Support my work on Patreon. Join for free.

Graham’s Hierarchy of Disagreement: How to Argue Like an Expert

Sigmund Freud famously wrote that “the person who first flung a word of abuse at their enemy instead of an arrow was the founder of civilisation.” The “word”, he noted, became the “substitute for the deed”. Since then, humanity has perfected the art of civilised disagreement. I kid of course. The arrows we fling nowadays tend to be digital. Spotting, deflecting or catching them requires a refined skill set. Thankfully, in 2008 author Paul Graham attempted to put contemporary disagreement into perspective by classifying its most common forms. Graham’s Hierarchy of Disagreement was born. Let’s explore how to argue like a true expert and what forms of disagreements we should rather avoid.

Table of Contents

What Is Graham’s Hierarchy of Disagreement?

Graham’s Hierarchy of Disagreement is an informal ranking of seven common strategies of dissent; from low-quality forms of disagreement to the highest-quality ones. It was developed by Paul Graham in his essay How to Disagree. Paul’s reasoning for writing the piece was pretty straightforward.

The web made the writer-reader relationship more interactive, he thought. As a result, disagreement by way of the written word in the form of online comments surged. While ideas being challenged was a wonderful thing, we might want to develop a nuanced understanding of what quality disagreement looks like.

Graham’s Hierarchy of Disagreement

Let’s revisit his hierarchy, expanding its scope to the spoken word. We’ll look at each of the seven stages individually and reflect on them holistically in the end. I’ll also attempt to show how we might handle each of the ways to disagree. Let our climb to the pinnacle of cultivated conflict begin.

DH0. Name-calling

The lowest form of disagreement, and probably also the most common.

As Freud correctly observed, abusive language is a more civilised way of disagreement than violence. Still sounds terrible, but let’s be honest. We’re all doing it. This is why I probably won’t have to give an example. Name-calling is a quick and easy way to dismiss everything about an argument in one go: What was said, how it was said and most importantly that ignorant blockhead who said it. Today, it even comes in the form of gifs and memes.

Graham's Hierarchy of Disagreement
Graham’s Hierarchy of Disagreement

Despite its common nature, name-calling can still take us by surprise. Especially when it’s wrapped in sophisticated language. As Graham points out, “The author is a self-important dilettante,” falls into the name-calling category, too. Yet it says nothing about the validity of someone’s claim. Technically, I can be a self-important dilettante and still be correct.

Regardless, I think there’s some merit to name-calling — apart from it not being an arrow. If used in a tongue-in-cheek manner or for humorous effect it can be a legitimate way to make a point. With one important caveat: It should be obvious to the audience that you know how to argue and how to behave yourself. If you’re Richard Dawkins, you can get away with indirectly telling Neil deGrasse Tyson to “f*** off” – and make everyone laugh in the process.

When reacting to name-calling, I think it’s worth seeing it as a form of disagreement, albeit an emotionally loaded one. The breaking of conventions may signal frustration or helplessness. There’s really no point in launching into a long-winded rebuttal that one is in fact not ‘an asshat’. Labelling the negative is a better strategy: “Sounds like you’re really passionate about this topic,” could elicit substance you can actually work with.

DH1. Ad hominem

Attacks the characteristics or authority of the writer without addressing the substance of the argument.

An ad hominem feels like it could also belong in category one of Graham’s Hierarchy of Disagreement. But it’s a bit more sophisticated attack on your credibility. As Mill’s Trident suggests, when it comes to the argument itself, it can only be wrong, partially correct or 100% accurate. But surely, if we’re deemed not competent enough or morally reprehensible, it means our argument can be conveniently ignored. It’s quite reminiscent of the legal strategy to discredit the integrity and reliability of a witness to dismiss his or her testimony.

Although this form of disagreement feels insincere, I think Graham is right in not entirely dismissing it. His example is that of a senator being in favour of increasing the salaries of parliamentarians. “Of course, he would say that. He’s a senator,” is a response akin to a mental shortcut that has a point. But how exactly? Let’s untangle this a bit further.

It looks like the missing links between a speaker’s characteristics or authority and the substance of a claim are bias and competence. The senator is clearly biased when it comes to his own salary, so it has some bearing on the validity of his argument. He may have fallen into an intuitive trap rendering his claim unsound. A similar attack on one’s authority denies the writer or speaker the competence to put together a valid point. Oddly enough, the proof of competence can be found in the argument itself, which is conveniently ignored.

Today, there seem to be ever more metrics by which characteristics and authority are judged: Group identity and presumed political affiliation. What someone looks like. What they don’t look like. With whom they happened to share a stage in the past. The number of likes and shares, who they follow, or who follows them. In short, any conveniently (un)favourable label. When it comes to assessing if a claim makes any sense, those metrics don’t get us too far. The same goes for the number of Twitter followers. But of course, I would say that. I only have a few.

