Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Better Disagreement

67 Post author: lukeprog 24 October 2011 09:13PM

Honest disagreement is often a good sign of progress.

- Gandhi

 

Now that most communication is remote rather than face-to-face, people are comfortable disagreeing more often. How, then, can we disagree well? If the goal is intellectual progress, those who disagree should aim not for name-calling but for honest counterargument.

To be more specific, we might use a disagreement hierarchy. Below is the hierarchy proposed by Paul Graham (with DH7 added by Black Belt Bayesian).1

 

DH0: Name-Calling. The lowest form of disagreement, this ranges from "u r fag!!!" to "He’s just a troll" to "The author is a self-important dilettante."

DH1: Ad Hominem. An ad hominem ('against the man') argument won’t refute the original claim, but it might at least be relevant. If a senator says we should raise the salary of senators, you might reply: "Of course he’d say that; he’s a senator." That might be relevant, but it doesn’t refute the original claim: "If there’s something wrong with the senator’s argument, you should say what it is; and if there isn’t, what difference does it make that he’s a senator?"

DH2: Responding to Tone. At this level we actually respond to the writing rather than the writer, but we're responding to tone rather than substance. For example: "It’s terrible how flippantly the author dimisses theology."

DH3: Contradiction. Graham writes: "In this stage we finally get responses to what was said, rather than how or by whom. The lowest form of response to an argument is simply to state the opposing case, with little or no supporting evidence." For example: "It’s terrible how flippantly the author dismisses theology. Theology is a legitimate inquiry into truth."

DH4: Counterargument. Finally, a form of disagreement that might persuade! Counterargument is "contradiction plus reasoning and/or evidence." Still, counterargument is often directed at a minor point, or turns out to be an example of two people talking past each other, as in the parable about a tree falling in the forest.

DH5: Refutation. In refutation, you quote (or paraphrase) a precise claim or argument by the author and explain why the claim is false, or why the argument doesn’t work. With refutation, you're sure to engage exactly what the author said, and offer a direct counterargument with evidence and reason.

DH6: Refuting the Central Point. Graham writes: "The force of a refutation depends on what you refute. The most powerful form of disagreement is to refute someone’s central point." A refutation of the central point may look like this: "The author’s central point appears to be X. For example, he writes 'blah blah blah.' He also writes 'blah blah.' But this is wrong, because (1) argument one, (2) argument two, and (3) argument three."

DH7: Improve the Argument, then Refute Its Central Point. Black Belt Bayesian writes: "If you’re interested in being on the right side of disputes, you will refute your opponents' arguments. But if you're interested in producing truth, you will fix your opponents' arguments for them. To win, you must fight not only the creature you encounter; you [also] must fight the most horrible thing that can be constructed from its corpse."2 Also see: The Least Convenient Possible World.

 

Having names for biases and fallacies can help us notice and correct them, and having labels for different kinds of disagreement can help us zoom in on the parts of a disagreement that matter.

Let me illustrate by labeling excerpts from Alvin Plantinga's critical review of Richard Dawkins' The God Delusion.

DH1, Ad Hominem:

Dawkins is not a philosopher... [and] you might say that some of his forays into philosophy are at best sophomoric, but that would be unfair to sophomores...

DH2, Responding to Tone:

[In this book] the proportion of insult, ridicule, mockery, spleen, and vitriol is astounding. (Could it be that his mother, while carrying him, was frightened by an Anglican clergyman on the rampage?) If Dawkins ever gets tired of his day job, a promising future awaits him as a writer of political attack ads.

DH4, Counterargument:

What is Dawkins' reply [to the fine-tuning argument]? He appeals to 'the anthropic principle,' the thought that... "we could only be discussing the question in the kind of universe that was capable of producing us." ...But how does that so much as begin to explain why [our universe] is fine-tuned? One can't explain this by pointing out that we are indeed here — anymore than I can 'explain' the fact that God decided to create me (instead of passing me over in favor of someone else) by pointing out that if God had not thus decided, I wouldn't be here to raise that question.

DH6, Refuting the Central Point:

Chapter 3, 'Why There Almost Certainly is No God,' is the heart of the book... [Dawkins says] the existence of God is monumentally improbable... So why does he think theism is enormously improbable? The answer: if there were such a person as God, he would have to be enormously complex, and the more complex something is, the less probable it is: "However statistically improbable the entity you seek to explain by invoking a designer, the designer himself has got to be at least as improbable..."

...What can be said for this argument? Not much. First, is God complex? According to much classical theology... God is simple, and simple in a very strong sense... More remarkable, perhaps, is that according to Dawkins' own definition of complexity, God is not complex. According to [Dawkins] something is complex if it has parts that are "arranged in a way that is unlikely to have arisen by chance alone." But of course God is a spirit, not a material object at all, and hence has no parts. Therefore, given the definition of complexity Dawkins himself proposes, God is not complex.3

 

Of course, even a DH6 or DH7 disagreement can still be wrong. But at the very least, these labels can help us highlight the parts of a disagreement that matter for getting at the truth.

Also see: Causes of Disagreements.

 

 

 

__________________

1 This article is an update to my earlier post on CSA.

2 Sometimes the term "steel man" is used to refer to a position's or argument's improved form. A straw man is a misrepresentation of someone's position or argument that is easy to defeat: a "steel man" is an improvement of someone's position or argument that is harder to defeat than their originally stated position or argument.

3 For an example of DH7 in action, see Wielenberg (2009). Wielenberg, an atheist, tries to fix the deficiencies of Dawkins' central argument for atheism, and then shows that even this improved argument does not succeed.

Comments (84)

Comment author: xv15 25 October 2011 06:06:18AM *  18 points [-]

If the goal is intellectual progress, those who disagree should aim not for name-calling but for honest counterargument.

and

DH7: Improve the Argument, then Refute Its Central Point...if you're interested in producing truth, you will fix your opponents' arguments for them. To win, you must fight not only the creature you encounter; you [also] must fight the most horrible thing that can be constructed from its corpse."

