Blogging FAQs

Originally posted 2023-07-11

Obligatory disclaimer: all opinions are mine and not of my employer

How long does it take me to write an essay?

Definitely the most FAQ.

I spend anywhere from 5-100 hours spread over 1-12 calendar weeks. “Why did Google Brain exist?” and “Data Oriented Programming in Python” took on the longer side, and “A Research Codebase Manifesto” took on the shorter side (but only because I’d already done so much thinking on the topic). This FAQ took 2 hours in two sittings.

There is no clear separation between ideation/writing/editing; I think by writing. I’ve deleted as much as 80% of my writing before publication. Most of my essays are based on insights that I’ve already vaguely understood from the work I’ve done over the years. The writing time goes into isolating the insight, testing its limits, and finding the clearest way to present it.

This is a lot of time to spend per essay. I think it just reflects my personal expectations on how thoroughly I want to understand an idea before moving on. My expectations have risen over time, probably more quickly than my ability to meet my own expectations, leading to overall fewer essays and more time spent per essay. Still, as long as I manage to publish a few essays a year, I’m happy with my output. I wouldn’t want to publish anything I didn’t feel proud of.

Why do you write?

There are three main reasons. First, I like thinking, and writing is the best way to think critically and honestly. Second, I like teaching, and writing is a great way to pass on wisdom. Third, it gets me street cred. It’s like a resumé, but one that people actually enjoy reading. It opens doors to conversations that wouldn’t otherwise happen.

How do I get started writing?

Write. It’s really that simple.

It doesn’t matter what platform or tech you use. The set of people who talk about the right way to start a blog will have stopped blogging within a year on average, and it is precisely these people who seem to be drawn to arguing about whether Medium or Substack is a better place to blog. Ignore them. In ten years, people will still be reading my blog and as they stumble on this FAQ, they will wonder what Substack was.

My only recommendation is to (eventually) own your content in a platform-neutral format, so that you can migrate platforms at will. I started on Blogger.com; moved to a personal domain (a django site hosted by… Hostgator, IIRC?), then rewrote it as a static site generator hosted on Google Cloud Storage. The latest iteration is a static site hosted on Firebase, for its SSL support. You can see the static site generator and raw essay files on GitHub..

What should I write about?

Write what you know and have experienced. The world is a better place when you delete your hot take on a flamebait topic where you have no particular expertise.

For many of us, this means job experience. As a result, you might feel like you would be violating confidentiality or that you might be offending your employer by writing about your work. Your gut instinct is probably correct. So, instead of writing about company strategy, write about the general strategic landscape that would enable someone to understand the company strategy (if they had access to it). Instead of complaining about your employer, think through how the situation might be improved. It’s more positive-sum, lets you roleplay your boss’s job, and helps you level up your soft skills. Finally, you can write on a time delay. “Deep Learning on EMRs is Doomed to Fail” was an essay I could have written in 2017, but at that time it would have been nonobvious and definitely confidential. In 2022, it was something that everybody in the field already knew, and having quit Verily in 2018, I was able to say it out loud, since there was no chance it would be mistaken as an official company stance.

Not all of my writing is public. I’ve written well-reviewed documentation at Google on TensorFlow and ML readability, and was fortunate enough to have some of it published as official TensorFlow guides.

How do I get better at writing?

Write more. Be intentional about the editing process. Read your blog posts out loud. Try to explain the idea to a friend. Ask a more experienced writer to give you feedback. Decide who your target audience is and decide what their level of background is. (I target a relatively sophisticated audience because it’s honestly tedious to write too much introductory material.)

My early essays are bad. I look back at them and think, “wow, I have no idea what this guy is trying to say”. I do occasionally go back and re-edit them, and I have even deleted some that I deemed unsalvageable. It’s now been over ten years since I started writing, and my writing has improved tremendously. Yours will too.

The good news is that at first, nobody will read your blog, so there’s really nothing to be ashamed of by publishing your writing.

To get people to read your blog, share your work on Reddit, Facebook, Linkedin, HN, wherever you can get eyeballs on your work. It’ll probably disappear into the noise; sometimes you’ll get snarky comments on how bad your writing is and how horrible of a person you must be. I used to get upset when I saw those comments. Nowadays, I endorse the “u mad bro?” school of philosophy.

Have more questions?

Email me.