If we were to steelman the ad hominem attack, we could see it as a comment on the likelihood of a claim having substance. That is how much weight it should be given. It would be impractical to divorce an argument from the speaker or writer entirely. (Just imagine a claim about rocket science being tweeted by Elon Musk or yours truly.) Even though disagreements cannot be solved by comparing characteristics or weighing up credentials alone, they have a place when used in good faith.


Really, if a writer’s expertise and standing are so questionable, it should be a breeze to debunk their arguments. So what can you do about it? Perhaps just lean back and internalise an aphorism by Nassim Nicholas Taleb: “You never win an argument until they attack your person.”

DH2. Responding to Tone

Criticises the tone of the writing without addressing the substance of the argument.

We’re making our way further up the pyramid. The air is getting a bit thinner and we’re now talking about the content of what was being written or said. Kind of. Because at this level, disagreement is only about how an argument was conveyed.

There’s a dimension of tone to our earlier name-calling example, the one involving Dawkins and deGrasse Tyson. The astrophysicist criticises the evolutionary biologist for his rather uncompromising promotion of atheism and science. Dawkins responds with a quote that he attributes to an unnamed former editor of the New Scientist magazine:

Science is interesting, and if you don’t agree you can f*** off.

At level DH2. of Graham’s Hierarchy of Disagreement, we would ignore Dawkins’ underlying point entirely and spend all our energy on rebuking the man for the harsh, authoritarian and combative tone of his words. Of course, this was all said in good humour. Though, it’s an extreme example of a phenomenon with which we’re probably all too familiar.

Graham's Hierarchy of Disagreement

Whether we like it or not, tone invariably plays a role in both written and spoken language. Is it friendly and curious or hostile and condescending, personal or distanced? Is the tone implied towards the audience or the subject? Tone carries meaning and can change a sentence considerably. Just think of irony or sarcasm.

Paul is correct in saying that tone is hard to judge and eventually in the ear of the beholder. Whatever the perceived meaning of an argument is, it’s usually more productive to prioritise the substance than to engage in a meta-argument about tone. With the exception of asking for clarification, I’d say.

The antidote to this form of disagreement is similar to the previous ones. Bad-faith actors can always find ways to respond to tone. If they never circle back to your argument, it may have just been a distraction for their lack of quality disagreement. So label the negative and probe for substance.

DH3. Contradiction

States the opposing case with little or no supporting evidence.

I’d be committing a crime if I didn’t reference Monty Python’s Argument Clinic at this point.[1] Not sure what Paul Graham was thinking. As early as 1976, John Cleese and Michael Palin opined:

Palin: Argument is an intellectual process. Contradiction is just the automatic gainsaying of anything the other person says.

Cleese: No it isn’t.

Yes, it is. Though it ranks somewhat in the middle of Graham’s Hierarchy of Disagreement. It’s the first level that addresses the actual substance of a claim. Well, poorly. If we go with the SExI method, an argument consists of a Statement, an Explanation and an Illustration, that is supporting evidence. Contradiction is nothing more than a low-resolution counter-claim that rejects an argument in its entirety. However, as Graham points out, it is not without merit.

It may be a good opener to establish your opposition to a statement you consider outlandish or obviously (un)true. In order to address the substance of an argument, there also has to be substance to begin with. If there’s none, Christopher Hitchen’s dictum may apply: “What can be asserted without evidence can be dismissed without evidence.”

In a best-case scenario, contradiction is enough for the other party to concede your point. In a slightly less good scenario, you prompt your counterpart to provide more supporting evidence. That’ll give you something to work with. Alternatively, you’ll be trapped in an “automatic gainsaying” forever.

On a more serious note, the antidote to mere contradiction is the classic ‘why’? This is annoyingly illustrated by the Why Game. The Why Game is an underrated debating activity that will lead you to appreciate the Münchhausen Trilemma, a thought experiment about the impossibility of proving the truth. If you prefer to be less confrontational, a “What caused you to think that?” will do. Even less confrontational, mirroring what the other person said with a curious and questioning tone followed by a pause should prompt them to expand on their idea.

DH4. Counterargument

Contradicts and then backs it up with reasoning and/or supporting evidence.

We’ve now reached the “first form of convincing disagreement” according to Graham’s Hierarchy of Disagreement. Hooray! Unfortunately, a counterargument tends only to give us the illusion of constructive objection. No doubt, it’s an improvement to level 3. But there’s the danger of missing the point of the original argument and ending up strawmanning each other.

Worst case, the counterargument doesn’t directly relate to the original argument. “Peanut butter is healthy because it tastes fantastic,” can be countered with “No, because my cat is allergic to it. Here’s its health record,” but that doesn’t get to the core of the matter. The same is true for the provision of a link as a counterargument. It’s too unspecific and too unrelated to the original claim to deserve a higher rank.

