I was once helping a child with her homework. She was supposed to write about a place that was important for her, and had chosen her family’s summer cottage.

Every now and then, she would get distracted from the writing, tell me something about the cottage, and then complain that she didn’t know what to write next.

Me: “Well, you could write the thing that you just told me.”

Child: “Oh! I can?”

Then she would. After a bit, the same thing happened again: she got stuck, distracted herself by telling me something about the place, and complained about being stuck. I then pointed out that she had just told me something that she could write down. This happened about 3-4 times in total, after which she had enough sentences for the assignment.

It’s not just children. On a few occasions, an adult I’d be chatting with would tell me something like, “I want to explain this concept, but I don’t know how”. Then I would ask them what they want to explain, they’d explain it to me, and I’d point out that they just gave a perfectly good explanation.

And lest it sound like I was thinking of myself as being better than any of these people, it totally happens to me too. I will be stuck thinking about how to explain something or how to write an essay, and then I’ll start explaining to some imaginary person what exactly I’m blocked on. If I can’t think of anyone else, I’ll explain it to Claude. Then I will often go from being blocked to just smoothly coming up with an explanation.

Also, I once had a paper accepted for publication that I wasn’t entirely happy with. It was a little abstract and dry to read, but I thought it was at least okay. Soon after, I was asked to give a talk about it.

I sat down to think how I wanted to present it. I found some illustrative examples to help better explain the ideas and thought of a narrative that would carry through the talk, one that even had a bit of a dramatic arc.

As I was doing that, it didn’t take me too long to have the thought of, “fuck, this is how I should have structured and written my paper in the first place”. Something about the act of being about to explain it to people put my mind in an entirely different kind of mode. (I then understood why in some fields, papers start as seminar talks.)

In my previous post, I talked about the difference between Doing One Neat Thing and Getting Some Work Done. I talked about how there is a mode that’s easy to go into, where I feel that I should Get Work Done and treat it as this big task that’s important to push through. As a consequence, the work often feels laborious and, for creative tasks, isn’t necessarily of very high quality.

I contrasted this with Doing One Neat Thing, where you just do one thing that feels neat and then you might end up doing much more work.

A subtype of Getting Work Done is Getting Writing Done, where I relate to writing as this big Thing one needs to do, and it starts feeling hard…

…even though, if I just stopped to explain what I wanted to express to another person, it would involve much less effort and be of a higher quality. (Call this Explaining To A Person.)

Why does this happen?

My guess is that Getting Writing Done often fails because one doesn’t have a target audience in mind. When I am Explaining To A Person, I have some expectation of the kinds of arguments this person will understand, and what will make sense to them. I can then query my intuitive model of “what do I need to explain for this person to understand”, while also tracking my sense of “does my current explanation feel like it is understandable to this person”.

In my previous post, I mentioned that one problem with Getting Work Done is that one becomes disconnected of any internal signals of quality. If I start off from the expectation that I am doing something Neat, then I can kind of hill-climb on that sense of Neatness - keep going in any direction that continues to feel equally or more Neat. Whereas if I just start forcing myself to do work because I should and I’m not optimizing for what feels valuable, there isn’t necessarily any point where my mind would start finding it valuable.

Likewise, if I start imagining a specific person who would read my writing, I can start with an explanation that would make sense to them, and have my mind generate writing that continues to feel like that. It often also works to imagine a more general audience - thinking that I’m writing something “to my Facebook friends” will produce a kind of gestalt feeling of everyone on Facebook who reacts to my posts, that also serves the same function.

In my experience, the choice of target audience also changes the writing. If I imagine that I will post something primarily to my Facebook audience, it will produce a different style than if I am writing something primarily to my blog, or primarily to LessWrong, or primarily to Twitter. This is shaped by my previous history of posting content on that platform - I will feel instinctively drawn toward the kind of content and argumentation that was well-received there before.

This might seem obvious - of course, your choice of target audience affects your writing choices! But while this involves some degree of conscious thinking about what I should make explicit and what I shouldn’t, it’s only to some degree. Much more seems to happen just from having the felt sense of a particular audience in my mind, and then the word-production systems in my brain automatically adjust the “shape” of the text that I want to produce.

Hugo Mercier & Dan Sperber’s influential 2011 paper “Why do humans reason” holds that the process of reasoning - which they define as the process of justifying and giving reasons for particular beliefs - evolved for the function of convincing others, and for spotting flaws in the arguments of others. This doesn’t mean that reasoning would necessarily be hostile. Suppose that I have in mind a course of action that I think would benefit us both. It is then in our shared interest that I try to convince you to join in by offering reasons for my belief that make sense to you. And it is also in our shared interest that you evaluate my argument for flaws and try to poke holes in it, so that we don’t pursue my thing if I happen to be wrong about it being a good idea. Similar to the prosecution and defense in a court arguing for their respective positions, taking up adversarial roles should ideally lead to a process that is truth-seeking overall.

One of the experimental findings that Sperber & Mercier discussed involved asking people to explore views like “What are the causes of school failure?” or “Would restoring the military draft significantly increase America’s ability to influence world events?”. The general finding was that people will produce pretty shallow and superficial reasoning in support of their thinking. However, if they then have to debate their view and be challenged on it, they will show significant improvement in their arguments.

Mercier & Sperber interpret this to mean that arguments for given positions are produced on an as-needed basis, and that this is rational behavior if the purpose of arguments is to convince others. If I offer a superficial argument and you believe it - maybe because you know me and trust me to know what I’m talking about, even if I don’t always explain it in detail - then it would be a waste of time for me to think about a more detailed argument. I only need to generate a better argument if you question it. However, I should still focus on the specific objections that you raised, rather than bother with hypothetical objections someone else might make.

The paper is full of these fascinating kinds of reframings of how seeming failures of rationality make sense if you see reasoning being primarily social rather than individual. One overall sense that emerges from it is that argumentation always needs to be tailored to a particular recipient - the arguments that will convince my grandmother may not convince my friend from school - and that the reasoning machinery in our minds is shaped to do this tailoring automatically.

Unless, of course, I don’t have any clear target audience in mind, and am just Getting Writing Done because I’m “supposed to write something”. In that case, it makes sense that the writing would feel laborious and unrewarding, since that reasoning machinery lacks some of the decision criteria it has evolved to use.