Arguing without warning

Arguing without warning

Mar 2022

I think tipping is bad, but I still tip when expected, because… what’s the point?

Probably most social norms are like that—even if we disagree with them, they’re too entrenched to change. That’s why it’s important to watch for critical points where norms could go either way.

Here’s one such case: Say someone wrote something. You think they’re wrong, so you write a manifesto arguing that they’re wrong. Are you supposed to:

  1. Name and link to them?
  2. Not name or link to them?
  3. Let them know before you publish?
  4. Run it past them to catch misunderstandings?
  5. Give them a chance to write a rebuttal?

Are the rules different if you’re friendly with the person you disagree with?

Of course, in modern social media, the prevailing attitude is no quarter asked or given. If that’s what you’re used to, you probably think it’s adorable that I imagine any of this is open to debate. But in the genteel corridors of long-form internet writing, people really do seem to have varying expectations about this stuff. Just personally, I often wonder what rules I’m supposed to follow.

Here, I’ll suggest the following norms:

  • It’s usually good to name and acknowledge what you’re refuting.
  • There’s zero obligation to do anything else.

More controversially:

  • We should push back against people who imply they’re entitled to the other stuff.

Even more controversially:

  • We should maybe intentionally not do the other stuff, even when it would be easy, just to help establish that it’s not required.

Why? A few reasons.

We need more red teams

There’s just isn’t enough checking and counter-arguing, full stop.

Brian Caplan points out that no one cared about the spreadsheets he published to support his argument that education is about signaling rather than learning. This is true even though his book got lots of attention, and even though his conclusion is at odds with the popular consensus.

Why? Caplan suggests it’s because no one believes quantitative social science is meaningful. Maybe, but I suspect the explanation is simpler: There’s not much glory in checking someone else’s work.

Say you pored over Caplan’s spreadsheets and you found some technical error. Would that buy you some significant reward or attention? Unless you already have a high profile, probably no one would understand or listen to you.

The same pattern holds elsewhere. While many say peer-reviewed research is the gold standard of reliable truth, I think this trust is mostly delusional. Reviewers are anonymous and—aside from altruism—have little incentive to go digging for mistakes. They might be very eager to reject a paper, yes, but it’s easier to do that for superficial reasons rather than finding a mistake in the depths of some proof or experimental design.

The world needs more disagreement.

Or… OK, that’s not true—there’s disagreement everywhere but it’s superficial and unfair and sarcastic and mean. So how about: The world needs more high-quality disagreement.

Economically speaking, careful, fair, substantive critiques are undersupplied. We need to lower the barriers to making them, not raise them.

It doesn’t particularly matter what you think

One argument for running your manifesto past the person you disagree with is that maybe it’s all a big misunderstanding.

Maybe it is, but so what?

Say that somewhere in my brain I have a flawless argument that watermelon is bad. However, everyone who reads my Case Against Watermelon thinks I’m wrong. If they talk to me, I can clarify what I meant, and then they’ll often agree with me. But the writing alone never does it.

In this case, does it matter that I have a perfect argument in my brain? I don’t think so—most people will only see what I wrote.

The mistake here is seeing a counter-argument as being against a person, rather than as against a particular artifact. What matters is what’s out there.

We shouldn’t trust anything that’s not stress tested

Go find any popular article that makes factual claims. Then, read comments about it on some high-quality forum. Almost always, you will see comments that are deeply problematic for the original thesis. (“If watermelon is bad, why do 73% of children choose it over honeydew or cantaloupe?”)

How much should these comments decrease your confidence in the original article? I claim: Not much, at least on average.

Say that for most articles you read, you go through this process:

  1. You read it.
  2. You’re convinced that it’s true.
  3. You search for criticisms.
  4. They are devastating, so you no longer think the article is true.

That would be silly, for two reasons. First, if everything you read has devastating criticisms, you should expect them to exist. They should only decrease how much you trust something if the criticisms are stronger than average.