Counterarguments work when everyone’s paying attention to what was being said or written. When both sides are willing to do the legwork themselves by reflecting on how a counterargument might disprove their point of view. It also has a place in debate when you want to pivot to a whole new line of argument.

Don’t get me wrong. I love to focus on self-righteously hammering my own points home as much as the next person. It’s just not quality disagreement. If you realise you’re arguing past each other and you don’t want to merely agree to disagree, start by levelling up to DH5.

DH5. Refutation

Finds the mistake and then explains why it’s mistaken using quotes.

As Graham points out, refutation is the highest form of disagreement. That’s because it’s difficult, which is why it’s scarce. To pull it off, you have to master the art of dissecting an argument and establishing why parts of it might be flawed.

It helps to have a conceptualisation of what an argument consists of. If we go with the SExI method again, you’d be analysing if the statement has flaws, whether the explanation of it is lacking or if supporting evidence does not hold up. “Peanut butter is healthy because it tastes fantastic. This has been shown in a 2018 study,” gives us plenty of aspects to refute. The alleged correlation between health and taste for example. Or the sample size and methodology of the study.

The prerequisite for refutation, however, is your ability to pay close attention to what’s being written or said for that matter. Paul is right (again) in saying that whatever you quote has to match with what you’re refuting. Otherwise, you run the risk of strawmanning your counterpart, which essentially sets you back to level 4 or lower.

I’d add a related pitfall. Be careful not to lose yourself in pointing out faulty logic and reasoning for its own sake. This kind of nitpickery rarely advances a debate or gets anyone closer to the truth. It can also feel more like a clandestine ad hominem or name-calling exercise. So don’t get carried away if you like this kind of stuff (note to self).

Unlike most of the earlier levels, being confronted with a refutation can be a welcoming experience. Done right, it truly tests your ideas and encourages you to think through and refine your position. That being said, refuting parts of an argument still doesn’t get you all the way to the top of the pyramid.

DH6. Refuting the Central Point

Explicitly refutes the central point.

Here we are at the top of Graham’s Hierarchy of Disagreement. We’re pretty much alone up here. We’re still in the realm of refutation, but instead of refuting aspects we’re identifying and demolishing the central point of an argument. Similar to DH5., this requires skills that can be practised with more advanced debate games: Pay attention to what was being said, repeat it back to the speaker until he or she agrees with your characterisation of the argument and then refute the central point.

The highest form of disagreement comes with its own difficulties. It’s hard to refute a central point when the other side doesn’t have one. Also, refuting a single argument may cause you to win the battle, but it’s usually only one piece of a set of claims to establish a proposition. Though, if you’re facing a formidable opponent who knows how to disagree, you have a guaranteed constructive intellectual sparring session at your hands.

In fact, this level of disagreement is so vital in solving problems and discovering the truth that it’s often generated somewhat artificially if it doesn’t come naturally. The prime example being forms of institutionalised devil’s advocacy such as the Tenth Man Rule.

Finally, what can you do if your carefully crafted central point is taken apart? Perhaps just concede that you were wrong, be grateful for learning something new and get out of your head. Take up a new hobby, maybe. I hear archery is quite good for improving your patience, focus and confidence.

The Legacy of Graham’s Hierarchy of Disagreement

How to Disagree was intended to help writers and readers have better disagreements. By showing them how to “see through” and defeat “intellectually dishonest arguments”, Graham hoped to make interactions less “mean” and people “happier”. “Well, that didn’t work out, did it?” a cynic might say in 2022. If anything we managed to diversify and perfect the ugly art of dishonest disagreement at almost every level.

On a positive note, I’d like to think Graham was underselling his hierarchy. The timeless power of refutations is undisputed and should obviously be pursued as the highest good in written and spoken disagreement. The problem is that it’s very rare to solely argue at this level. No level is without merit. If used in good faith, all objections have the right to exist, especially in dialogue.

Closing Thoughts

The art of intellectual sparring lies in your ability to use the entire range of Graham’s Hierarchy of Disagreement. In order to understand your counterpart’s arguments and craft your own. Because disagreement is about more than reasoning and logic. It’s also about people’s emotions, pace, timing, wit and collaboration. It addresses the substance of an argument while taking into account how something was said, the person who said it and the reasoning behind it.

You should be able to adapt to your counterpart and the dynamic of an exchange. The teasing out of an argument, its probing from all angles requires you to be well-versed at every level – even though you might never use them all. What you’re aiming at is getting you both to DH6. Where you understand each other’s central points as well as how and why you see the world the way you both do. Who says you can’t have a bit of banter along the way?


37