(The parent is a long comment and makes several points, so I’m going to answer it in several parts. This is part 2.)
I hope it is also clear how the original comment in this chain is not a good pointer towards this presentation of a way in which Benquo’s post could be deeply mistaken.
The point implied by the original comment in this chain is absolutely critical. It disappoints me to see it (mostly) dismissed, because I think it is emblematic of a deep and pervasive problem with current trends in “rationalist” thought. (I elaborate somewhat further in this comment.)
One hypothesis that seems too simplistic to be right is something like “Said is generally skeptical of counterfactual reasoning.”
In a sense, this is, indeed, accurate. I do not mean to recapitulate, here, in this comment thread, the entirety of the philosophical debates about counterfactuals, but my view (which to me to be relatively uncontroversial) is this:
… suppose I perform some behavior (like mashing potatoes with a glass), and we then discuss what would have happened if instead I performed a different behavior (like mashing potatoes with a whisk).
It seems to me that we can have a meaningful discussion about this counterfactual scenario to the extent that we can transform it into questions like:
“What has happened in the past, in situations where I have mashed potatoes with a whisk?”
“What has happened in the past, in situations similar to the above—for example, cases where I have mashed turnips with a whisk, or mashed potatoes with a fork?”
“What do I predict will happen in the future, if I do any of the above things?”
#1 and #2 concern mundane facts, which are known to us. (We can also modify them along the lines of “not I, but my friend Ann, mashed potatoes with a whisk”, or “I heard from my mother about someone who mashed potatoes with a whisk”, etc.) #3 concerns predictions, which become informative once made and then tested.
In the “time-travel story case”, #1 doesn’t apply for obvious reasons; #2 doesn’t apply unless you relax the standard of similarity so far that it becomes useless; #3 also doesn’t apply because—at least for the foreseeable future—we can’t do this experiment. So any discussion about “what would happen” in this imaginary scenario is nonsense; nothing we can say about “what would happen” can be true, or false, in any meaningful way.
It seems to me like you have many responses, ranging from “I agree, because X” to “I disagree, because Y” to “I don’t think this question is resolvable” to “I don’t think this question is interesting.” It seems to me like there’s interesting material in the first two responses, even if the third response is in fact valid.
The question is unresolvable and uninteresting; I don’t see what there is to agree or disagree about.
A potentially absurd example: it seems to me like there’s a consistent view in which the mathematical technique of proof by contradiction is classed with “generalizing from fictional evidence.”
I agree that this is absurd. What you’re talking about here is formal systems; that is a very different case. (I originally typed out an extensive rebuttal of your example, but to be honest, it seems to me like it’s simply a non sequitur, which makes the rebuttal moot.)
The point implied by the original comment in this chain is absolutely critical.
There are two tracks here:
1) If the point is critical, implying it is perhaps not sufficient, and the thing should be spelled out. Short comments often imply many distinct generators.
In particular, the generator that I think is most obvious is “there’s a genre mismatch between ‘the ordinary means by which people navigate their lives’ and ‘time-travel stories’, as the first is nonfictional and the second is fictional; don’t generalize from fictional evidence!”, but the position that I saw as a more serious objection from Benquo’s point of view was “while there are powerful models here, there’s also insight porn that feels powerful but isn’t; it is not clear that the dimension you highlight separates the two as opposed to leading you towards insight porn.”
2) Whether or not the point is correct. As it happens, I think the second concern (that this doesn’t reliably distinguish insight porn from true insight) is interesting and important, and that the first concern (that time travel stories are completely distinct from ordinary lives) is mistaken.
A brief comment on why it’s mistaken: Robinson Crusoe is a fictional example, yes, but it’s a fictional member of a real class, and in explanatory pieces you should expect the author to use examples the audience will know, and those will generally be fictional examples because of both higher audience recognition and fictional examples can more crisply separate out the real thing. The ordinary means by which people navigate their lives includes losing some foundations of support, venturing into the unknown, and making tools out of their constituent parts; there is a meaningful way in which any programmer who opens up a new text window is doing something cognitively similar to Robinson Crusoe.
[A frame people can adopt, which is sometimes useful, is that they’re an amnesiac time traveler from 3018; what thing can they do now, even though they don’t remember what it is? See Archimedes Chronophone: the point here is a subtle one that’s not “remember something from the future” because you’ve forgotten it, it’s “what happens if I take seriously the possibility that there are major opportunities that are accessible to someone now just because they know something, and what modes of thought might lead to discovering that thing?”.]
[This is just responding to points that are easy to respond to contained in the parent; my overall sense is “it might take a post to point at what’s going on here, and so I’m going to try to write that post instead of handle it here.”]
This is as good a time as any for me to mention that this term “generators”, which you’ve used a few times now, is not familiar to me in this context. I think I can sort of guess the general meaning from context, but I’m really not sure. Where is it from? Is it idiosyncratic to you, or…?
Anyhow, as to your #1, both objections you describe are important. Yes, don’t generalize from fictional evidence, and also avoid insight porn.
Robinson Crusoe is a fictional example, yes, but it’s a fictional member of a real class …
What class? People who’ve gotten shipwrecked? Or people who’ve gotten shipwrecked and managed to sustain themselves via their resourcefulness, etc.? Or something else?
If an author chooses to use a fictional example, then the specific real class of which the fictional example is a member should be identified explicitly, and as many examples as possible of real members of that class should be provided.
… and in explanatory pieces you should expect the author to use examples the audience will know, and those will generally be fictional examples because of both higher audience recognition and fictional examples can more crisply separate out the real thing.
I do not at all agree that this is a reasonable expectation. In fact, I think that reliance on fictional examples is a deep and pervasive problem in “rationalist” writing (and thought), and one which has done much to corrupt the epistemics of the rationalist community/movement. I can hardly think of terms too strong in which to object to this practice. I think it would be a very good idea to excise it, root and stem. (Perhaps, one day, we may trust ourselves with the use of fictional examples once more; but not now, and not for some time.)
The ordinary means by which people navigate their lives includes losing some foundations of support, venturing into the unknown, and making tools out of their constituent parts; there is a meaningful way in which any programmer who opens up a new text window is doing something cognitively similar to Robinson Crusoe.
I’m sorry, but I think that this is an absurd analogy. This is “in some sense…” type reasoning taken much, much too far.
(As for the bit about the chronophone, well… I don’t think it’s critical to your points, so I won’t take up more space and time with my views on it. But if you think it’s a critical point, then I’ll respond to that, as I certainly do have opinions on it.)
I agree time travel is nothing which applies to #1,#2 or #3. However Im curious:
Are time travel stories not taking points #1 and #2 in order to create a szenario to explore #3? Meaning I create a situation and in order to make it explorable and detatch it from people’s expectations I use the literary device of time travel.
(The parent is a long comment and makes several points, so I’m going to answer it in several parts. This is part 2.)
The point implied by the original comment in this chain is absolutely critical. It disappoints me to see it (mostly) dismissed, because I think it is emblematic of a deep and pervasive problem with current trends in “rationalist” thought. (I elaborate somewhat further in this comment.)
In a sense, this is, indeed, accurate. I do not mean to recapitulate, here, in this comment thread, the entirety of the philosophical debates about counterfactuals, but my view (which to me to be relatively uncontroversial) is this:
It seems to me that we can have a meaningful discussion about this counterfactual scenario to the extent that we can transform it into questions like:
“What has happened in the past, in situations where I have mashed potatoes with a whisk?”
“What has happened in the past, in situations similar to the above—for example, cases where I have mashed turnips with a whisk, or mashed potatoes with a fork?”
“What do I predict will happen in the future, if I do any of the above things?”
#1 and #2 concern mundane facts, which are known to us. (We can also modify them along the lines of “not I, but my friend Ann, mashed potatoes with a whisk”, or “I heard from my mother about someone who mashed potatoes with a whisk”, etc.) #3 concerns predictions, which become informative once made and then tested.
In the “time-travel story case”, #1 doesn’t apply for obvious reasons; #2 doesn’t apply unless you relax the standard of similarity so far that it becomes useless; #3 also doesn’t apply because—at least for the foreseeable future—we can’t do this experiment. So any discussion about “what would happen” in this imaginary scenario is nonsense; nothing we can say about “what would happen” can be true, or false, in any meaningful way.