I would add that the goal of intellectual progress sometimes extends beyond you-the-rationalist, to the (potentially less than rational) person you're arguing with. The goal is not just to "produce" the truth, or to recognize the truth with your own two eyes. The goal is to both locate the truth and convince the other person that it is in fact the truth.

Often, I find myself in the following scenario: Someone says, "X and Y, therefore Z!" And off the bat, I have a good idea of what they're thinking and where the logic goes bad. But in point of fact, they are being loose with semantics, and there exist definitions of X and Y consistent with their original (loose) statements which would imply Z. I could ask them clarifying questions and get them to pin down their position further...but alternatively, I am free to say, "Surely you don't mean this one thing [which they really do mean] because here's how that would go bad. Perhaps you really meant this other thing? Am I understanding you correctly?"

This makes it much easier for people to "back down" from their original position without losing face, because they are framed as not having ever committed to that position in the first place. The reality is that we often have a choice between nailing someone in place and offering them up as a sacrifice to the logic gods -- in which case we don't really win since the logic gods can't actually touch people who don't submit to their power -- or deliberately leaving them untethered, so that they will more willingly adjust to new evidence.

Here it's not so much that I'm constructing the best argument from the corpse of their fully formed argument and striking it down. It's more like encouraging the growth of an adolescent argument in a direction that does not require it to be struck down, in the process striking down the bad argument that the original argument would have grown into, and trying to ensure that my "opponent" doesn't end up getting slain along with the bad argument.

This would be out of place in the above post, but I thought it was worthy of a discussion on Better Disagreement. Because I used to think the way to win was to pin people into logical corners, but if you're goal is partly to convince people, and those people are like most people, then in my (limited) experience, this way works So. Much. Better.

Comment author: a_gramsci 31 October 2011 03:07:42AM 0 points [-]

I find this is the most constructive way to resolve a debate between two people (see: http://lesswrong.com/lw/881/the_pleasures_of_rationality/) But in long-running debates, or ones with heated debaters, this is much harder. Firstly, because many debates are long running precisely because this strategy cannot be applied to the,. The issue with heated debaters is that this requires an open mindset of looking for truth versus looking to prove yourself right, which I find lacking in many debates.

Comment author: kilobug 23 October 2011 09:00:19AM 12 points [-]

Interesting hierarchy, but it lacks one important point which is often fundamental in some debates, like the theism/atheism one, or even many political one : the difference between "belief" and "belief in belief". For many theists nowadays, the core of their belief in God is not really due to deep reasons that God has to exist - they believe that Adam and Eve, Noah's Ark and most of the Bible are philosophical tales, not that they really happened, they believe in evolution, when they are sick they go see a doctor, not to Lourdes... but they still believe in God, because if they feel that if they don't, life has no meaning, ethics will crumble, and the horror of death will become unbearable without the hope of afterlife.

When someone is thinking like that, no DH5, DH6 or DH7 level disagreeing on the existence of God will do much. But building a line of retreat could. While under your hierarchy, attacking those belief in belief, building a line of retreat, would be only a DH1, while it can (in some cases, obviously not in all of them) be the most efficient way to disagree : point to the core reason behind a disagreement.

Comment author: Louie 23 October 2011 12:12:14PM 8 points [-]

It's much the same concept, but I will point out that I prefer to build people new temples instead of helping them construct lines of retreat.

Comment author: STL 23 October 2011 08:39:48AM 10 points [-]

Honest disagreement is often a good sign of progress. - Gandhi

Source? I can't find any that look reliable.

Comment author: Grognor 23 October 2011 09:23:33AM *  9 points [-]

I like this post, even though it doesn't add much to Paul Graham's original essay. I mean, I wouldn't have seen this content if it were not posted on Less Wrong.

For all of my life until this year, I've been confounded by DH2 arguments against myself. Why are my opponents ignoring what I say because I said it angrily, or sadly, or confrontationally, or in passing, or whatever? Well, I don't like when people do that, but it doesn't change the fact that people do it, so I've started to adopt a more pleasant, acceptable tone.

I still don't like it. I don't like that I have to adopt a certain style to be taken seriously. But oh well.

Edit, 4/04/12:

I've started to adopt a more pleasant, acceptable tone.

I was deluded when I said this.

Comment author: lukeprog 23 October 2011 03:11:26PM *  12 points [-]

I like this post, even though it doesn't add much to Paul Graham's original essay.

I debated whether to bother writing it, but in the end I felt it was worth it to:

  • Compress the content explaining each level of heirarchy
  • Include DH7, which was at the time only available on Internet Archive
  • Give concrete examples from actual articles
Comment author: Vladimir_Nesov 25 October 2011 12:24:00AM *  7 points [-]

Include DH7, which was at the time only available on Internet Archive

On LW, this idea is well-known as The Least Convenient Possible World.

Comment author: lukeprog 25 October 2011 12:36:09AM 0 points [-]

Added, thanks.

Comment author: GilPanama 23 October 2011 08:33:35PM *  17 points [-]

Why are my opponents ignoring what I say because I said it angrily, or sadly, or confrontationally, or in passing, or whatever?

The way you say something may signal that you are trying to diminish their status. If you say it with a sufficiently negative tone, it may even be taken as a signal (a generally reliable signal) that you care more about diminishing their status than about having a truth-seeking discussion.

In other words, what wedrifid said, but less simply and more explicitly.

Comment author: wedrifid 23 October 2011 09:51:45PM -1 points [-]

Exactly!

Comment author: fubarobfusco 25 October 2011 12:11:39AM 9 points [-]

Sometimes, talking about tone is merely a poor rebuttal — a DH2 argument. Sometimes, it's a request for a more pleasant conversation: it's simply unpleasant to have a casual chat with someone who comes across as contemptuous, hopeless, or bigoted. Tone does exist, after all, and it is possible to be an unpleasant, hostile conversationalist; so sometimes when people talk about tone, they really mean it.

(For instance, it is insufferable to have a conversation about (say) race and IQ with someone who keeps using racial slurs in the conversation; or about the relative importance of different academic disciplines with someone who keeps referring to engineering students as "pencil-necked dorks" or liberal-arts students as "poem-fag hipsters". Obviously, their choice of tone does not prove anything about their actual arguments; but it does come across as hostile and unpleasant. People having a casual conversation cannot be expected to put up with arbitrarily high levels of unpleasantness.)

The tone argument becomes seriously toxic, though, when what the tone-arguer means by it is: your argument is wrong; therefore, gathering and presenting evidence for it, and making your argument clearly and boldly, are signs of a character flaw on your part.

For instance, consider the following exchange:

A: Gostaks are distimming doshes! We should stop them!
B: I don't believe you. What's the big deal?
A: Well, I've gathered all this evidence of a hundred different cases just in the last year where a gostak distimmed a dosh. Take this one case, for instance; the dosh in question was in a coma for six weeks and to this day faints at the sight of a distimming-widget.
B: Wow, you're really obsessed and hostile about this. I mean, what kind of person would go and gather "evidence" to accuse gostaks of supposedly distimming doshes? You must really hate all the doshes who get along just fine with gostaks. And if there was a problem, well, seriously, you'll get more flies with honey than you will with vinegar. I don't have to listen to any more of this anti-gostak ranting.

Comment author: wedrifid 23 October 2011 02:31:26PM 3 points [-]

Why are my opponents ignoring what I say because I said it angrily, or sadly, or confrontationally, or in passing, or whatever?

Because for most practical purposes how you say it is the important part.

Comment author: Grognor 25 October 2011 12:19:41AM 1 point [-]

It upsets me that people are replying to this comment as though it were a problem I still have.

Comment author: thomblake 25 October 2011 12:50:05AM 2 points [-]

Probably because of:

I still don't like it.

It sounds like you haven't really gotten over the issue.

Comment author: thomblake 24 October 2011 07:19:20PM *  0 points [-]

To put it another way: You are upset that they are misinterpreting what you say because of your tone. Instead, you could recognize that you are speaking their language incorrectly. That tone seems important to other speakers of the language and not to you is not a problem with their understanding of the language.

Comment author: DSimon 24 October 2011 01:01:03AM *  7 points [-]

I really like this article a lot, and want to adopt these terms into more general usage. An important step to accomplishing this is to stop referring to the items by number, and use their gloss descriptions. "False positive" is a lot easier to remember than "type I error" (or is it type II....)

Comment author: GilPanama 23 October 2011 08:22:55AM *  37 points [-]

DH7 should be kept internal, at least at first. Being misinterpreted as trying to construct a straw man when you've been trying to do the opposite can derail a conversation. To actually believe that you've made a steel man, not a straw man, the person you're arguing with would have to admit that you've created a stronger argument for their own position than they could.

It's probably best to practice up to DH7 internally, and only up to DH6 vocally.

If we imagine arguments as soldiers, as they tend to be, the problem becomes even clearer:

(A and B are about to fight.)

A. Ah! My worthy opponent! I shall send my greatest soldier to crush you... GOLIATH! ATTACK!

B. His sword's a little wimpy. Let me give him a bazooka.

If I were A, I wouldn't trust that bazooka on B's word alone, I'd be annoyed at the slight against my blacksmiths, and, even if it turned out to be a totally legitimate bazooka, I would, at the very least, consider B a tactless grandstander.

(Though if the bazooka did work, I'd use it, obviously. I just wouldn't like using it.)

Comment author: nazgulnarsil 25 October 2011 03:46:29AM *  17 points [-]

You can be gentle about DH7 by attributing the improved argument to someone with high status. This is my typical strategy and seems to work well. It's a double whammy because you're implicitly associating them with someone of high status e.g. "it's funny you say that, it's very similar to an argument by <famous person>". I'm NOT saying that you actually have to know a bunch of famous arguments offhand, the better argument can be attributed fallaciously to anyone who has spoken on a topic and can have little to do with the person's original argument. Few notice and you have the out of being mistaken even if they do.

Comment author: JoshuaZ 25 October 2011 03:51:45AM 18 points [-]

That is a fascinating border case between Dark Arts and rational discussion.

Comment author: Jack 25 October 2011 02:38:12AM 16 points [-]

The way this is done in (good) academic philosophy is "6 then 7". First you show that their central point fails for reason x. Then you suggest how their position can be improved upon then you refute the new position.

Comment author: orthonormal 23 October 2011 05:15:09PM 12 points [-]

DH7 does happen between mathematicians now and then. Person A has an idea of a proof for X. Person B could show a problem with Person A's proof (DH6) or an unrelated disproof of X (DH4? DH6?), but the best response is to show A a disproof of X that makes it clear why A's strategy is futile.

This is often done well enough that it doesn't even hurt feelings. But math is kind of a special case.

Comment author: Will_Sawin 23 October 2011 09:11:07PM 7 points [-]

In particular, in math it is clear which arguments are more dubious. DH4 arguments are often perfectly acceptable, as a simple and clear counterexample refutes a complicated argument that could easily have a subtle flaw.

The ability to make simple, irrefutable arguments is tremendously beneficial to sane arguing, for instance because it enables you to use the The Emperor Has No Clothes defense and avoid studying the details of your opponent's argument.

Comment author: GilPanama 23 October 2011 07:08:34PM *  3 points [-]

In this case, I'd even drop my initial thoughts about rudeness. If you can prove that somebody's gone down mathematical blind alley, it's downright polite to do so, since there's no ambiguity about the relevance of the steel man here.

Comment author: fiddlemath 23 October 2011 09:05:30AM 5 points [-]

I'm pretty sure it depends on who you're arguing with. If either of you is trying to /win/, rather than /find the truth/, then DH7 is tough to do. But if you and your interlocutor both care more about being correct than sounding correct, and you both respect each other, then you can and should attempt DH7 aloud.

Comment author: GilPanama 23 October 2011 09:36:38AM *  5 points [-]

I can respect the person I'm arguing with, and consider them to be truth-searching, and still not want to antagonize the part of their hardware that likes winning. I also dislike having my primate hardware antagonized unnecessarily; I tolerate it for the sake of truth-seeking, but it's not fun.

I see two likely cases here:

A) I come up with a tougher version of their argument in my head, in order to be as careful as possible, but I still have a good way to refute it. This is DH7.

In this case, announcing the tougher version doesn't get us any closer to the truth. A dead steel man is as dead as a dead straw man. I might as well refute what was actually said, rather than risk being unnecessarily smug.

B) I come up with a tougher version of their argument in my head, and I can't actually defeat the tougher version.

In this case, I definitely ought to announce this problem.

But this is not DH7 as posted. This is my actual purpose in making a steel man - the possibility that the steel man may actually force me to change my mind. I'm not trying to argue with my opponent on a higher level when I do this, I'm trying to argue myself out of being cognitively lazy.

A good rule of thumb: DH7 should be really really REALLY hard to do well if you're arguing with reasonably smart people who have thought carefully about their positions. In fact, it is so hard that anybody who could do it consistently would never need other people to argue with.

EDIT: In the interests of dealing with the worst possible construct, I should add:

A) In the case where openly announcing DH7-level arguments lets both parties see that they've misinterpreted each other, going to DH7 is a net win.

B) An expert DH7-level arguer may still need other people to argue with if they have been exposed to very different sets of evidence.

But generally speaking, the cognitive effort needed to communicate a steel-man version of someone else's position is better spent on expressing one's own evidence.

Comment author: gjm 23 October 2011 09:47:06AM 5 points [-]

If you come up with a better version of the other person's argument but keep it to yourself and only refute the original version, then later on they may think "Now, in all honesty Gil was right about X ... but no, wait a moment, that's just because I didn't get it quite right. If I'd said X' instead then his argument wouldn't have worked." and stick with their position rather than changing it.

I doubt that this outweighs the effect of antagonizing them at the time by saying "You should have said X', and I'm now going to refute that" in most cases, though.

Comment author: GilPanama 23 October 2011 10:13:48AM 1 point [-]

Ideally, a reasonable counterargument that applies to the strong form will also apply to the weak form without significant editing. If the person one was arguing with would have been receptive to DH7 in the first place, that alone should stop them from making the strong form argument - the countering evidence has already been provided.

Where this fails... well, I said "at first" in my thread-starter for a reason.

Comment author: Will_Sawin 23 October 2011 09:14:58PM 2 points [-]

Some DH7, or at least DH7-like thinking, can be relatively easy. For instance, there will often be gaps in someone's argument that they do not consider significant, or a general case they hadn't bothered to think of. You can't make it perfect, but you can patch it up a bit.

Comment author: GilPanama 23 October 2011 11:24:18PM 0 points [-]

Point taken - in some cases, the significance of the gaps is more evident to the outside view.

Comment author: HughRistik 24 October 2011 11:00:55PM *  2 points [-]

DH7 should be kept internal, at least at first. Being misinterpreted as trying to construct a straw man when you've been trying to do the opposite can derail a conversation. To actually believe that you've made a steel man, not a straw man, the person you're arguing with would have to admit that you've created a stronger argument for their own position than they could.

This.

DH7 is of limited use in an adversarial debate, unless your opponent is open-minded. It could convince fence-sitters, but only if they are open-minded.

The problem with DH7 is that it's too easy for your opponent to accuse you of a straw man. Even if that's not true, they may be able to delude some of the audience.

Analogies are another debate tactic in this category: they are only useful towards listeners with an open-mind, otherwise, they make you open to attack be the other person rejecting your analogy.

A great time to use DH7 or analogies is against the argument of someone who isn't present to convince a third-party. Since your opponent isn't there, they can't reject your attempts at charity or analogies as straw men, and you can use those tools to convince your audience that you are correct, and you've given those arguments the best consideration you can.

Of course, if you're going to do this, try to make sure you are right, because if you are wrong (e.g. you misunderstood what your original opponent was saying), then they won't be around to clarify.

EDIT: Actually, there is a way to do DH7 with your original interlocutor. You have to lead them to admitting that the steel version actually follows from their argument, and then you knock it down. E.g. you start by "are you suggesting Y?" which you think follows from their original position, X. This can make you look like you are genuinely working to understand them (which, of course, you are). Then when they take the bait, you knock it down, and they can't complain.

And you can't be too confrontational or accusatory, because that will tip them off that you are going to knock Y down. If they catch a hint of that, then they will never admit that Y follows from their original position X.

Comment author: dlthomas 24 October 2011 11:35:40PM 1 point [-]

You need to be sure that your rebuttal applies both to the argument they have presented and to the steel man argument you have constructed (which you can spell out or not, depending on context), and ideally to any men of straw or steel others are likely to construct for themselves on hearing your opponent's argument.

Comment author: thomblake 24 October 2011 11:16:47PM 0 points [-]

I think it bears repeating that this matters if you're trying to win an adversarial debate, but not so much if you're trying to learn the truth of the matter.

Comment author: TheOtherDave 23 October 2011 06:27:12PM 2 points [-]

Depends on how it's done, IME.

I often find that "Hm. So you're saying XYZ? That doesn't really work, because of ABC. But now that I think about it, X'Y'Z' would be consistent with what you're claiming, and not have that problem. Even there, though, A'B'C' suggests it's false." can work all right, although I'm often tempted to add "But of course by this point I've wandered off into a corner and started arguing with myself, which seems antisocial."

Comment author: GilPanama 23 October 2011 08:25:39PM 3 points [-]

I think that in some contexts, like arguing over mathematical proofs (as orthonormal noted), spending a little time arguing with yourself to bring out X'Y'Z' is polite and a sign of good faith. In other cases, I'd rather just trot out A'B'C' early on, as long as it doesn't require too much effort, and deal with both arguments at once without ever explicitly raising X'Y'Z'.

Comment author: MarkusRamikin 23 October 2011 09:14:37AM *  0 points [-]

The danger you point out is real, but an unqualified dictum such as "DH7 should be kept internal, at least at first" is very specific advice that IMO is going too far. A great deal depends on the quality of the sub-argument to do with strengthening the opponent's position, and the opponent's (and/or audience's, if any) receptiveness to that. You seem to be saying we should always have low confidence in both, while I'd say it depends.

If we imagine arguments as soldiers

Maybe let's not. Applying DH7 already assumes we're interested in truth, not just in winning a debate.

Comment author: GilPanama 23 October 2011 10:04:17AM *  4 points [-]

I don't ALWAYS have low confidence in the other arguer's ability to tolerate a steel man version of their own argument. I do have low confidence in the ability of most people, especially me, to decide what constitutes a non-gratuitous steel man. I have an unfortunate, but understandable, bias in favor of my own creations, and I suspect that this bias is widely shared.

Comment author: Johnicholas 23 October 2011 03:30:01PM 13 points [-]

There's a valuable difference between two different kinds of counterarguments (that I encountered in Drescher's Good and Real, but I presume it's widely known):

To correct someone, it is not sufficient to offer an argument for the opposite conclusion. At that point you merely have an apparent paradox - two arguments for opposite conclusions. You must also point out where in their argument they went wrong.

If everyone followed this policy, it could break certain circular or repetitious disputes which are trapped in a cycle of: A offers argument for X, B offers argument for not-X, A repeats their argument more clearly or more loudly, B repeats their argument in turn, and so on.

Comment author: Will_Sawin 23 October 2011 09:05:23PM 5 points [-]

Doing this is often surprisingly difficult. There's something about analytical frameworks, where to me, A feels like a flaw in the argument to B, whereas to you, B feels like a flaw in the argument to A.

I think one reason for this is that decisions are made in terms of cost-benefit analysis, which tends to make that distinction fuzzier.

Comment author: lessdazed 23 October 2011 03:40:16PM 4 points [-]

Maybe they didn't go wrong, and there is no paradox, and it is a matter of relative values as to which is more important.

In which case screaming makes a lot of sense as a strategy.

Comment author: GilPanama 23 October 2011 08:59:19PM *  7 points [-]

In that case, we can replace "point out where in their argument they went wrong" with "point out where our underlying value judgments seem to diverge."

If they then try to argue that your values are wrong and theirs are right, either you have to move the discussion up a meta-level or, yes, screaming.

Comment author: drnickbone 13 February 2012 05:31:57PM *  3 points [-]

The reference to Wielenberg's argument as an example of DH7 is an interesting one. This anlysis might be a bit off-topic, but the poster mentioned it so here goes.

Wielenberg is saying a couple of things in the article:

  1. Dawkins' central argument won't work if God is conceived as a necessary being, rather than a contingent being.
  2. Also, it won't work if God is conceived of (or just defined as) a simple being rather than a complex being.

Now on point 1, to be fair to Dawkins, he probably has not met real theists who have wanted to seriously defend (in an open debate, with a straight face) the idea that God is a necessary being. This is because the claim is rhetorically very vulnerable to a couple of simple counter-arguments:

"What? If you're going to insist that God exists necessarily then you're just being dogmatic, and shutting off all reasonable discussion on the subject. I could just as easily claim that an anti-God (such as the most evil being possible) exists necessarily. Or that the universe exists necessarily without being created. So now we're at a stalemate. To get past the stalemate, we need to concede the possibility that we each are wrong. I'm quite prepared to do that. Will you?"

Or this:

"What? There is no such thing as a necessary being. It seems quite possible to me, and to anyone listening with an open mind that there could have been nothing in existence at all. We all understand the question 'Why is there something rather than nothing?' and we are puzzled by it precisely because we understand that there could have been nothing. But, in that case there cannot be any being which exists necessarily. So if you're going to define God as existing necessarily, then you've just defined a contradiction in terms. When writing my book, I was assuming that you'd have the good sense to define God as something possible, rather than something contradictory."

Notice that either of these should convince anyone who's not already been sucked into the Plantinga-inspired black hole of modal theism. In that black hole, you just deny as possible various inconvenient states of affairs which any normal human being would judge to be possible (including the overwhelming majority of actual theists). And, better still, you don't have to give any real proof that the denied state of affairs really is impossible (in the same way as proving that there is no largest prime number). That would be too hard. Instead you just have to assert it, and claim that you are within your epistemic rights to assert it.

As I said, this black hole is not rhetorically very effective... And it is certainly not rational as LW would understand the term.

On point 2, Dawkins does actually address this simplicity reply in The God Delusion: he is just mystified by it. The reason is that he is really convinced of his proposition that any designer of a complex being is itself a complex being.. So in his mind, defining God to be simple (when he has already shown it to be complex) is again just defining a contradictory being.

Now I think a legitimate criticism of Dawkins is that he doesn't provide a detailed argument for the principle that a designer of a complex being must itself be complex. So to that extent his argument has a hole, though there are a number of approaches to patch the hole. If Wielenberg were really trying DH7, he would have tried such patches.

For example, we could collate inductive evidence from lots of designers we know about like humans, chimpanzees and CAD tools (they are all complex) or from lots of simple things we know about like quarks, electrons and atoms (they are not designers). Or we could try to use formal complexity definitions (such as logical depth, which seems the closest analogue of physical and biological complexity as Dawkins describes it).

Notice that If the hole does get closed, or looks like getting closed, the theist is in trouble. And then it won't do to present a "simple" God as a counter example to Dawkins' principle, since this will be like presenting "phlogiston" as a counter example to the principle that objects gain mass when burned. Or worse, presenting a "squircle" (defined as a square circle) as a counterexample to the principle that all squares have straight sides...

Again sorry if off-topic.

Comment author: Swimmer963 24 October 2011 08:33:21PM 3 points [-]

I came up with something like DH7 out of a rather irrational reason: that is, I hate disagreeing with people, and I used to especially hate criticizing people, but I also get (perhaps understandably) annoyed by people being wrong about things. So my typical strategy has been to point out the things that they are right about to them, and look for a version of their central argument which is right, whether because its claim is weaker and its area of interest is narrower, or because there was a logical flaw which was simple to fix.

I know there's at least one person who found my discussion style really irritating and started trying to make me argue 'against myself' and then point that fact out. But I don't get especially attached to most of the ideas or points I argue about, so I remember thinking it was kind of pointless that he did that.

Comment author: Robert55 23 October 2011 03:11:52PM *  3 points [-]

These are 7 useful rules of thumb but the "Art of Controversy" (http://en.wikipedia.org/wiki/The_Art_of_Being_Right) is always going to be slippery. The tract by Schopenhauer is useful to read to see how more subtle moves can be played out. It is recommended not as a handbook but to be better armed against the spectrum of ploys that an opponent may play.

Comment author: MarkusRamikin 23 October 2011 08:51:47AM *  6 points [-]

Good post.

I wonder if it'd be out of place for us on LW to start explicitely noticing these argument levels in disputes here. Though it'd probably be a better norm to use this positively (point out "good DH7 argument there") rather than negatively, which... come to think of it, would just be something of a DH2 argument in itself, harmful ammo for the sophisticated arguer.

Do I understand right, that the difference between DH4 and 5 is that DH4 doesn't directly engage the previous person's argument? And if only DH6 is "refuting the central point", then Dh5 means "nitpicking"? So it seems to me a solid DH4 could tower above DH5.

If a senator says we should raise the salary of senators, you might reply: "Of course he’d say that; he’s a senator." That might be relevant, but it doesn’t refute the original claim: "If there’s something wrong with the senator’s argument, you should say what it is; and if there isn’t, what difference does it make that he’s a senator?"

This is true but still, I'm inclined to think an ad hominem argument has some place in honest debate. If the real reason someone is advocating something is an unstated one, to do with their vested interests or psychological makeup or whatever else might invite an ad hominem argument, then if you defeat their stated argument, you'll get nowhere. They'll just come up with another one. Then another one, if that is defeated again. That's hacking at leaves rather than striking at a root. Exposing the real reason might help with actually getting somewhere.

Then again, little is more annoying when someone, instead of engaging your argument, starts judging your psychological makeup...

So what's optimal to do here? I'm not sure. Perhaps hacking off a few leaves first creates an opening for an attack at the root. At this point you've given the other person the courtesy of engaging their stated arguments for a while, and gathered some evidence in the process: "look, I defeated your last three arguments, and you ignore that and keep coming up with new ones, clearly we're dancing around the real issue here".

Comment author: orthonormal 23 October 2011 05:17:51PM 12 points [-]

Ad hominem attacks do constitute some Bayesian evidence. However, human beings tend to dramatically and unconsciously overweight them (because our powers of reasoning are optimized for navigating social battles rather than seeking abstract truths), so the proper rationalist norm among humans should be to eschew them.

Comment author: thomblake 24 October 2011 07:13:33PM 2 points [-]

Indeed; this can probably be a stronger claim: Even when attempting to discount evidence like ad hominem, we will overweight them.

Comment author: Johnicholas 23 October 2011 03:48:10PM 7 points [-]

The first few levels of the hierarchy, DH0, DH1, DH2, DH3, are not useless, if you're concerned with getting to the truth efficiently. Pointing out that the author has no credentials in the area they are speaking in would allow people to limit their attention to credentialed speakers (due to bounded rationality). The senator example is similar - if you can't listen to everything and have to prune someone, choosing to forgo listening to obviously biased speakers might be a reasonable choice.

DH2 and DH3 are sometimes valuable as well, because we can do some inference about the world, taking a piece of communication as evidence. For example, we may infer something about the author's personality or beliefs from their tone; and then (bounded rationality again) decide whether to read further based on this (limited, probabilistic) estimate of the author.

Contradiction works similarly; it's a performance, putting the reputation of the speaker behind the negated claim. If a third party hears a college freshman make a point, and the professor flatly contradicts it, then the third party may be able to decide who to believe, shortcutting detailed analysis (scarce processing again). In some cases, this kind of contradiction argument can even be supported solely on tone, without leaning on reputation. A scholarly-sounding contradiction to a casual claim may be sufficient to sway me in some situations - perhaps the stakes are low.

Comment author: potato 24 October 2011 10:41:11PM 2 points [-]

If you’re interested in being on the right side of disputes, you will refute your opponents' arguments. But if you're interested in producing truth, you will fix your opponents' arguments for them. To win, you must fight not only the creature you encounter; you [also] must fight the most horrible thing that can be constructed from its corpse.

This is amazing! I will def incorporate this into my daily arguments.

Comment author: J_Taylor 24 October 2011 08:08:44PM 4 points [-]

This is not a statement of disagreement, but merely a stylistic concern. I understand that prefacing one's work with a quotation from a respected figure is often good form. However, if it does not interfere with our purposes too much, could we not make a habit of it at Lesswrong?

Comment author: roryokane 02 November 2011 03:26:14AM 0 points [-]

I personally don’t really care whether posts have quotations at the top or not, but could you clarify why you don’t like those quotations? Is it something to do with appeal to authority being a fallacy?

Comment author: J_Taylor 07 November 2011 06:04:25AM 0 points [-]

I did not mentally categorize the Gandhi quote as being an appeal to authority. Rather, I categorized it as a decoration which contains irrelevant, and potentially false, information.

However, in all honesty, I cannot appeal to any higher principle for why I dislike essays being preceded by, let us say, generic quotations from certain figures. My dislike is probably merely due to aesthetics.

Comment author: GilPanama 23 October 2011 08:52:50PM *  3 points [-]

An application of this hierarchy:

Jack the Scarecrow. My crystal healing pills will give you eternal life. For $50.00 each, you need never die, suckers.

--

DH0: "I'm not interested for myself, but can I buy you a border collie and give her some? If you're going to live forever, you're going to need a smart friend to make the really tricky decisions."

DH1: What, exactly, is your profit margin on these crystal healing pills? If we don't live forever, would you still make money off of them?

DH2: Any post that ends in the word "suckers" directed at the readers is difficult to read charitably.

DH3: > My crystal healing pills will give you eternal life.
WRONG.

According to this hierarchy, D3 is arguing on a higher level than DH0, DH1, and DH2. And, well, maybe that's correct. "Higher" doesn't necessarily mean "more subtle." But in the absence of this hierarchy, I'd have been tempted to order the arguments from low- to high-level as 0 < 3 < 2 < 1.

(EDITED to remove potential applause light; EDITED again to better hit the idea of DH2)

Comment author: Xom 25 October 2011 01:44:04AM 5 points [-]

The DH tier seems to correspond to an upper-bound on the effectiveness of an argument of that tier.

Comment author: kpreid 23 October 2011 09:54:45PM 0 points [-]

Your example makes me wonder where "Their argument is not even wrong." (i.e. semantically incoherent) fits. It seems to me to be in the vicinity of DH2 and DH3, but not exactly either one.

Comment author: GilPanama 23 October 2011 11:40:34PM *  3 points [-]

I'd say that the incoherent speaker is arguing at DH(-1). DH0 would be an improvement. You would be counterarguing at DH(No) - argument by pointing out conversational emptiness.

(edited to clarify that it is the person who makes the incoherent argument who is arguing badly, and the person arguing against that who is doing something entirely outside the hierarchy.

Other DH(No) arguments-that-are-about-non-argument include "We aren't actually arguing about the same thing" and "let me take some time to do more reading before I reply.")

Comment author: byrnema 23 October 2011 03:22:03PM *  2 points [-]

I really like DH7s.

When I'm reading through LW posts, they're pretty much the only arguments I'll update on. Otherwise, I'll reserve the possibility that the initial argument wasn't argued well. Especially when a position is just dismissed as 'silly', I'll assume that the inferential distance was too large between that person's point of view and ours. For example, here.

Comment author: lessdazed 23 October 2011 09:22:30AM 2 points [-]

those who disagree should aim not for name-calling but for honest counterargument.

I agree that people can use the scale to sort among their arguments and filter out the lower ones, thus raising the level of discourse. But that is not the primary way I use the list. (I think I saw it on your site originally.)

People already subconsciously have a decent handle on what you explicated. What needs the most help is consciously understanding how good the various forms of argument are.

I say that people subconsciously understand these things because when pressed to rationalize, justify, or explain a position, they tend to give the available argument highest on the scale. This is counterbalanced by unique advantages name calling has, such as taking fewer words and lowering the status of the opposing arguers directly. But at least from D2 to D6, people basically get it.

That people are poor at comparing lists of evidence, one for and one against a proposition, does not mean that within each list, they are poor at sorting arguments from best to worst (although such a ranking will include already factored in for each sub-argument how likely it is to be wrong). The scale is most useful for testing the conclusions of entities by seeing where on the scale their best arguments fall, and how many such they have, in addition to checking the quality of them (i.e. how well they prove the sub-conclusions they are claimed to prove). "Entities" includes other people and one's inner voice.

Comment author: mingularity 16 September 2013 05:41:52PM *  1 point [-]

I note that a DH7 argument requires suspension of disbelief.

I also propose a DH8 argument, which is a collaborative version of DH7: Work together to improve both arguments, identify the strengths and weaknesses of each, and if one cannot be collaboratively chosen as the winner, identify what experiment can be conducted or what information can be obtained to move forward on the issue.

Comment author: Kevin 27 October 2011 10:44:52PM 1 point [-]
Comment author: kwackarole 03 December 2011 06:13:54AM *  2 points [-]

Original Hacker News discussion is here: http://news.ycombinator.com/item?id=149052

This contains I think the first reference to DH7, by 'tungstenfurnace'. He attributes the practice of strengthening an opponent's case before refuting the central argument to Karl Popper.

This practice is noted in Bryan Magee's Popper (Fontana, 1973) :

http://books.google.co.uk/books?id=6sw1GGBUYjkC&pg=PA91&lpg=PA91

What Popper aims to do, and at his best does do, is to seek out and attack an opponent’s case at its strongest. Indeed, before attacking it he tries to strengthen it [..]

Comment author: shminux 24 October 2011 04:29:11PM *  1 point [-]

Even DH7 presumes that the argument is wrong to begin with, which is not overly rational.

How about:

DH8, Clarifying When the Central Point is Indeed Valid. E.g. "A model of a supernatural entity as an ancestral simulator can be derived from the Simulation Argument framework. The validity of this framework and its approach to the question of Origin is now examined..."

or even

DH9, Update Your Model Based on Opposing Views. E.g. Given the <opposing argument>, which appears to be valid provided the <following conditions> hold, I have updated my priors to account for the Universe as described by <opponent>. The next order of business is to jointly examine our priors and come up with a more reliable model.

Comment author: JoshuaZ 24 October 2011 05:01:29PM 6 points [-]

One problem with your proposed DH8 and DH9 is that sometimes they just aren't possible. Sometimes people are just wrong and no update is necessary. The rest of the hierarchy is always possible regardless of the strength of the argument. DH8/9 not so much.

Comment author: shminux 24 October 2011 05:15:17PM 4 points [-]

As I mentioned, unless your opponent is stupid or trolling, in which case any level of engagement is probably wrong, there is a high chance that there is something to their claims, limited in applicability though they might end up being, so DH8/9 (and maybe higher?) are worth at least considering.

Comment author: pedanterrific 24 October 2011 04:38:27PM *  4 points [-]

("DH" stands for disagreement hierarchy.)

Comment author: JoshuaZ 24 October 2011 04:41:20PM 2 points [-]

That doesn't seem to diminish shminux's point. The highest level of disagreement should be when one no longer has a disagreement. That should be the goal.

Comment author: pedanterrific 24 October 2011 04:46:21PM *  1 point [-]

I understand the basic idea, AAT and all that. I'm just saying that if I set out to describe and rank the ways in which people express disagreement, "I agree" wouldn't be on the list.

Edit: That is, it's not that DH7 assumes the argument is wrong, just that you disagree with it. As long as you disagree, it's generally "better" - less logically rude - to use higher levels of the hierarchy than lower. If you find that you can't, then it might be time to update towards your opponent.

Comment author: JoshuaZ 24 October 2011 04:48:49PM 5 points [-]

So maybe rename it a dialogue hierarchy?

Comment author: Nornagest 24 October 2011 04:44:10PM 2 points [-]

I see what you're saying: opposing arguments should not be parsed from a presumption of falsity, which amounts to writing the bottom line and working backwards from there. Which is quite true as far as it goes. But it seems to me that your proposal contains an implicit bias in the opposite direction: by placing situations where the opposing argument is valid above all disagreements but on the same scale, you're privileging concordance relative to dispute.

