explain, though, why you think that statement is meant to be Harry taking the piss? (Within the text, that is—Eliezer does rightly frame it as a joke in those links). Harry’s surrounding statements are sincerely put, and the next paragraph suggests to me that Harry believes that the induction argument should have refuted Dumbledore:
In any case, I do hope that, at some point, Harry has to face down the taboo tradeoffs (to be topical to the current arc) implied by universal immortality.
Oh, dear. I suddenly seem to have acquired the desire to write a HPMOR/Do the Math crossover.
I don’t think Harry is actually taking the piss, and nor does he see it as a literal proof. It helps to remember who he’s talking to. He’s trying to get Dumbledore to consider not just death, sometime in the far abstract future, but a thing that you might actually welcome even just one day after a day when you didn’t welcome it. Not a proof but a rhetorical device.
Well, and here’s where it gets interesting: are there any other places where we see Harry use logic that he knows (or should know) to be unsound in an instrumental fashion? That is, where he makes a tactical choice to argue nonsense, believing it to have a better chance of convincing someone who disagrees with him?
Harry should consider the possibility that he “might actually welcome [death] even just one day after a day when [he] didn’t welcome it”—if he can’t anticipate the possibility of his utility function changing based on an infinity of new evidence, he should stop pretending to be solely rationalist. Which, interestingly, Chapter 82 seems to be hinting at.
Honestly, that comes across as a flaw in Eliezer’s worldview more so than Harry’s. I’ve seen him make the same argument in his own name, and it’s pretty transparently false(cf. anyone committing suicide, ever). Being forced to die is evil and ought to be opposed, but I have a feeling that literal immortality would appeal to many fewer people than might be expected.
Please distinguish behavior from preference. See for example this post: Urges vs. Goals: The analogy to anticipation and belief.
Are you actually trying to suggest that literally nobody who has ever committed suicide has genuinely wanted to not be alive? I mean sure, there’s the “cry for help” gone wrong, and similar, but there’s also the ones who actually want to die.
True, suicide by mistake is probably rare, but I don’t think that’s what Vladimir meant.
If I’m tortured and I can’t find a way of making it stop, I might “want to die”. But that’s not because I don’t want to live, it’s because I don’t want to live in torture.
I don’t really have much knowledge of the subject, but my impression was that most suicides are of the “I can’t live like this anymore” kind, not “dying would be kinda cool”, albeit with the “like this”=“torture” being a YMMV issue in many cases.
I’m not saying there aren’t other possibilities, just that simply pointing to suicides without more careful analysis is not clear proof of Harry’s induction being wrong.
Harry’s induction is only correct if either wanting to live one more day implies wanting to live two more days as a mathematical law(which is not true) and that it’s impossible to change your mind partway through a two-day period(which is also not true), or if he knows what he will desire at every possible future date. Since neither of those conditions holds, mathematical induction fails.
I’m drawing attention to the reasoning steps, not the conclusion. My point is that it’s generally incorrect that people’s behavior contrary to some goal implies that goal not being held, that it’s only weak evidence.
Killing yourself is weak evidence of wanting to be dead? Pray tell, what would strong evidence be? Killing yourself twice?
Strangely, yes. The trouble is, for a human it’s not that simple, there are many senses of having a goal that may well disagree with each other. The post I linked distinguished “urges” (more immediate tendencies that control one’s behavior) and abstract goals, but urges could be further subdivided into wants and likes, what one tends to seek vs. what one enjoys having been done, and the structure of abstract goals is potentially much more complicated. Any single reading, however dramatic, doesn’t reveal the details of this picture.
I say again, you seem to have ruled out any possibility of strong evidence being presented for the thesis. What’s an observation that you would consider strong evidence, if actual suicide is weak? And if there isn’t one, then why should I assume that you’re arguing in good faith?
You perceive this as an argument, but I have no interest in arguing about anything in particular, except to point out that technical inaccuracy about the meaning of “wanting” in your original comment (which you should judge on object level, based on what you conclude from reading the linked posts, if you choose to do so, not from how you perceive the context of me linking to those posts). That is all I meant to do, but arguing about what I meant to do is similarly beside the point, so I’m bowing out.
The point is, I can conceive of no stronger evidence that a person wants to die than to observe that they committed suicide. Mocking the strongest evidence available for being weak is nonsensical.
(I don’t think that a single action, however dramatic or final, is the strongest available evidence about goals. This comment is then a second out-of-context technical remark that shouldn’t be taken as a relevant argument in the preceding context.)
Please see Katja’s post Estimation is the best we have. The strongest available evidence may well be weak, which doesn’t mean that you don’t go with the strongest available evidence, but going with it also doesn’t make it strong.
“Death is bad” is a true conclusion that Harry has arrived at through legitimate means. “I will always want to live forever” is utter nonsense, but it’s not necessary for the True Patronus.
And my problem, here, is that “death is bad” cannot be an unqualified truth. “Human death is bad” can be aspirationally true, and I am willing to believe that the unprecedented depth of Harry’s aspiration might be the key that unlocks the power of his Patronus—but it does look like EY means it literally, and that means that Harry should at some point need to distinguish between his ideology and that of, as Edward Abbey puts it, “the ideology of the cancer cell.”
If (is a day) then (want to live one more day) If (is day after day) then (want to live one more day) Today, want to live one more day.
Therefore, want to live forever.
2⁄3 of those are false claims. Seriously, for a rationalist site, this is an astonishingly poor argument to see getting thrown around.