The question is unresolvable and uninteresting; I don’t see what there is to agree or disagree about.
I agree that this is absurd. What you’re talking about here is formal systems; that is a very different case. (I originally typed out an extensive rebuttal of your example, but to be honest, it seems to me like it’s simply a non sequitur, which makes the rebuttal moot.)
There are two tracks here:
1) If the point is critical, implying it is perhaps not sufficient, and the thing should be spelled out. Short comments often imply many distinct generators.
In particular, the generator that I think is most obvious is “there’s a genre mismatch between ‘the ordinary means by which people navigate their lives’ and ‘time-travel stories’, as the first is nonfictional and the second is fictional; don’t generalize from fictional evidence!”, but the position that I saw as a more serious objection from Benquo’s point of view was “while there are powerful models here, there’s also insight porn that feels powerful but isn’t; it is not clear that the dimension you highlight separates the two as opposed to leading you towards insight porn.”
2) Whether or not the point is correct. As it happens, I think the second concern (that this doesn’t reliably distinguish insight porn from true insight) is interesting and important, and that the first concern (that time travel stories are completely distinct from ordinary lives) is mistaken.
A brief comment on why it’s mistaken: Robinson Crusoe is a fictional example, yes, but it’s a fictional member of a real class, and in explanatory pieces you should expect the author to use examples the audience will know, and those will generally be fictional examples because of both higher audience recognition and fictional examples can more crisply separate out the real thing. The ordinary means by which people navigate their lives includes losing some foundations of support, venturing into the unknown, and making tools out of their constituent parts; there is a meaningful way in which any programmer who opens up a new text window is doing something cognitively similar to Robinson Crusoe.
[A frame people can adopt, which is sometimes useful, is that they’re an amnesiac time traveler from 3018; what thing can they do now, even though they don’t remember what it is? See Archimedes Chronophone: the point here is a subtle one that’s not “remember something from the future” because you’ve forgotten it, it’s “what happens if I take seriously the possibility that there are major opportunities that are accessible to someone now just because they know something, and what modes of thought might lead to discovering that thing?”.]
[This is just responding to points that are easy to respond to contained in the parent; my overall sense is “it might take a post to point at what’s going on here, and so I’m going to try to write that post instead of handle it here.”]
This is as good a time as any for me to mention that this term “generators”, which you’ve used a few times now, is not familiar to me in this context. I think I can sort of guess the general meaning from context, but I’m really not sure. Where is it from? Is it idiosyncratic to you, or…?
Anyhow, as to your #1, both objections you describe are important. Yes, don’t generalize from fictional evidence, and also avoid insight porn.
What class? People who’ve gotten shipwrecked? Or people who’ve gotten shipwrecked and managed to sustain themselves via their resourcefulness, etc.? Or something else?
If an author chooses to use a fictional example, then the specific real class of which the fictional example is a member should be identified explicitly, and as many examples as possible of real members of that class should be provided.
I do not at all agree that this is a reasonable expectation. In fact, I think that reliance on fictional examples is a deep and pervasive problem in “rationalist” writing (and thought), and one which has done much to corrupt the epistemics of the rationalist community/movement. I can hardly think of terms too strong in which to object to this practice. I think it would be a very good idea to excise it, root and stem. (Perhaps, one day, we may trust ourselves with the use of fictional examples once more; but not now, and not for some time.)
I’m sorry, but I think that this is an absurd analogy. This is “in some sense…” type reasoning taken much, much too far.
(As for the bit about the chronophone, well… I don’t think it’s critical to your points, so I won’t take up more space and time with my views on it. But if you think it’s a critical point, then I’ll respond to that, as I certainly do have opinions on it.)
I agree time travel is nothing which applies to #1,#2 or #3. However Im curious:
Are time travel stories not taking points #1 and #2 in order to create a szenario to explore #3? Meaning I create a situation and in order to make it explorable and detatch it from people’s expectations I use the literary device of time travel.