Writing to Understand

You’ve probably picked up by now that I’m a big fan of writing. I wrote You should write more six years ago, and I’ve been advocating a daily writing practice more recently.

Most of that has been advocacy for writing to be read - stuff you’ll publish, probably on a blog. Although the writing itself has many benefits in and of itself, the ultimate purpose of it is to produce something that someone else will read.

Today I’d like to flip that and talk about a different approach. You might produce something you want other people to read at the end of it, but that’s incidental. The writing is being used more directly as a tool, and the artefact it produces exists only to enable something else.

Specifically, I’d like to instil in you the following good habit: If you want to understand something better, write about it.

I don’t mean you should write a fully polished piece ready for public consumption about it. You can do that if you want - and indeed I’d encourage that for things you really want to understand - but there are a lot of easy writing techniques that you can and should reach for to improve your understanding. The resulting writing may never be read by anyone but you, and chances are you will throw it away afterwards, but by the time you are done with writing you should understand things better than you did when you started.

Why? Because you can think things while writing that you cannot think while not writing, and you can use those thoughts to improve your understanding.

Writing as Clearer Thinking

Writing is a form of thinking - you do not generally think what to write and then write it, the very act of writing is itself the thought.

This is just how cognition works - you don’t think separately from acting, instead the thought is an integral part of the act. Consider catching a ball - you don’t think to catch it, then instruct your body to do so, you just catch it (or, if you do try to think to catch it and instruct your body to do so, more likely than not what happens is that you miss it).

Similarly, when having a conversation you don’t generally rehearse everything in your head before saying it and then say it, you just speak. The thought is the verbalisation, rather than the verbalisation being an expression of the thought.

You can think purely in your head of course, and we often do, but speaking out loud or writing it down often forces us to have differently shaped or more complex thoughts than we would in our head. In particular written or spoken thoughts tend to be more coherent than those in your head - they may not have the polish of a finished piece, but they’re a lot more coherent than the version that exists solely in your head, which tends to dart around, skip things, etc. This coherence often forces you to spot things that you were skipping over.

One way in which this is visible is with rubber duck debugging. Siderea explains this very well, but in short rubber duck debugging is a very straightforward technique for solving many problems: You explain them to a rubber duck.

"Duck," I continued, "I want to know, when you use a strap hanger, what keeps the sprinkler pipe from jumping out of the strap when the head discharges, causing the pipe to..."

In the middle of asking the duck my question, the answer hit me. [...]

I turned to look at Bob. Bob was nodding. "You know, don't you," he said.

The process of spelling out all of the steps forces you to have thoughts you were skipping over when you tried to have them in your head, and spelling out those detail can help you make sense of them. This isn’t fool proof, but generally you’ll improve your understanding of the problem even if you don’t solve it.

An example where this often comes up for me in practice is the site mathoverflow, which is a place for asking research level maths questions. In order to show that I’ve cleared the bar for being suitable for the site, I put a lot of work into my mathoverflow questions - explaining things clearly, saying the things I’ve tried and why they didn’t work, etc. The goal is to get the explanation of the problem to the point where it’s relatively unassailable - nobody can look at it and say “Have you tried doing this obvious thing, you idiot?” and be right.

Then, nine times out of ten, I delete the question, because explaining it that clearly was enough to understand it well enough to solve it.

As well as forcing you to think things through clearly, writing also just gives you access to intellectual capabilities that you can’t have when you try to think a problem through bare brained, because you can organise a lot more thinking on paper (or electronic document) than you can in your head because your working memory is so sharply constrained.

This is particularly visible with things like mathematics, where “showing your working” (in reality doing your working) on paper allows you to escape the constraints of working memory in your own head. A lot of mathematics (particularly the conceptual bits) can be productively done in your head, but maintaining a long and complex proof or calculation without writing it down can be almost impossibly difficult.

This increased working memory from thinking on paper effectively allows you to hold more things current in your thoughts at a time. Without writing it down you will lose track - by the time you’ve thought of the tenth thing, you’ve forgotten the first.

With this increased working memory comes increased power and clarity of thought, because one of the key advantages of writing over thinking out loud is that it turns your thoughts into objects that you can manipulate. You can look at the things you’ve written, you can manipulate them, you can go “Is that right…?” and correct them, you can make connections that you wouldn’t have seen without making them explicit.

Tools for Thinking with Writing

There are some easy tools that are applicable to most problems or things you need to understand that are worth deploying in most circumstances.

