A longtime reader asked me a few weeks ago about my writing process. Today I’ll finally address the question. He asked me whether I correct as I go, or whether I write as fast as I can and correct later. But my experience over the last few weeks makes me think there’s something a lot more important for me to say.

Do what works, and don’t overthink it.

A few weeks ago, an assignment came down from management to write 12 pages. Typically, spewing out 12 pages isn’t a problem for me. If it’s something I know, I ought to be able to do that in about six hours. If it’s something I know and care passionately about, it can take less than that. If it requires some research, it might take closer to 18. But it’s doable.

Except they had a specific way they wanted me to write it. Outline it. Outline again. Then write. Then rewrite. Then rewrite.

After a grueling 11-hour session this weekend–after spending 20 hours on it over the course of the previous two weeks–I had something that resembled completeness. But it was so bad, I’m glad my name isn’t on it anywhere. The rigid process had me questioning my ability as a writer.

Now, the rigid process might help some people, and that’s fine if it does. But I ended up spending more time outlining than I would have spent just writing the thing in the first place.

I think it comes from someone’s effort to apply process engineering to writing. Process engineering certainly has its place. I’ve written many processes in my time, and for things that need to be done the same way every time–like building a server, and especially when building a specific type of server–it saves an immeasurable number of hours. Because if you have a fleet of web servers sitting behind a load balancer and one of the servers has one tiny piece misconfigured, isolating the problem is going to be extremely difficult.

The problem is that there’s a lot more variance in writing than there is in building a web server. When you’re building web servers, the end goal is for all of them to be very nearly identical, and when they diverge, they diverge in predictable ways, like which DNS and database servers they point at.

Writing isn’t like that. And when you try to force that model on it, the result sounds like a computer wrote it. It’s not a coincidence, because in the end there isn’t much difference between a computer following a program and a human being following a process, except that you have to give the human being health benefits and paid time off.

So if there’s one thing I can say about writing, it’s to not overthink it. Figure out what you need to say, then find a way to say it, but don’t get hung up in the details. Like Don Hicks, the former publisher and editor of Amazing Computing once told me: “Just sit down and explain something to me.”

So, back to that question. Should you correct your mistakes in the middle of your writing, or should you just keep on rolling, then come back and fix it later? I don’t know. Try it both ways. I’ll fix my mistakes if I’m writing slowly, but if I’m on a roll and the words are flying off my fingertips, I just keep rolling with it and let the words come as they will, and fix the stuff later. Stopping when I have momentum is counterproductive. Then again, when I don’t have momentum, sometimes fixing stuff helps me build momentum.

That works for me. It might work for you. Or it might not. Only you can find out. But it helps when you can set it aside long enough to forget what you meant to write, so you can read it and see what’s actually on the paper.

I spent about four years in a world where there isn’t time to outline. Sometimes I had exactly one hour to write a story for publication. I would write parts of it in my head as I walked from the courthouse to the newsroom. Then I’d sit down, bang out a few hundred words, rearrange the paragraphs in order of importance once it had all come to me, spend a few more minutes reworking it, do my fact check, then send it to the editor to do some minor rework and a sign-off. By the time I was going to sleep, it was rolling off the press.

I’ve said it a hundred times or more, and most every other published author has as well, but it’s still worth repeating. If you want to be a good writer, then write. Some of it will be good, some of it will be terrible. Creating a lot of both–hundreds of thousands of them, at least–is the best way to learn to recognize the difference.

I do think a step-by-step, enforced process may be helpful for goading someone who doesn’t want to write into writing. I adamantly do not think a step-by-step, enforced process will make bad writing good, but I’ve seen it do the opposite.