How to Disagree with “How to Disagree”
Graham’s Hierarchy of Disagreement
http://paulgraham.com/disagree.html
The above diagram is a visual representation of the article “How to Disagree” by Paul Graham. Noting how the internet has magnified our opportunities to disagree with each other publicly, notably “in comment threads, on forums, and in their own blog posts”, Graham creates a ‘Hierarchy of Disagreement’ with ‘name calling’ at the bottom rung, and the much rarer and more respectable ‘refuting the central point’ at the top.
Nothing below DH4 (Counterargument), we are told, counts as in any way ‘convincing’.
The hierarchy is intended as a tool for readers to evaluate what they read. Graham’s hope is that they will use it to see through intellectually dishonest arguments and to immunise themselves against mere eloquence. He also believes that it may help writers, since ‘most intellectual dishonesty is unintentional’ and ‘seeing his current position on the disagreement hierarchy may inspire him to try moving up to counterargument or refutation’. It is best understood, however, as an attempt to provide an overall civilising effect, with the ultimate goal of making people happier:
“ ...there is a lot more meanness down in DH1 than up in DH6. You don’t have to be mean when you have a real point to make. In fact, you don’t want to.”
A Twitter friend of mine, having read my piece The Fallacy Fallacy, asked what I made of this hierarchy, given that my key intellectual engagement with my readers in the above piece was to implore them not to be arseholes.
...The answer is that I have a Lot of Opinions. This is going to be a long post.
Below, I defend the lowest three categories as potentially legitimate contributions. In the process, I criticise the position that ‘refutation’ and especially ‘refuting the central point’ are intrinsically the highest and most praiseworthy levels of disagreement.
The overarching theme you’ll see is: commenters on written works can have epistemic projects of their own which are not reducible to proving the original author wrong. The criticisms they make will be connected to their own projects, and disproving the original author’s central point may not be necessary - or indeed even relevant - to those projects.
Let’s start with...
DH0. Name-calling.
“This is the lowest form of disagreement, and probably also the most common. We've all seen comments like this:
u r a fag!!!!!!!!!!
But it's important to realize that more articulate name-calling has just as little weight. A comment like
The author is a self-important dilettante.
is really nothing more than a pretentious version of "u r a fag."
It absolutely is something more, but that has nothing to do with how articulate it is.
Firstly, we should note that the proper objection to a homophobic slur is that it is a homophobic slur, not that it is accompanied by a poor vocabulary and presentation. It would still be worthless if it were written in iambic pentameter.
(Contrariwise, ‘inarticulate’ name-calling can serve a powerful rhetorical purpose. Harry G. Frankfurt rejected the dainty term ‘humbug’ in favour of the visceral ‘bullshit’ and ‘bullshitter’ because he considered that the very roughness of those terms better expresses our evaluation of certain behaviours which make someone’s testimony untrustworthy.)
Whilst Graham acknowledges that it isn’t the inarticulateness of the slur that makes it objectionable, he does so by categorising ‘name calling’ itself as objectionable, as the lowest form of disagreement, regardless of how articulate it is.
Certainly, many or most examples of name-calling are nothing more than meanness. But context is all, and I am unwilling to dismiss the whole category as such. ‘Name calling’ can simply be categorising people according to habitual behaviour. That behaviour may be epistemically relevant; for example: “self-important dilettante”.
‘Self-importance’ is a tendency to value one’s own opinion above that of others regardless of how justified this is in a given instance; whilst a “dilettante” is a non-expert in the field who is merely dabbling without putting the work in to understand the subject. Both of these are relevant to a speaker’s reliability as a source.
I will allow that if you are responding to an individual, and hoping to engage with that individual, then calling them names seems counterproductive.
In the ‘Fallacy Fallacy’, one of my objections to the practice of simply naming the fallacy in your opponent’s argument and calling it a win is “you’re being an arsehole”. However, I think my readers will stay engaged in this case because (a) it’s not really addressing them personally, but rather a behaviour that they may or may not actually engage in, and (b) it operates on the good-faith assumption that if they are engaging in this behaviour, they probably don’t recognise it as arsehole behaviour. In Aaron James’ terms, “someone can act like an asshole - in a particular situation... - without really, ultimately, being an asshole” (p. 8, Assholes, a Theory).