Explanatory writing.

Explanatory writing is just a written version of the rubber duck debugging technique. Explain, as clearly as possible, something that you already understand, but that leads to something that you don’t understand. e.g. the technical problem you’re trying to solve, how it feels in a particular situation, what the theorem you’re trying to prove is about and what you would need to complete the proof, what the symptoms of the bug are and why you haven’t been able to track it down.

I’ve written about how to explain things before, and one of the key things I emphasise there is that you have to have some idea of who you are explaining to. For writing to understand, my recommendation is that you pick one of the following two audiences:

  • An expert who knows a lot more about the domain than you do but nothing about your specific problem.

  • A version of your past self who has not yet encountered the thing you are trying to explain at all.

The former is particularly useful for things that you are stuck on, the latter for things that you kinda understand but are trying to achieve clarity on.

List Making

Rather than writing prose, write lists. Come up with a heading such as “Questions I wish I knew the answer to” or “Things that could be causing this” or “Ideas I could try”. Something that gives you a list of things to potentially work through that would improve your understanding of the problem.

You should then pick a target number of items - it should be slightly unreasonably large. e.g. guess the number you think you could easily come up with, double it, and round up to something nice seeming. You should certainly aim for at least ten.

Now write that many items.

Importantly, don’t worry too much about whether the things you’re writing on the list are good. The point is to fill the list - you’ll throw away most of it.

This is a technique I got from De Bono’s book, Lateral Thinking. It’s annoyingly quite a good book, but it’s written by De Bono so it’s also a book that’s good at being annoying. In it his point is that there are essentially two types of thinking: Lateral thinking, and vertical thinking (I tend to use “critical thinking”, which doesn’t mean quite the same thing but never mind). Lateral thinking you generate ideas, vertical thinking you prune them down.

De Bono points out that one of the difficulties people often have is that they move these two things too close together - they try to avoid generating bad ideas in the first place, rather than trying to generate a mix of good and bad ideas and decide later which is which. This list-making tool is designed to break you out of that.

This is one of the nice thing about writing based tools, which is that you can make use of their explicitness to shape particular directions of thinking.

Free Writing

Free writing combines some of the advantages of the previous two techniques, but is its own thing.

Free writing is writing about a subject more or less without structure or purpose. You pick a set amount you’re going to write (a word count, or a page count) and then you write that much.

It doesn’t have to be good. It might contain some good bits, but that’s not the point. The point of free writing is that it encourages you to get down everything you can think of. Much of that might be garbage. That’s OK. You’re not going to show your free writing to anyone, you’re just going to write it to find out what you think.

As with the list making, it is worth setting your goal with free writing to be about twice as long as seems reasonable. I generally find it’s only when I feel like I’ve run out of things to say that the really good stuff starts coming out.

Free writing can be a mix of ideas, explanation, follow on thoughts, recording a literal stream of consciousness, whatever you like. Often it’s a good opportunity to connect up different things - “that reminds me of”, or explaining more things you’ve already written.

But it doesn’t matter. The only two rules when using free writing this way are:

  • It has to be on topic (optional).

  • You have to write at least your target amount.

This essentially puts a bit of very light pressure on you to generate new and interesting thoughts on the subject, and see how they fit together, and it’s almost always productive.

You can also do this without any particular goal in mind, which is why the topic is optional: It can be a good technique for figuring out what you need to understand next, by basically letting you write down some stuff you’re thinking about and iterate on it. I’ve been doing two pages of hand written free writing every morning as part of a thing called morning pages (suggested by the book The Artist’s Way, which is supposedly good but I haven’t read it).

Doing recurring free writing (even if it’s not morning pages) seems particularly useful, because it turns it into an ongoing process.

Or you could write a blog post

Honestly, it’s a really good approach. Try all of the above first if you’re not comfortable writing in public, but there really is something uniquely good about writing something that is intended to be read by other people to read. It serves as a good forcing function that puts all of the advantages of writing together in a hard to imitate way.

But the above are all still useful too, because they are much lower effort, and they can help you explore your thinking on subjects and in directions that you are not quite ready to take public yet.

Obligatory Plug

On the subject of thinking better: As mentioned last week, my friend Lisa McNulty will be running her online course “How (not) to think like Sherlock Holmes”. It starts this week (Thursday July 30th), and it runs for five weeks but you can also buy a taster ticket for just the first week if you want to see what it’s like before you decide whether to sign up for the rest. You can buy tickets here.