On Good Writing
What I want to read.
I want to write things that people want to read. Today, I thought about the exact end state my writing should have to fulfil that goal. To describe exactly what would make people want to read it.
I was inspired by an exercise from the great scientist Richard Hamming. Spell out the biggest thing you want, then list what stops you from getting there. This is simple to state but hard to do.
What would I actually want to read?
This essay is the result of that exercise. It comes down to novelty, experience, and opinion.
Novelty
What is written should be new.
Novelty in your writing gives readers a chance to discover new things themselves. This is because new ideas give way to even more new ideas. And that amplifies the strength of your writing. New things are like budding branches on a tree. If you can write something new, it gives your readers a chance to find more ‘budding branches’ that might stem from it. They might take what you have written, and write something new themselves. Conversely, if something is old, then the branches are likely to have already grown.
I started having a lot of new ideas about spectroscopy when I talked about recent problems in the area with others since I was on the fringe. On the other hand, learning about the classical things in physics didn’t help me come up with ideas for research. Sure, learning about the history of electricity is fun and inspiring, but if the presentation isn’t novel, then there isn’t much I can do with it.
Novelty is also efficient for society. It would be a shame if there were a thousand people who all wrote about the same photosynthetic mechanism of the English pea plant. I try to check if what I write is new by writing a sentence of the form ‘this is new because X’ to myself.
The hardest form of novelty is to make a discovery, because of the time it takes to do so. I myself am trying to write something novel in this sense. I want to write about my work creating datasets for protein spectroscopy prediction. I know from my supervisors that this work is highly likely to be novel.
But it has been hard, and right now it is incomplete. No definite results have been made, and I still have to do some analysis on the predicted spectra.
I have spent hours doing mathematical exercises from an old book by Szabo and Ostlund. I’ve had to learn the syntax of Slurm, a language used to run simulations on supercomputer. I’ve also had to go through the painstaking work of manually labelling orbitals from images, and I’m not a chemist. It will take a while to come up with something good. So far, I have spent about six months working on this problem. I expect to go on for another year.
Ironically, once I complete it, I might try to publish it in a journal. It will probably be read less than the content I have written on this blog.
If the hardest form of novelty to achieve is making a discovery, then this is especially hard in the natural sciences. Regardless if it’s in theory or experiment. Judging by the six years it takes to do a PhD, it takes a very long time. And it usually requires the help of an experienced supervisor who tells you where the problems are.
When people achieve this ‘discovery’ novelty, I want to read about it. This used to be something only academics did. But now there are independent researchers doing their own thing, and publishing it. There has been a renaissance of independents doing experiments in AI safety. I also find this novelty when journalists decide to do their own investigations.
This is where I find many popular science books to be heavily disappointing. Many authors typically only review the classic topics not on the fringe of science, not offering new discoveries. If most of the content would already be known to an undergraduate, or is easily found on Google, then I wouldn’t consider that novel.
You can find this ‘discovery novelty’ in abundance by just looking at scientific journals. But do I enjoy reading them? Not really. For something to be something that I ‘want’ to read, it requires a lot more than that. A different kind of novelty which I will call experience, which brings me to my second point.
Experience
Whilst novelty from discovery is hard, it isn’t the only way to be novel. An equally good way is to write about personal experience. I think of personal experience as the life outside of the technical thing you are writing about. It is highly unlikely that your personal experience matches up with anyone else’s.
Personal experience that I like to read could just be mundane details of the setting in which you worked. Like where you lived, what kind of food you ate, or the exercise routines you might have had. Suppose you walked three hours a day to think about a result (Andrew Wiles). Maybe you chain-smoked cigarettes when doing math (Oppenheimer). These are details that I want to know in writing. You don’t even have to justify why you did them. Plain descriptions give me enough information to be happy reading, and the ambiguity of whether they made you a better person is exciting. I might question how your health habits might have affected your work.
Personal experience might be about your internal motivations. I want to read about your own reasons to pursue research. For instance, do you have a desire to alleviate suffering? How did you feel when your experiements had failed? What was it like when you win a prize (Venki Ramakrishnan)?
It is the best when the feelings are surprising. I was surprised that Haruki Murakami’s desire to write a novel came suddenly from hearing the sound of a bat hit a ball at a baseball game.
This is why I love to read science memoirs. And why science memoirs help to calibrate me in my own work. The most memorable examples include Fei Fei Li’s autobiography and Venki Ramakrishnan’s autobiography.
Personal experience offers practical advantages. When personal experience is written, it breaks down a confidence barrier that otherwise might have made you stop reading the material.
For example, if someone wrote about a piece of mathematics, and I’ve started reading the first sentence, I might prematurely stop if I thought it was too hard. This would be unfair to both the author and the reader. If the author gave some context about who they were and their background, it would give me a chance to then think ‘oh, we have a similar background, maybe I am capable of understanding what you are about to say’.
Method
Part of personal experience is your method. Method to me is explaining the system of rules that drove you to your conclusions, or the problems you want to solve. If you spend eighty percent of your time thinking about problems and twenty percent trying to solve them, then that is something that I want to know. This is how the reader gets hooked, because they will think that maybe if they tried mimicking your method, then they too will also find something worth discovering. I also want to know how much luck you think was involved in being successful.
There is great variation in method. There is a great essay about the difference between ‘theory builders’ and ‘problem solvers’ by Tim Gowers. Gowers talks about how some mathematicians solve problems for the sake of understanding mathematics, and others understand mathematics for the sake of solving problems. Two very different methods.
Part of method is writing with opinion. Opinion is a prior about the world that hasn’t been proven yet. For a reason I can’t really explain, opinion really draws me in. I think this is because there is great value reading about someone’s priors, since I’ve come to believe that value judgements and normative statements are just as important as the scientific method.
Enjoyment
Above all, I want to feel like the author has enjoyed writing their piece of work. The piece of work should feel like it was written without constraint, and without pressure. Almost as if it was a hobby that they were doing in their spare time. If you are writing after your working hours, then that freedom allows you to take risks because your life doesn’t depend on it.
I almost feel like enjoyment is more important than the gravity of the subject matter itself. Writing that deals with the mundane in novel ways is especially exciting to me, because those things are fun to think about. I remember when one physicist suggested he write about the equations to make a perfect roast chicken. I wished he ended up writing that piece of work. Or when my friends mention their theories about philosophy in physics, I hope they eventually write about it. That stuff is golden to me.
So now I think I know what makes good writing, I just have to do it.