It’s possible that (b), the assumption of good faith, might suffice to keep the reader engaged even if you are addressing them personally. One might use name-calling as a sharp, strong way to draw attention to behaviour that an individual would find abhorrent in themselves if they noticed what they were doing. But you would have to focus on the behaviour itself: “you’re being an asshole by doing XYZ” rather than “you are an asshole”. Perhaps this focus shift is enough that Graham would no longer characterise it as ‘name-calling’ at all.
(Also, in order to be effective, it would probably have to be based on some existing trust and respect between the writer and commentator, which might be better preserved by a private conversation rather than a public one).
However, all of this is to ignore a significant shift between “u r a fag!!!!!!!!!!” and “The author is a self-important dilettante”: the first statement addresses the author, whilst the second does not.
Instead, the second commenter is addressing other readers, and issuing a warning about the author. Their goal is to draw attention to certain habitual behaviour on the part of the author which makes them untrustworthy in general. Now, according to Graham, the ultimate (legitimate) goal of critical comments is to refute the central point made in the original argument. But a genuine concern that “the author is a self-important dilettante” is actually compatible with the commentator agreeing with the central point of the original argument.
One potential scenario: perhaps the central point isn’t contentious to those in the know, but has been stripped of context and presented as the author’s original brilliance, making it less likely that lay readers will ever find the real context or understand it. The commenter’s project is therefore to address this behaviour in the author.
Or quite possibly the commenter has met this annoying habit in several different writers, and their project is to address this behaviour in general.
To urge the commenter in this scenario to raise their position in the hierarchy, to ‘inspire him to try moving up to counterargument or refutation’, is to ignore their actual concerns, which don’t exist anywhere on the hierarchy.
Rather than treating ‘refutation of the central point of the original argument’ as the One True Goal of Disagreement, it’s more appropriate to recognise what the commenter’s epistemic goal actually is and assess how well their approach helps them to achieve it.
In the scenario I’ve laid out, the appropriate critical assessment might be:
“If you want to warn readers about this issue properly, you’ll have to give more detailed objections. Could you point to missing citations, perhaps?”
Our commenter may retort, however, that before they can convince fellow readers of a problem, they must get their attention in the first place. The sort of name-calling that provides a quick, rough assessment of the situation serves this initial purpose well.
In this sense, name-calling is in fact the most convincing form of disagreement at their disposal.
DH1. Ad Hominem.
So far, I’ve argued that some versions of name-calling have worth because they are a succinct way of communicating some epistemically relevant flaw in the author to other readers who may not be aware of that flaw. In short, then, I said that they counted as ad hominem attacks.
But the Hierarchy of Disagreement gives an extremely low position to ad hominem attacks themselves:
DH1. Ad Hominem.
An ad hominem attack is not quite as weak as mere name-calling. It might actually carry some weight. For example, if a senator wrote an article saying senators' salaries should be increased, one could respond:
“Of course he would say that. He's a senator.”
This wouldn't refute the author's argument, but it may at least be relevant to the case. It's still a very weak form of disagreement, though. 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?
Saying that an author lacks the authority to write about a topic is a variant of ad hominem—and a particularly useless sort, because good ideas often come from outsiders. The question is whether the author is correct or not. If his lack of authority caused him to make mistakes, point those out. And if it didn't, it's not a problem.
I’ll start by happily admitting that ad hominems can be lazy attempts to discredit the author’s central point without putting in the necessary work. If that is the commenter’s goal - if they want to disprove the central point, and believe they have done enough to achieve this - then they are mistaken in this belief, and redirecting them to the actual requirements of D6 is appropriate.
However, when a commenter questions a writer’s epistemic character and credentials, we should be open to the possibility that the commenter’s goal is... to evaluate the writer’s epistemic character and credentials.
It’s true that the commenter is probably also concerned about the likelihood of the writer being incorrect in their claims. After all, the reason we care about someone’s epistemic character and credentials is that we want to be able to believe what they say.
However, this doesn’t mean that assessing someone’s character is a mere stepping stone to refuting their argument. To see this, think about the success criteria. If one's concern genuinely is to establish whether a writer is epistemically untrustworthy, it is not sufficient to refute this particular argument. Even the most reliable source will make genuine errors from time to time. The reverse also holds: showing that the author is correct in this instance is not conclusive evidence that they are epistemically trustworthy. It’s just one sample, and they might be frequently unreliable elsewhere.
