3 ms·
An argument in favour of steelmanning is that it is efficient. Using Paul Graham's hierarchy of disagreement[1][2], one reason steelmanning is efficient is bec
by dredmorbius 5y ago
An argument in favour of steelmanning is that it is efficient.
Using Paul Graham's hierarchy of disagreement[1][2], one reason steelmanning is efficient is because if your counterparty's best or most central argument is wrong then all of it is wrong.
It's also highly likely that numerous arguments hinge on that same point. After all, creationists continue to pull out Pangloss's "noses are for eyeglasses" argument despite the original being proposed as a parody in the first place.... Having a refutation of such a point can be useful in countering future bogosity.
Your counterparty may not participate. I'd attempted to establish any credible basis for a key point in a recent discussion, to which the response was an ever-escallating set of rationales for not answering the question. I wasn't even interested specifically in winning the debate, only in whether or not there was a foundation for the claim presented. Apparently there isn't.
It's also worth considering that reason and debate are not an especially useful method for arriving at truth. This has been a challenge since the days of the philosphers and sophists. Ultimately, it was Francis Bacon who provided an alternative: scientific method based on empirical observation and hypothesis testing. Here again the focus is on key propositions or empirical behaviours from which other consequences follow.
If you're engaging with someone who acts in bad faith, or who is simply incapable of grasping, accepting, or admitting truths, steelmanning won't do much to convince them. It may be of use when you're carrying on that discussion before others, however.
________________________________
Notes:
1. http://www.paulgraham.com/disagree.html http://www.paulgraham.com/disagree.html
2. Diagram: https://issuepedia.org/File:Graham's_Hierarchy_of_Disagreement.svg https://issuepedia.org/File:Graham's_Hierarchy_of_Disagreeme... [3]
3. Issuepedia itself is a useful resource for exploring didactic discussion and methods. Graham's Hierarchy is referenced on the "Hierarchy of Truth" article: https://issuepedia.org/Hierarchy_of_truth https://issuepedia.org/Hierarchy_of_truth RationalWiki is another generally useful site, with a broader coverage and authorship: https://rationalwiki.org https://rationalwiki.org
- notahacker 5y agoBut often the counterparty's best or most central argument isn't wrong, or is a core belief which is particularly hard to undermine. Easier to defeat the detail of a tangential argument of the form "noses are for eyeglasses" than contest the creationist's core claim that it's fundamentally reasonable to believe God exists (a belief shared by many who have been persuaded the detail of creationism and especially young earth creationism is silly misunderstanding). And it's certainly more efficient to point out areas where Qanon is an incoherent mess with a long history of failed prophecy than to attempt to convince people that people in power can be trusted and definitely never have any relations with sex traffickers. If the strongest form of an argument can be defeated with such effortless efficiency it's a waste of time delving into the weaker parts of the argument, it's probably not a very strongly held belief in the first place. Indeed people attempting "steelmanning" of more complex beliefs like the Scott Alexander invermectin example often concede the semi-strong bit to gain trust for the rest of what they have to say.
- dredmorbius 5y agoThere are a few takes on this: - If someone's just bullshitting you straight up, and you can find multiple instances of this by examining the more straightforward parts of their statements, then you can choose this route. This also often applies where there is no there there --- where an "argument" is simply an endless firehose of muck, as with the Qanon case. - If the core is solid, and your goal is gaining knowledge rather than Winning Imaginary Internet Points (or playing at more substantive rhetorical games, of which there are in fact many) ... then you've won by learning something. - If the core proves difficult to untangle or validate --- pointing this out is a valid criticism as you're putting additional valid burden of proof on the party arguing that case. Without substantiating their points, the argument is based on confusion, uncertainty, or ignorance rather than empirical merits. - In many ways, empiricism is a Steelman argument of the form "evidence trumps theory". If there's some theoretical argument and you can show that the evidence clearly contradicts it ... I'd argue that's a form of steelmanning. Of course in reality such evidence may be tremendously hard to obtain, or not clearly decisive, or require extensive interpretation to judge. - There are arguments which come down to statements of values. As was pointed out on the Ezra Klein podcast earlier this year, it's really hard to definitively argue for the superiority of one set of value judgements over another (though not necessarily impossible). You can point out that this is the nature of the disagreement, however. In the sense of steelmanning, this would be a case of pointing out begging the question in its correct sense --- if you presume that your argument's conclusion is a superior value, then you cannot invoke that assumption to prove the conclusion as the argument is the conclusion. I'd had a recent discussion of tenant vs. landlord rights on HN where this seemed to be the case. You can state that property rights are important, but if you're using the assumption to prove that property rights are superior to tenancy rights ... you don't have a logically sound argument.
- Jiro 5y agoIf creationists argue based on a belief in God, yes, it's a core belief that's hard to undermine. But a lot of creationists argue in contexts relating to teaching creationism in schools. Youre not allowed to teach religion as truth in schools, so the creationists use flimsy scientific arguments, not religious ones.
- adhesive_wombat 5y ago> It's also worth considering that reason and debate are not an especially useful method for arriving at truth. And this is why all those "XYZ is good/bad: debate me" arguments that make up so much of politics, religion and philosophy are so, so tedious. They do nothing to advance anything or find anything new.
- dredmorbius 5y agoThey may be useful as a training ground. Otherwise, I largely agree. The more I've dug into the history of philosophy, particilarly rhetoric, logic, moral philosophy, and many of the debates from religious, enlightnement, and modern philosophers, the more I'm coming to recognise ancient tropes which keep re-emerging in these debates. I'm not sure if that's deliberate or simply because the patterns are so wired into both human psychology and cultural knowledge and behaviour. But the exploration itself tends to be rather more illuminating than the debates, most of which retread not only the same ground but the same very basic methods and arguments. It's a bit like the old saw about the difference between ten years of experience, and one year repeated ten times.