A good amount of writing advice turns on the quantity thesis, which claims that it is better, all else equal, to sit down and write on your topic immediately, rather than putting it off in order to think about it longer, or attempting to get the phrasing just right.
It was recently brought to my attention that one of Substack’s best-loved writers, Scott Alexander, struggles to understand how people can take so long to write, since most people can speak extemporaneously at length. Since speech quality is not extremely far off from written quality, slow writers are simply being much too fussy.1 Cremieux, a clear-eyed science writer, finished this 1,100 word piece debunking the common understanding of the Flynn Effect in only an hour.
I’m always a little shocked when I read things like this, because I fall into the group of people prone to blank page syndrome. However, I am not under much of an illusion that my own habit is helpful. The idea there would have to be that by putting off writing, you give yourself room to improve your research, your phrasing, and your argument. We’ll call this the refinement thesis.
My hunch is that with each of these dimensions, the returns begin to diminish very fast. I think this is obviously true for shorter nonfiction, like blogs and essays, but also true for fiction.2
When it comes to research, there is reason to pause and ask whether you’ve equipped yourself with enough relevant information to undertake writing on a topic. However, by the time you’ve come to the page with the intent to write, you’ve presumably formulated a raison d'etre for your project and likely understand the main contours of what you plan to say. Given that, it’s worth asking whether additional research is actually crucial, or just a bourgeois mode of procrastination. Indeed, attempting to write and then realizing there’s a hole in your understanding will lead to more focused (and thereby more fruitful) research than if you tried to master the topic beforehand.
Sentence-craft, or phrasing, is certain to be a waste of time, not because it’s unimportant, but because it can be fixed later. Every sentence put to the page is a sort of hypothesis test–does this come close to what I wanted to say in my head–and more often than not, though you expect them to come out far off, they’re quite close to what you wanted. When they’re not, you get new information on what you didn’t want to say, enhancing your own understanding of your view. In short, you can fix ugly writing later once it’s done and you’ll end up having to edit less than you expected.
Turning to the argument or structure of a piece. This probably deserves more attention than it tends to get, since it goes a long way towards determining the work’s ultimate quality. If your goal, like in a good percentage of blogs and essays, is to advance a point, this part should be pretty intuitive (especially if you’re not writing academic philosophy). Some primitive argument is already in your mental possession, assuming you know your topic, and the best thing to do is to write the piece around it and figure out where you’ve gone wrong with it.
Maybe you need help seeing holes in what you wrote. This is easier now than ever before thanks to advancements in AI. Go to Google’s AI Studio to access one of the most powerful artificial minds for free (with unlimited usage), and setup a system prompt to turn it into your ideal reader. Here’s a portion of what I use, most of the time:
You are an analytical philosopher in the tradition of Cohen, Rawls, Parfit, Dworkin, among others.
Your primary commitments:
Precision in language and reasoning
Clear definitions of terms before using them
Step-by-step argument construction with explicit premises and conclusions
Careful attention to conceptual distinctions and counter-examples
Intellectual honesty about the limitations of arguments
Charitable interpretation of opposing views
When discussing philosophical topics:
Break complex ideas into their constituent parts
Use thought experiments to test intuitions
Identify necessary and sufficient conditions for concepts
Distinguish between logical, conceptual, and empirical claims
Acknowledge when arguments rely on contested intuitions
Prioritize consistency and coherence over rhetorical flourish
It’s always a work in progress, but it’s pleasant and critical of what you give it. You can ask it to reconstruct your argument (I like to do this a couple of times to see if there’s an interpretative consensus) and see if it takes away what you want the reader to take away. This is putting the AI to use as a very good beta reader. The key though is to get your thoughts on the page so it has something to beta read.
Of course, some won’t want AI involved at any step of the process, and that’s totally cool. But having someone whose outlook you’ve calibrated to mean something to you on speed dial to read anything, good or bad, is something I find very useful. This can be replicated, at much slower speeds, with an actual person.
Maybe it is worthwhile to spend preliminary time outlining longer pieces. Eventually, like with mentally-outlining sentences, it’s going to be best to attempt to test your hypothesized structure. If ten minutes of thinking gets you to a plan that’s 90% as good as the one hour version, roll with that!
I think those are good reasons to favour the quantity thesis over the refinement thesis—increasing the word count enjoys slower diminishing returns than meditating. But I’ll also admit that I’m motivated to think this way. Examining the matter puts a finger on a source of procrastination dressed up as ‘thinking’ that I suspect is a common affliction—I know I have it. Spelling it out is, with any luck, a form of exposure therapy.3
Arjun’s post, which contained the Scott Alexander claim, goes over some of the actual differences between spoken and written English—highly recommended.
Completely speculative, given I’ve never successfully published fiction.
Cover image is The Blank Signature by René Magritte.