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.
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.
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".
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.