I suspect that Graham’s rejection of ad hominems as a category is connected to a kind of individualism about knowledge. The hierarchy of disagreement carries an assumption that it is our job as individuals to assess the arguments personally, in terms of both structure and content. Graham encourages us to find the flaws in the argument ourselves; and if we cannot, then to say nothing.
But this kind of rugged intellectual individualism fails to engage with the reality of knowledge as a social phenomenon. As individual readers, we have limited resources of time and energy to do independent research, especially in areas outside our own specialisms (i.e., most areas.) It is normal and necessary to rely on other people for most of what we know. We do, however, have some degree of leeway as to which people we rely on. When we choose these people, our concern is not merely whether they are correct in a given instance, but whether they can be relied upon to be correct. Therefore our ability to assess arguers has, if anything, a greater impact on the health of our epistemic environment than our ability to assess arguments.
Yet Graham’s hierarchy actively discourages commenters from assessing arguers. In particular, he discredits:
Drawing attention to bias,
Drawing attention to a lack of authority.
I’ll look at each of these in turn.
Drawing Attention to Bias
It is my position that drawing attention to a bias that other readers may be unaware of can be a helpful contribution from a commentator even if they can say nothing else.
One useful thing it can do is simply to prompt readers to pay more critical attention to the argument than they might have done otherwise. For example, perhaps the backbone of the senator’s argument was the claim that a senator’s salary compares unfavourably with alternative career options for highly qualified candidates, and this lack of competitiveness was detrimental to the Senate. Suppose I pass over this claim on first reading with no more engagement than “huh, didn’t think about that before, seems reasonable enough”.
Then I read a comment along the lines of ‘Yeah, this was written by a senator, though!’
I might at that point think ‘Oh, I didn’t spot that. Hmm. Hang on, is that central argument sound? Might some cherry-picking of data or misleading use of language be at play here?’ I’d be snapped to attention.
Now, it is possible that this could lead me to re-read the article more carefully and spot some flaw in the central point of the original piece: so that the D1 commenter inspires D6 levels of disagreement in someone else. (How this process fits into the hierarchy is anybody’s guess.)
Suppose though that I am still unable to “say what it is” that’s wrong with the argument. Perhaps I lack the necessary contextual knowledge to spot cherry-picking whether it’s there or not. In that case, I might seek out additional sources that lack (this particular) bias to help me see if my suspicions hold water. Or, if that is not feasible or I decide it isn’t worth the time investment, then I might simply suspend judgment on the issue of senator’s salaries for the time being: not ideal, but I’d still be proportioning my beliefs to my available evidence in a more appropriate way.
I suspect Graham assumes that the comment ‘Of course he would say that. He's a senator’ simply aims at producing a knee-jerk response in the reader of believing the opposite of whatever the senator says. If that is the goal, then it reflects badly on the commenter; if it is the outcome, it reflects badly on the reader. But for a reader to categorise a comment as an ‘ad-hominem’ and feel like their intellectual work is done and they can safely dismiss it as ‘unconvincing’ is its own kind of knee-jerk response: namely, automatically refusing to take any ad-hominem seriously.
Taking an ad hominem comment seriously, as a reader, does not mean passively reversing the conclusion of the original argument, but allowing yourself to be ‘snapped to attention’ by the presence of a bias you were previously unaware of, and undergo the process of critical evaluation described above. This process is surely to the reader’s epistemic advantage even if the commenter themselves were just being intellectually lazy (and it is unjust to assume that they were).
Drawing Attention to a Lack of Authority
Graham regards commenters who object to an author’s lack of authority as guilty of a ‘particularly useless sort’ of ad hominem, because, he says, ‘good ideas often come from outsiders’.
In some situations, his objection holds. If experts in a field were to automatically refuse to take the ideas of ‘outsiders’ seriously, then they would lose the chance to gain valuable new perspectives through interdisciplinary exchange. This would be unfortunate. At a more everyday level, I concede it is not reasonable to demand that people only write about things that they are experts in. Amateurism is not a crime. We are allowed to have opinions on art or music or politics, say, without having a degree in any of those things, and we can write about them on our blogs if we want to. We may even have some really good ideas.