Second, the critiques themselves often have serious flaws, and then those have serious flaws, and it sort of goes on forever. I find this aspect of reality very frustrating, but most discussions seem to be an infinite regress carried on until someone gets exhausted and leaves.

So, if the critiques of an article find only “moderately bad flaws”, that often increases my trust because my prior was that the flaws would be even worse. If the claim that some types of watermelon taste bad to certain people at some times of the year survives the skeptics, then that might be an above-average outcome.

There’s value to having everything in public

Maybe when I publish my anti-watermelon manifesto you’re skeptical. So you contact me with your concerns, and I’m able to convince you that, yes, watermelon is bad, so you drop the issue.

The problem is that maybe some other people

  1. wouldn’t think of the concerns you had, but
  2. wouldn’t accept my counterargument against them.

If you had made your concerns in public, those people would now think I was wrong. But as it is, they’re left believing me.

There might also be people that

  1. would think of the concerns you had, and
  2. would be convinced by my counter-argument, but
  3. wouldn’t bother to contact me about it.

It was nice of you to politely write to me and accept my argument. Yet, your failure to argue with me in public has paradoxically left them less likely to agree with me.

To be sure, maybe after you contacted me, we could work out all our differences and write something with a consensus view. Private adversarial collaborations like this are great! But public arguments are also important.

The median person is not receptive

A year or so ago, I read a journal article on some education research that seemed misleading. They had done a certain regression but not included all the coefficients in the paper. I wrote to them several times asking what those coefficients were but got no response. I tried contacting them on Twitter and got no response. I wrote a critique which got no response. I asked Freddie deBoer about it, who emailed the authors and got no response. Then he wrote about their research and got no response. Then Matthew Yglesias wrote about it and you can probably see the pattern here.

The fact is, most people do not respond positively when their results are questioned.

Of course, you might not be like that. You might accept that everyone makes mistakes, and you might see us as engaged in a collective search for truth, rather than a zero-sum battle to prove which human brain is righter.

But how is the person who disagrees with you supposed to know that? We need to design our norms around the unfortunate reality that most people are not helpful when criticized.

Attacking something is a compliment

Why bother critiquing something? Probably you read things that you think are bogus all the time, but you don’t bother arguing against them, because they don’t rise to the level of being worth refuting.

It’s pretty rare to rebut something that’s totally wrong. If it were total nonsense, there wouldn’t be much glory in debunking it. We tend to criticize things that we think are almost right, so close to right that we’re worried other people will accept them.

This is a compliment. If someone criticizes something you wrote, you should feel gratified that you were “right enough” to be worth arguing with.

Proposed norms

So, given all that, here are the norms I suggest:

  • Go ahead and criticize whatever you want, with no need for any kind of warning.
  • Try to phrase your criticisms as against the artifact that you think is wrong, rather than against the person who created it.
  • If you’re tempted to complain that someone is criticizing one of your artifacts without talking to you first, that’s lame, don’t do that.

If you accept that these norms are good, how to push them forward?

Perhaps you should make a point of “arguing without warning” even when it would be really easy to do otherwise. Say I disagree with you, but we’re good friends, and I know you’ll be helpful if I run my ideas past you. The problem is, if I do that, then it looks like that’s what people do when they are friendly. So, arguably I should make a point of not running my ideas past you, just to make it clear that this isn’t aggressive or rude.

Now, I do have one concern about my suggested norms. Say I’m in high school and I wrote my anti-watermelon article for the school paper. If you’re a New York Times columnist and you dedicated a column to systematically tearing my ideas apart… that would be weird.

So maybe if you’re much more famous or powerful than me, you should be more gentle. Perhaps, but I worry about a slippery slope here where the norm flows downward and adds friction to everyone. So use your judgement, I guess?

(Watermelon is, in fact, good.)

Comments at hacker news.

new dynomight every thursday
except when not

(or try substack or rss)