I am generally still very bad at steelmanning, but I think I am now capable of a very specific form of it. Namely, when people say something that sounds like a universal statement (“foos are bar”) I have learned to at least occasionally give them the benefit of the doubt instead of assuming that they literally mean “all foos are bar” and subsequently feeling smug when I point out a single counterexample to their statement. I have seen several people do this on LW lately and I am happy to report that I am now more annoyed at the strawmanners than the strawmanned in this scenario.
It sounds to me like it has a lot in common with the noncentral fallacy. There’s a general tendency to think of groups in terms of their central members and not their noncentral ones. This both makes sneaking in connotations by noncentral labels possible, and makes “all central foos are bar” feel like the same thing as “all foos are bar”.
Even more so with the “No foo is a bar”. Those statements are most probably either very common definitions like “no mammal is a bird” and therefore not very informative, either they are improbable. Like “no man can live more than X minutes without oxygen, ever”.
In the last case, even if the X is huge, we can assume that maybe it can be done under some (unseen yet) circumstances.
In other words, don’t be too hasty with universal negations!
Why can’t people say “some foos are bar” or “foos tend to be bar”? My default interpretation of “foos are bar” is “all foos are bar”. I tend to classify confident assertions that “foos are bar” with clear counterexamples as blustering. We already know from Philip Tetlock’s work that hedgehogs who make predictions based on simple models tend to be more confident, more widely quoted in the media, and more wrong than foxes who make equivocal, better-calibrated predictions based on more complicated models.
I think there may actually be a bit of a group coordination problem here—hedgehogs gain status from appearing confident and getting quoted in the media, but they’re spreading low-quality info. So it’s a case of personal gain at the expense of group loss. I’m inclined to call people out for hedgehog-style behavior as a way of dealing with this coordination problem. (In case it’s not obvious, I frequently see hedgehog-style predictions from LW-affiliated people and find them annoying and unconvincing.)
Why can’t people say “some foos are bar” or “foos tend to be bar”?
I mean, of course they can, but sometimes they won’t. People aren’t careful with their language and it’s uncharitable to assume that people mean what you think their words should have meant instead of what they most likely actually meant.
I also think you have a different prototypical case in mind than I do. I’m thinking the kind of nitpicking where someone says something like “fire is hot” and someone responds “nuh-uh, there’s a special type of fire you can make that is actually cool to the touch” or something like that.
People can’t/don’t say that “some foos are bar” or “foos tend to be bar” because it is often less accurate than “all foos are bar” or better yet “foos are bar”. This is because truth is fuzzy, not binary or digital. For example, “some humans have two arms” gives you very little information. Do 10 out of seven billion humans have two arms? 6.99999 billion out of 7 billion? Maybe half of humans?
By contrast the statement, “humans each have two arms” or even “all humans have two arms” is mostly true, probably better than 99% true, despite the existence of rare counter examples. You can make useful plans based on the knowledge that “all humans have two arms”.
If we see truth as binary, and allow a mostly true statement to be invalidated by a single rare counterexample, we have lost a lot of real information. If I know that 100% of humans have two arms, I have a more complete and accurate, though imperfect, view of the world than if I know only that “some humans have two arms”.
Best of all of course is if I know that 99.9834% +/- 0.0026% of humans have two arms. However absent such precise information, the statement “humans have two arms” is a pretty accurate and useful representation of reality.
I am generally still very bad at steelmanning, but I think I am now capable of a very specific form of it. Namely, when people say something that sounds like a universal statement (“foos are bar”) I have learned to at least occasionally give them the benefit of the doubt instead of assuming that they literally mean “all foos are bar” and subsequently feeling smug when I point out a single counterexample to their statement. I have seen several people do this on LW lately and I am happy to report that I am now more annoyed at the strawmanners than the strawmanned in this scenario.
It sounds to me like it has a lot in common with the noncentral fallacy. There’s a general tendency to think of groups in terms of their central members and not their noncentral ones. This both makes sneaking in connotations by noncentral labels possible, and makes “all central foos are bar” feel like the same thing as “all foos are bar”.
Even more so with the “No foo is a bar”. Those statements are most probably either very common definitions like “no mammal is a bird” and therefore not very informative, either they are improbable. Like “no man can live more than X minutes without oxygen, ever”.
In the last case, even if the X is huge, we can assume that maybe it can be done under some (unseen yet) circumstances.
In other words, don’t be too hasty with universal negations!
Why can’t people say “some foos are bar” or “foos tend to be bar”? My default interpretation of “foos are bar” is “all foos are bar”. I tend to classify confident assertions that “foos are bar” with clear counterexamples as blustering. We already know from Philip Tetlock’s work that hedgehogs who make predictions based on simple models tend to be more confident, more widely quoted in the media, and more wrong than foxes who make equivocal, better-calibrated predictions based on more complicated models.
I think there may actually be a bit of a group coordination problem here—hedgehogs gain status from appearing confident and getting quoted in the media, but they’re spreading low-quality info. So it’s a case of personal gain at the expense of group loss. I’m inclined to call people out for hedgehog-style behavior as a way of dealing with this coordination problem. (In case it’s not obvious, I frequently see hedgehog-style predictions from LW-affiliated people and find them annoying and unconvincing.)
I mean, of course they can, but sometimes they won’t. People aren’t careful with their language and it’s uncharitable to assume that people mean what you think their words should have meant instead of what they most likely actually meant.
I also think you have a different prototypical case in mind than I do. I’m thinking the kind of nitpicking where someone says something like “fire is hot” and someone responds “nuh-uh, there’s a special type of fire you can make that is actually cool to the touch” or something like that.
Fair.
People can’t/don’t say that “some foos are bar” or “foos tend to be bar” because it is often less accurate than “all foos are bar” or better yet “foos are bar”. This is because truth is fuzzy, not binary or digital. For example, “some humans have two arms” gives you very little information. Do 10 out of seven billion humans have two arms? 6.99999 billion out of 7 billion? Maybe half of humans?
By contrast the statement, “humans each have two arms” or even “all humans have two arms” is mostly true, probably better than 99% true, despite the existence of rare counter examples. You can make useful plans based on the knowledge that “all humans have two arms”.
If we see truth as binary, and allow a mostly true statement to be invalidated by a single rare counterexample, we have lost a lot of real information. If I know that 100% of humans have two arms, I have a more complete and accurate, though imperfect, view of the world than if I know only that “some humans have two arms”.
Best of all of course is if I know that 99.9834% +/- 0.0026% of humans have two arms. However absent such precise information, the statement “humans have two arms” is a pretty accurate and useful representation of reality.