Epistemic status: this is an attempt to steelman the case for the death penalty
...
I do not believe in vengeance or justice. I do however believe in fixing problems. And it’s clear the only way to fix this problem is to put such people in positions where they cannot do anyone any harm.
Some people have complained that, when their opponents “steelman” their position, in practice it can mean they steelman a particular argument that is not their main argument. This struck me as a remarkably explicit and self-aware example of that.
I don’t know what the solution is. Maybe tell people not to use “steelmanning” in such cases, maybe tell people to stop expecting “steelmanning” to necessarily mean it won’t miss a central argument. Maybe decide that you should, e.g., say “I’m steelmanning this particular argument”, because if you say “I’m steelmanning the case for this conclusion” then that means you’re supposed to capture all important arguments for that conclusion.
Steelmanning is not the same as passing the ideological Turing test.
ITT is successful when your opponent agrees with you, or when your opponent cannot distinguish you from their actual allies.
Steelman is successful if you, or your allies, find something useful in the ideas of your opponent. Whether your opponent approved of the result, or not.
In ITT, your opponent is the judge (of how much it passes). With a steelman, you are the judge (of whether you have extracted something useful). When steelmanning, you cherry-pick the good parts, and discard the rest. When passing an ITT, you need to pass all checks.
For example, an ITT of a religion is… speaking like a true believer. A steelman of religion is e.g. saying that we don’t really know how the universe came to existence, and that there are some social benefits of religion.
Steelmanning is writing retcon fanfiction of your interlocutor’s arguments. As such it necessarily adds, omits, or changes elements of the source material, in ways that the other person need not accept as a valid statement of their views.
Some people have complained that, when their opponents “steelman” their position, in practice it can mean they steelman a particular argument that is not their main argument. This struck me as a remarkably explicit and self-aware example of that.
I don’t know what the solution is. Maybe tell people not to use “steelmanning” in such cases, maybe tell people to stop expecting “steelmanning” to necessarily mean it won’t miss a central argument. Maybe decide that you should, e.g., say “I’m steelmanning this particular argument”, because if you say “I’m steelmanning the case for this conclusion” then that means you’re supposed to capture all important arguments for that conclusion.
Steelmanning is not the same as passing the ideological Turing test.
ITT is successful when your opponent agrees with you, or when your opponent cannot distinguish you from their actual allies.
Steelman is successful if you, or your allies, find something useful in the ideas of your opponent. Whether your opponent approved of the result, or not.
In ITT, your opponent is the judge (of how much it passes). With a steelman, you are the judge (of whether you have extracted something useful). When steelmanning, you cherry-pick the good parts, and discard the rest. When passing an ITT, you need to pass all checks.
For example, an ITT of a religion is… speaking like a true believer. A steelman of religion is e.g. saying that we don’t really know how the universe came to existence, and that there are some social benefits of religion.
Steelmanning is writing retcon fanfiction of your interlocutor’s arguments. As such it necessarily adds, omits, or changes elements of the source material, in ways that the other person need not accept as a valid statement of their views.