On Rewriting

LIV

2024.07.07

Today, we’re connecting acting to coding and writing. If you want to get good at anything, you just need more reps.

The Secret Superpower

Rewriting is one of the most powerful tools any writer has at their disposal, whether they write essays, novels, spec docs, or code. And by rewriting, I don’t mean just “editing,” I mean taking it from the top again.

This truth is table stakes in the performing arts, but there’s some sort of bias away from it in endeavors like writing that inherently create artifacts as a by-product of effort. Getting over that bias has yielded dividends for my ability to write in english and code. 

A note: This post, in itself, can be read as a sort of rewrite of Paul Graham’s "Hackers and Painters.” He communicates many of the ideas found here better than I can.

Acting

Like performing arts in general, acting is quite distinct from writing in the sense that there is not inherently a permanent artifact of your work produced by virtue of you doing the work. You can keep saying your lines in a mirror, and, unless a camera is rolling, those moments are lost to time, like tears in rain. 

So, why even do it? In the case of acting, the answer is pretty obvious: for every rehearsal you do, you get better. Through a mix of being honest with yourself about your performance and having other people give you feedback, you’re compounding your ability to perform.

The 100th take has the benefit of everything you learned from takes 1-99. You know both what went wrong and what went right.

Granted, if you’re making a movie, when it comes time to actually shoot, they’ll be video evidence of what you said and did. And, in all honesty, recording yourself before the actual performance to tighten your feedback loop can be very very beneficial, too. But, camera in the loop or not, the notion of any actor not rehearsing before they actually shoot or expecting their first take to be so perfect that they don’t have to shoot again is quite absurd.

The expectation is that you will practice before the camera even starts and then, once the camera does start, you’ll keep iterating until you get it right. You have the freedom to “take it from the top” and do the whole thing again, reworking all of the nuances of your performance, including the enunciation, articulation, your physical presence and body language, and, depending on the director, perhaps even parts of the script.

Each later take is better by virtue of the prior one having been done through this process of trial and error. Of course, that’s without mentioning the muscle memory you create by doing the same thing over and over again; as parts of your performance become “automated,” you make it easier to add even more nuance. You don’t have to keep making an active effort to focus on the more primitive pieces, like actually memorizing the lines.

Your repetitions free you to focus on adding even more color to the performance.

While this quite obviously maps to other performing arts and even athletics, there’s no reason it shouldn’t also map to any other endeavor executed by a human.

Re-Writing Words

It was a hard pill to swallow, but some of the best advice I ever received as a writer from a mentor was to rewrite things after I wrote them. Again, I don’t just mean edit things, either. The conversation in which I was introduced to the concept went something like this:

Mentor: “Noah, do you rewrite things?”

Me: “You mean like edit the stuff I write? If so, yes.”

Mentor: “No, no, I mean rewrite the whole thing.”

Me: “Like delete it and start again?”

Mentor: “Yes. Exactly.”

We’re talking about opening up a new document and writing the whole thing again, whatever it is. This is the equivalent of getting multiple takes in when acting or rehearsing before the camera even starts rolling.

This can seem strange and alien; unlike acting or the other performing arts or sports, writing inherently produces an artifact, a ledger of your work. You can scroll up and see your document and all the words you’ve produced. 

Really, though, seeing evidence of your past performance is no reason to get extra attached to it–nothing about producing an artifact inherently increases the probability that said artifact is immediately perfect. 

As a matter of fact, having an artifact should actually make you want to recreate whatever you’re doing even more, because you have the benefit of seeing exactly what worked and being able to reuse the best parts directly. With performing arts, you might have to do some more work juggling those nuances in your head and might ask for the assistance of a director. With writing, you can use the benefit of your artifacts to go further alone. 

I’ve rewritten essays, scripts, applications, and even a novel. If it’s a longer piece of work, I’ll break it into smaller chunks, re-read a chunk, and then rewrite it. It’s a little scary at first, but you can take solace in the fact that you know you still have the original in another file somewhere. I can promise you that you will rarely, if ever, fully return to the original. The process of rewriting can add in beautiful nuance and chains of thought that you very easily could miss before. 

Compounding Creation 

More generally, you’re never simply making art, you’re making yourself a better artist. Your degree of expertise in some field is a function of the quantity of your iterations and how much better you get with each iteration, which is itself a function of the quality of your feedback loop. 

Multiplier = (1+(% increase/iteration))^(iterations)

In the context of writing, if every paragraph you write improves your skill level by .1% on average and you write 1,100 paragraphs, you’ll 3x your writing level. Replace paragraphs written with any other sort of iteration in any field, and you’ll see the same level of compounding, whether it be scenes rehearsed for acting or functions written for programming.

While the reps are important, the increase in your skill level per iteration also matters. While this is more abstract and harder to figure out, I think it just involves more honesty with yourself and a willingness to get feedback on imperfect products. The hit to your ego for temporarily seeming naive is less harmful than the theft to your future self if you never put rubber to pavement.

If you’re honest with yourself and asking questions about the quality of your work, you can get pretty far. Some good questions when writing are:

• Do I understand what I wrote?

• Do I think someone else would understand what I wrote?

• If I removed this bit, would the point become more or less clear? 

That being said, literary writing and programming machine code are pretty similar endeavors. After all:

So, some questions you can ask to reflect on your programming abilities are:

• Does the code do what it’s supposed to do?

• Does the code do the thing it’s supposed to do when the real world is not doing what I expect it to?

• How long will it take me to interpret this code if I look at it again in a week or a month?

• How long would it take an arbitrary programmer of sufficient skill to interpret and edit this code if they saw it?

In programming, the first one is oftentimes a really easy question to answer, which is quite amazing. I’ve been learning rust lately, and the code won’t even try to run if the machine knows there’s going to be an issue with it. Talk about a feedback loop.

Of course, the self guided feedback loop is only part of the battle. You also have to ship your writing and your code, both to the consumers and people who know more than you as programmers or writers. That’s part of the selfish reason I write this blog every week, to keep my feedback loop alive so that every rep is worth more than just the feedback I could give myself. 

And, of course, because I hope that the content is valuable and useful for you, the reader :)

This post is 1,463 words; the document I wrote it on has about 2,000 words. The missing 500 words are from parts that I ruthlessly and callously rewrote. I can promise that this post would be much worse if it weren’t for my mini graveyard.

Live Deeply,