However, when a commenter draws attention to a writer’s lack of authority, it is likely because the writer has conspicuously avoided doing so themselves. Writing about something that you are not an expert on is not blameworthy in and of itself. However, presenting yourself as an expert in a field you know little about to an audience of people who aren’t in a position to tell the difference is irresponsible. For example: if your Ph.D. is in an area unrelated to the topic of your book, and you put ‘Ph.D.’ in big letters on the cover anyway, you know exactly what you’re doing. Calling someone out as epistemically untrustworthy for such behaviour seems appropriate.
To shrug off all issues of bias and lack of authority with:
‘If [their] lack of authority caused him to make mistakes, point those out. And if it didn’t, it’s not a problem’
is to place an unrealistic expectation on commenters and readers to go through and unpack arguments written by someone they already have a reason to distrust, on a subject they are not themselves experts in, before you will take their objections seriously.
In fact, it’s more than that: it is to force commenters to reshape the nature of their objections to the point where they are unrecognizable.
The supposed purpose of the hierarchy is that we learn to disagree well. But more specifically, its purpose is to restrict the behaviour of commenters in ways that cushion authors. I can’t exactly condemn this: it’s an understandable response to the worst of online behaviour. And yet... the way the hierarchy is framed, if a commenter has genuine concerns about the author’s epistemic character and credentials, then expressing them at all is intrinsically ‘low’ and unconvincing. Either they must fight on the territory laid out by the author, where the proper target is the central point of their argument - even if that isn’t the commenter’s central concern - or else they should stay silent.
When surely, if we are to have a hope of disagreeing well, we must first communicate the actual nature of our disagreement, whatever it might be.
DH2. Responding to Tone.
So far, we’ve seen that the Hierarchy of Disagreement is stringent in the position that attacking the author themselves is off limits. Whilst criticising the author’s tone is not technically an attack on the author, it nonetheless cannot count as ‘arguing well’, because it isn’t sufficiently focused on the author’s argument itself:
DH2. Responding to Tone.
The next level up we start to see responses to the writing, rather than the writer. The lowest form of these is to disagree with the author's tone. E.g.
I can't believe the author dismisses intelligent design in such a cavalier fashion.
Though better than attacking the author, this is still a weak form of disagreement. It matters much more whether the author is wrong or right than what his tone is. Especially since tone is so hard to judge. Someone who has a chip on their shoulder about some topic might be offended by a tone that to other readers seemed neutral.
So if the worst thing you can say about something is to criticize its tone, you're not saying much. Is the author flippant, but correct? Better that than grave and wrong. And if the author is incorrect somewhere, say where.
I’ll begin by responding to this line in particular:
“Is the author flippant, but correct? Better that than grave and wrong.”
This stands out to me as the reverse of the truth.
Flippancy is disrespectfulness, a lack of seriousness in one’s engagement with the subject at hand and in particular with opposing views. Grave, in this context, would be the opposite of that: engaging seriously with the subject at hand and with challenges to one’s position.
A flippant tone in writing is a red flag for a flippant approach to writing. An author could, of course, be flippant and correct. I am far less convinced that an author could be flippant and epistemically trustworthy. As I said in the previous section, assessing an arguer is not merely a stepping stone to assessing a particular argument, but a goal in itself, and part of the greater goal of shaping my epistemic environment into the healthiest one possible. Expanding my knowledge beyond my own experience and expertise requires trust. When deciding whether or not to trust an author, I don’t demand that they are flawless, that they never come to a false conclusion. I instead demand that they take the work of writing seriously and thus share a certain cognitive load with me such that I feel able to believe them without constantly having to assess whether they have bothered to put in the appropriate work this time.
It’s notable that Paul Graham chooses the example of ‘intelligent design’ as the thing that the author is guilty of being flippant about. Intelligent design isn’t evidence-based; presumably the author is being flippant about it because it isn’t evidence-based, this is probably all terribly sympathetic to his readership. In short, the choice of example draws us toward the idea that if the author isn’t taking the topic seriously, it’s because the topic doesn’t deserve to be taken seriously. But this is misleading.
I invite you to consider another example. Here is Ken Robinson, giving a talk on ‘changing educational paradigms’. Midway through, he starts using ADHD as a punchline to make his audience laugh, casually implying both that it is vastly over-diagnosed and that stimulant medication turns children into zombies, and my deepest objection isn’t that he was wrong (though he very much was) but that he was dismissive, flippant.
Someone who has a chip on their shoulder about some topic might be offended by a tone that to other readers seemed neutral.
I’m quite sure that the tone did ‘seem neutral’ to much of its audience. That is precisely the fucking problem.
