Hacker News Reader

Comment by nerdponx | Hacker News Reader

This is a good excerpt for those who don't have time to read this:

The Center for Applied Rationality promotes double-cruxing, a specific technique that helps people operationalize arguments. A double-crux is a single subquestion where both sides admit that if they were wrong about the subquestion, they would change their mind. For example, if Alice (gun control opponent) would support gun control if she knew it lowered crime, and Bob (gun control supporter) would oppose gun control if he knew it would make crime worse – then the only thing they have to talk about is crime. They can ignore whether guns are important for resisting tyranny. They can ignore the role of mass shootings. They can ignore whether the NRA spokesman made an offensive comment one time. They just have to focus on crime – and that’s the sort of thing which at least in principle is tractable to studies and statistics and scientific consensus.

View on HN

There's a nice general lesson from double cruxing that you can apply outside of one on one arguements if you care about holding consistent beliefs.

Always argue for your weakest crux. I.e. out of all the necessary premises of some position you hold, always keep the hardest to defend premise close to the front of your mind.

This is counterintuitive because people reflexively try to bury their weakest assumptions. What's more: we're rewarded for presenting strong arguments. But if there's some fact that, if disproven, would cause you to change your mind and you aren't really confident that that fact is true - you should care the most about finding out if it's false, or at the very least becoming more certain in your belief about it.

This is a very good point - as long as your goal is really and only to have the opinion that is closest to the truth, and you're completely open to changing it when new evidence comes. In other words, if you are really discussing because of the right reasons (which should of course always be the case).

But in many contexts, this is not the end goal. The end goal is to either make one feel better about its own beliefs, or to exert social pressure/gain social status.

You can often kind of tell who is really discussing to find the underlying truth based on this very observation.

I think it's useful to differentiate between "arguments" and "debates". An "argument" is a meeting of the minds, grounded in mutual respect, in which each side has an opposing a priori belief that they might be willing to change. A "debate" is a zero-sum game in which each person arguing is trying to change the mind of some third party^1. People often call interactions "arguments" when, in fact, they are actually "debates".

Doubele-cruxing can be useful in arguments, but the audience of a debate should approach any crux with incredulity. Anyone with half a mind toward "winning" a double-cruxed debate would choose an untenable position as their "crux". A practiced debater will find the scientific concensus and then work backwards to a crux. Not the other way around. Then they get the rhetorical benefits of acting "reasonable" without actually giving up anything useful to their opponent. In fact, if their opponent is clueless enough to accept the crux, they've just won the game. And if their opponent complains, well now the opponent is "meta-debating" which is apparently bad thing (and is a bad thing anyways because it makes you sound like an egghead).

So, double-cruxing is a great strategy for an argument between people who mutually respect one another. But audiences to a debate should be careful about whether they're being hood-winked by a rhetorical trick.

In short, a lot of this philosophizing about how to have a good argument really just boils down to "have the right argument with the right person". Two people who respect each other talking about something that both have fluid opinions on rarely devolves into a useless conversation.

^1: A debate in which there's no third party present might be called "a waste of time".

I dislike double cruxing along the lines of debating as a sport too, but I disagree that what you describe in the second half of your comment is a problem, in fact people should be encouraged to do what you describe.

The "double" part of double crux means that the crux you choose has to be a crux for both people, i.e. if it's established either way, that alone has to be sufficient to convince both people of the overall argument being true or false.

If you pick a crux that is a truly provably untenable position, and it turns out to also be a crux for the other person then they should happily change their mind on seeing the proof. If they don't it's not like they're committed to saying you've "won" against their will or anything, at worst it turns out that the crux they agreed on wasn't really a crux for them after all. At best, they discover that their belief rested on an untenable position.

By design it's pretty hard for a malicious actor to play double crux to win without also playing it to discover the truth, so long as they actually choose honest cruxes for themselves (and to some extent, even if they don't).

I'm unconvinced.

The rhetorical trick is to paint the opponent as unreasonable, pretentious, impractical, etc. if they don't agree with your double-crux. In this case, the resolution of the double-crux is not the strong argument (your opponent would never willinging agree to the double-crux)! The rhetorical characterization of your opponent is the strong argument. Again, it's just something you should watch out for as an audience member.

Also, inexperienced debaters will often agree to double-cruxes when they shouldn't because they don't understand the strategy that's being deployed agains them and the double-crux is presented in a "friendly" tone. And once it's deployed they realie that was a stupid thing to agree to. The truth we've discovered is that one of the debaters is inexperienced; we haven't learned anything about the underlying topic.

In practice, debates (acdaemic or out in the real world) are almost never searches for truth.

Again, as an audience member in an obviously zero-sum debate, I'm extremely skeptical of the double-crux. It's usually a steup for a rhetorical argument or a way of easily beating down an inexperienced person. It's rarely a geniune assertion that this one thing really is the fundamental point of contention.

In a friendly, respectful argument designed to search for truth, double-cruxes can be a great tool.

> Again, as an audience member in an obviously zero-sum debate, I'm extremely skeptical of the double-crux.

For some context, double-cruxing came out of the rationalist community (specifically CFAR), which commonly refers to persuasive rhetoric as the dark arts. They do not develop tools for use in the dark arts.

> So, double-cruxing is a great strategy for an argument between people who mutually respect one another.

That's a good point. The guide on double-cruxing[1] lists "epistemic humility" and "good faith" as pre-requisites for using the technique effectively.

[1] http://rationality.org/resources/updates/2016/double-crux

But the arguments on both sides would be somewhat crippled by removing the use of stats relating to gun control vs. crime.

Your quote may as well replace "gun control" with "religion" and "crime" with "morality". The arguments on both sides become superficial as you've removed the specific causes for each side to argue. You may as well continue this trend until both sides are arguing "good equals good" for differing values of "good".

This stuff isn't easily arguable by humans with opinions. Removing the example data that forms the opinions doesn't make the arguments on either side stronger or weaker, it makes the argument as a whole less relevant to society.

This is unfortunate because, as you quote, "that’s the sort of thing which at least in principle is tractable to studies and statistics and scientific consensus"

Breaking complex issues down to be "tractable to studies" is possible, but most techniques are fraught with errors of over-simplification.

> But the arguments on both sides would be somewhat crippled by removing the use of stats relating to gun control vs. crime.

In this case, the double crux was on crime prevention, so after agreeing on the crux they'd be only looking at gun control vs. crime statistics. Did you mean to say something else?

I think parent's point is that by searching for a double-crux you might obfuscate high-level generators of disagreement, so the conversation might progress but won't actually change anyone's mind even if there is a purely factual answer to the double-crus.

I think parent's point is made later in the paragraph after the section on double-cruxing:

> High-level generators of disagreement are what remains when everyone understands exactly what’s being argued, and agrees on what all the evidence says, but have vague and hard-to-define reasons for disagreeing anyway. In retrospect, these are probably why the disagreement arose in the first place, with a lot of the more specific points being downstream of them and kind of made-up justifications. These are almost impossible to resolve even in principle.

I.e., many people who have strong opinions on gun control would never agree to this sort of double-crux. And even if they did agree to it in the context of one conversation, and then "lost" the argument on that basis, they wouldn't actually go on to change their mind. They would return to the "High-level generator" and everyone just wasted their time.

The way I understand it, double-cruxing is a technique to be used (probably multiple times) during the debate, and isn't the entire debate itself. It kind of reminds me of "divide and conquer" strategies.

Reply on news.ycombinator.com