Now, I suppose you could argue for doing that consciously, on the grounds that people generally find it difficult or embarrassing to update based on their opponents' positions and need all the help they can get. But doing so would strictly be a psychological hack, so I don't think we should be arguing for it on rationality grounds.

Comment author: dlthomas 24 October 2011 04:48:49PM *  4 points [-]

I thought psychological hacks leading to better practical rationality was a big part of what this site is about.

Comment author: Nornagest 24 October 2011 04:55:23PM *  2 points [-]

It is. But hacking your mind by applying some countervailing bias is a workaround, not an actual fix, and referring to that workaround in isolation with terms like "overtly rational" risks overcorrection or misapplication: if you convince yourself that pushing for a DH8 or DH9 solution is generally rational, there are situations where that can actively mislead you.

Comment author: shminux 24 October 2011 04:52:47PM 2 points [-]

Unless your opponent has proven to be inferior and stupid time and again, it is reasonable to assume that their proposal has some merits. This may turn out to be false, in which case it ends at DH6 or DH7, but, more often than not, a disagreement between two smart and people can be traced to their priors (e.g. Talmud is the ultimate source of wisdom vs Experimental evidence is the final judge), and those are worth arguing over, not the specific argument, which tends to be many levels removed.

Comment author: pragmatist 26 October 2011 07:28:37PM *  1 point [-]

I would switch the order of DH1 and DH2. A tone argument is very rarely relevant to the substantive dispute. In most cases, the tone of an article shouldn't lead you to update your belief in the conclusion. An ad hominem argument, on the other hand, is often substantively relevant, especially given the power of motivated reasoning. It is entirely reasonable to lower your credence in the conclusion of an article arguing that senators are underpaid once you discover that the author of an article is a senator. Of course, if you have already evaluated the argument itself, and are fairly confident in your evaluation, then learning the identity of the author shouldn't significantly impact your belief in the conclusion (kind of like argument screens off authority), but that is true of tone arguments as well.

The justfication for placing tone above ad hominem in the hierarchy is that the former at least responds to the writing, not the writer. But surely this isn't adequate justification. One might respond to the writing in many ways that are entirely irrelevant to the disagreement, e.g. by reproducing the written piece in reverse order. The question should be, which of these responses is more often relevant to a proper assessment of the truth of the conclusion.

Comment author: Nornagest 26 October 2011 08:17:21PM *  4 points [-]

Like a well-motivated ad hominem attack pretty much boils down to an accusation of motivated cognition, a well-motivated attack on tone usually amounts to an accusation of bad faith. While that shouldn't affect your conclusion given a constant set of evidence, it can certainly change your or your audience's weighting of evidence presented and shape the way you approach it rhetorically: if you suspect your opponents might be trying to score political points rather than to present a coherent argument, it behooves you to be more careful in parsing their arguments for dog-whistle phrases or known lines of rhetorical attack. If you suspect them of being deliberately inflammatory to provoke an emotional reaction, that's a good cue to disengage, and so forth.

I don't know how I'd weight this relative to ad hominem, but I don't think I'd call it substantially less relevant in casual debate.

Comment author: TheOtherDave 26 October 2011 07:49:51PM 2 points [-]

The question should be, which of these responses is more often relevant to a proper assessment of the truth of the conclusion.

If my primary goal is to come to a proper assessment of someone's conclusion, that's certainly true.

Of course, in that case, this entire list only covers a vanishingly small fraction of the domain of discourse. The vast majority of my responses when that's my goal don't involve interacting with the speaker at all... I listen to what they've said, I assess it, end of process.

The nature of this list suggests to me that we're not talking about proper assessment... rather, it suggests that the assessment has already taken place, and now we're talking about successful expression of that assessment to an audience. That is, it suggests that the OP is talking about rhetoric, not analysis. In which case proper assessment of truth may no longer be the important question.

Comment author: pragmatist 26 October 2011 07:54:32PM 1 point [-]

OK, that helps. Perhaps this should be made explicit in the post, Luke?

Comment author: dlthomas 26 October 2011 07:41:58PM 1 point [-]

Updating beliefs regarding the conclusion of an argument based on discussion of tone is a poor idea, for sure. Discussion of tone is tangential, and doesn't do much to get you where you're going at the moment. I believe it is placed above ad hominem in the hierarchy because it is more likely to be adding something, however - updating beliefs about tone based on discussion of tone makes total sense, and may lead to better discussion in the future.

Comment author: TrE 25 October 2011 05:05:48AM 1 point [-]

I'm not certain whether DH-central-point (DH6) is so much different from DH-improve-then-disagree (DH7).

If DH5 is nitpicking (I'll call it DH-nitpicking), then DH-central-point is not. So DH-central-point would mean attacking what the author really wanted to say. But then, what's DH-improve-then-disagree? It can't really be filling the holes in the arguments of the author because then we'd arrive at DH-central-point, wouldn't we? Are there examples that clarify the proposed distinction between DH-central-point and DH-improve-then-refute?

Comment author: yew 27 October 2011 10:26:46PM *  0 points [-]

DH6 means attacked the central point of what the author actually said. Just filling in holes in their argument would still fall under that, I think. DH7 means substantially improving the overall structure of their argument (introducing totally new supporting reasoning, for example) and then attacking the result.

In the ideal case DH7 is about attacking the version of an argument that would have been presented by a superintelligence trying to convince you of something, since that's presumably the version of the argument that best relates to the truth.

Comment author: lessdazed 27 October 2011 10:58:19PM 1 point [-]

the version of an argument that would have been presented by a superintelligence trying to convince you of something, since that's presumably the version of the argument that best relates to the truth.

Not always, not even too often.

Comment author: yew 27 October 2011 11:33:28PM 0 points [-]

Perhaps amend that to "an honest superintelligence." If you still disagree with that I'd like to hear your reasoning.

Comment author: TrE 28 October 2011 05:11:19AM 0 points [-]

The person who applies DH7 is not a superintelligence? Just a rough guess.

Comment author: yew 28 October 2011 05:50:30PM 0 points [-]

Hence "in the ideal case." I'm guessing that they're referring to the idea that someone very much smarter than you could convince you of anything by simply lying very convincingly.