I was recently diagnosed with ADHD. I’m also a school governor for SEND, which means that I’m one of the people responsible for ensuring that the school has suitable provision for children with ADHD. I know how important supportive teachers are for children with ADHD, because I have interviewed children with ADHD to ensure that they are receiving the support they need from their school. I strongly suspect that a teacher's ability to be suitably supportive of these children would be negatively impacted if they didn’t actually believe that the children had ADHD. Ken Robinson is speaking to an audience that includes teachers, academic educationalists who train teachers, and undergraduates studying education.
Wildly surprised shall you be to learn that the tone did not seem neutral to me.
Since the tone seemed neutral to Robinson’s laughing audience but not to me, Graham invites me to consider that it’s the ‘chip on my shoulder’, my readiness to anger, which explains the difference in reaction. The disparity, their calmness next to my anger, is proof that I’m ‘not saying much’ in criticising the tone.
Oh, damn right, I’m angry.
If I wanted to disagree with Robinson in accordance with Graham’s hierarchy, I could not even aspire to achieve D6, ‘refuting his central point’, since this wasn’t Robinson’s central point, but merely a destructive little tangent. I believe I could reach as high as DH5 and refute his claims about ADHD. However, Graham counts even DH5 as an act of ‘deliberate dishonesty’ in cases where someone ‘picks out minor points of an argument and refutes those’, since ‘unless the opposing argument actually depends on such things, the only purpose of correcting them is to discredit one’s opponent’. There is no room for the idea that ‘minor points’ in one person’s argument might amount to major concerns for someone else.
In any case, refuting Robinson’s claims about ADHD would not get to the heart of my actual objection. My objection is that he was flippant. I have no desire to squash and squeeze and reshape that objection so that it resembles another form of disagreement higher up the hierarchy. I care that he was wrong, but I care far more that he was flippant, that he made a joke of it, because I’m worried about what that flippancy does to his audience.
It’s one thing to come to accept that a trusted expert on education made some false claims about ADHD. Academics aren’t always right. It’s quite another thing to come to accept that said trusted expert made a joke out of ADHD when he didn’t have an excellent justification for doing so.
Especially if you laughed.
(And why wouldn’t you laugh? He wasn’t being offensive. The tone seemed neutral to you.)
Robinson merely being wrong would still have led to his audience having false beliefs, but Robinson being flippant and wrong could lead to his audience having false beliefs which are significantly harder to extricate themselves from.
So yes, as I’ve said, I’m angry. Graham conflates anger with meanness and suggests that moving our disagreements higher up the hierarchy will rid us of both. As I’ll discuss in the next post, I don’t think that simply taking refuge in DH5 and DH6 makes it likely that we’ll escape either anger or meanness, but also, I’m not ashamed of my anger, or afraid of it. It’s not some destructive force. It’s simply a demonstration that I care about the things I’m arguing for. How could it be otherwise?
In conclusion...
Graham’s Hierarchy of Disagreement addresses the legitimate concern that a lot of argument (online or otherwise) is intellectually dishonest and amounts to little more than a personal attack. However, in the process of presenting an idealised form of disagreement, it ignores or dismisses the following:
Knowledge is social - permanently.
We are reliant on our sources of information being reliable and trustworthy. This isn’t a temporary childlike state we grow out of into True Intellectual Independence. It is a permanent human state of affairs. Assessing an author as untrustworthy is therefore epistemically valuable in itself. Passing that assessment on to other people is also epistemically valuable in itself - and sometimes best achieved with crude language.
The author doesn’t delineate the territory.
Rather than pointing at the conclusion of the author’s argument and saying ‘look, here, proving this wrong is the only legitimate target’, it is worth acknowledging that the minor points and yes the tone of an argument can have a broader impact, and are therefore legitimate targets of criticism in and of themselves. Arguments do not exist in a vacuum, and points made casually and in passing might be of central importance to someone else.
Evidence of emotion isn’t evidence of poor reasoning.
People tend to argue about things because they care about them. This is not in itself evidence that they are arguing badly. What Graham calls ‘having a chip on your shoulder’ could otherwise be described as ‘having personal experience of, or investment in, that thing’. In other words: having knowledge to contribute.
So, my disagreement with ‘How to Disagree’ is this: acknowledging the true scope of legitimate disagreement is crucial if we are to argue well.
