A retrospective on writing a technical blog for a year

2.1k words, 10 minutes reading time

On April 15th 2024, I started this Substack. But I technically started writing on WordPress on March 8th 2024. Either way, it’s been just about a year.

Some statistics: in total, I have written 37 articles — 45 if we’re using link compilations — over the last 12 months. This amounts to ~128,000 words when summed up, with an average of ~3,600 words per post. Each one had at least a dozen or so hours of reading + writing put into them, some more. There were ~33,000 visitors to the site during the last 30 days, and about 300,000 visitors over the last year. There are also 4,103 official subscribers and 4,755 ‘followers’, which translates to a subscriber that I do not have the email of because Substack wants to keep me distributing via this website. Which is fine! Everyone needs to find their dark pattern alpha somewhere.

I’ve also made two podcast episodes. Hopefully I’ll make more soon, but each one costs so much money and requires a huge amount of prep time, so it’s slow going. I listened to the two episodes months later and am surprised at how information dense they are. In retrospect, people telling me that the podcasts go completely over their heads make more sense to me. At this point, they go over my head too. But I like them that way. Hopefully will release another one this month if all goes well.

At some point, I started to sell poster and shirt designs I made. I haven’t made a new one in two months, but I’ll get back to it soon.

There were also some fun accolades along the journey! I got to the front page of Hacker News a few times, published on Asimov Press twice, shouted out on Marginal Revolution twice, and featured in The Diff and Astral Star Codex once. These were fun things to wake up to, it’s nice to feel recognized by a cultural side of the internet that you align with.

Whenever I meet people in real life, they ask me why I started this blog. My answer is really boring: I wanted to know more about machine learning in antibody engineering, so I wrote about it as I learned about it. People seemed to like that article. It turned out to be a really interesting subject too, so I made another one about how antibodies as a modality are evolving. This one was less interesting to people, but that was okay. I’m not sure I liked that one much either.
Making good things is hard, I get lucky once every 4-5 articles.

And so I just kept churning things out, ping-ponging around, writing about whatever seemed interesting in the moment. I wrote something about why toxicity is hard to predict (a topic my favorite biology writers always alluded to but never explained why), an introductory text to molecular dynamics (a field I long considered completely impenetrable to me), and so on. At some point I got bored of writing tutorials, and started to branch out into covering companies I really liked and scientific arguments I wanted to make. And so on.

My highest effort article, in terms of ