I don’t want your rationality. I can supply my own, thank you very much. I want your data. If you spot a logical error in my thinking, then please point it out. But short of that, among mostly-rational people, I think most disagreements come down to a difference of intuitions, which are rooted in a difference in the data people have been exposed to, and instead of presenting a logical counter-argument, you’re better off doing a “Rationalist mind meld” where you share your data.

I find myself making this mistake a lot. Let’s say I came across this (fake) comment:

There aren’t really classes in America anymore, because there’s no legal class distinctions and everyone has the same opportunities. And class mobility is very high.

They’re wrong and I know it. Maybe my instinct would be to reply like this:

Actually, a lack of a legally enforced class system doesn’t imply there are no classes. There is a lot of wealth inequality in America, and children born to poorer families don’t have the same opportunities as richer families. Class mobility is low in America, and classes are hugely significant.

This is some combination of:

Asserting what I believe without justifying it.
Justifying what I believe based on simple facts everyone in the conversation already knows.
Pointing out logical errors (“lack of legally enforced classes doesn’t imply there are no classes”)

The problem with this kind of response is that my own beliefs about class were not formed by this kind of logic. My own beliefs about class were largely informed by this Slate Star Codex essay on class, the linked essays about class, and Paul Fussel’s book Class. All this material describes in detail how the different classes act, look, earn their money, and how easy it is to move between classes. I’m expecting the person I’m talking to to believe what I believe about class while having heard almost none of what I’ve heard about class!

In the Rationalist community, there’s the concept of a crux, which is essentially the core reason why you disagree, and double-crux, a pattern where two people search for each other’s cruxes to find out the source of their disagreement. As has been pointed out, this doesn’t work so well in practice because disagreements often cannot be “traced to a single underlying consideration”. In the case of class above, I don’t disagree with them because of one specific fact or belief. I disagree because I have formed a complex and robust world model about class that the person I’m talking to just doesn’t have.

The correct way to resolve this is not with logical arguments, but with a mind meld!

Spock doing a mind meld

In Star Trek, Vulcans like Spock can create a telepathic link between themselves and someone else to exchange thoughts and memories directly. In the show this is used for various purposes including coming to a better understanding of the perspectives and desires of hostile alien species.

If I was to do a mind meld with the person who thinks class doesn’t exist in America, it might look like this:

My intuition is different than yours, so I’d appreciate if we could mind meld here. For my part, my intuition mostly comes from this Slate Star Codex essay on class, the linked essays about class, and Paul Fussel’s book Class. Probably the Slate Star Codex essay alone should be enough to give you a general idea. Are there any sources informing your intuition that you’d like to share?

Bam! MIND MELD!

It doesn’t have to be explicit like that, by the way. You don’t have to use the term “mind meld”. And it doesn’t have to be just essays. It can be any information that contributed to your understanding, whether in the form of scientific studies, books, blog posts, videos, datasets, podcasts, or even descriptions of personal experiences you’ve had. The latter is actually quite common on Hacker News. People there often provide little anecdotes instead of directly commenting on the main post. Here’s a short one from the many I saw just today:

Of all the impressive software developers I had pleasure to meet before 2012, I never met one who did it for money. They loved their work, the craft, the science, and sheer joy of the creative process. That culture ended quickly pretty around 2012-15, but I never figured out why.

Or this one:

This was a nice trick to protect text from copying. For instance, student assignments. Students could still use digital camera on CRT display, but 20 years ago cameras were costly and students did not have them. And typing text from scratch was a tedious job. So online served assignments were not shared too fast.

While you can’t necessarily trust a single anecdote, that’s not the point. The point is that instead of dealing solely with logical arguments — which have their place, such as if someone has visibly committed an error in logic — you’re also experiencing a sort of gradual mind meld with the whole community.

Most people don’t form their beliefs on the basis of pure logic. Instead, belief formation often looks like this:

People expose themselves to a large amount of data.
That creates an intuition.
People generate logical reasons to try to explain why their intuition is true.

The third step is the easy part. It mostly exists so you can communicate with others, and as a sort of sanity check on your intuition. Your intuition can lead you astray, and your logical mind exists to correct things when that happens. Think of your logical mind as your CPU and your intuition as your GPU. Your main goal is to train your GPU software to be rational, and your CPU exists to facilitate that training. You can’t rely on your CPU for many things because it’s too weak. No amount of reasoning about the rules of chess will allow you to beat me after I’ve played a few hundred games. No amount of reading analyses of fashion can compare to looking at 1000 pictures of well-dressed people. To train your GPU, you need to find good, high quality training data, and that’s where the mind meld comes in. If we focus not only on pointing out failures in logic, but also in sharing our training data, we’ll all end up more rational in the end.

Rich Text Copier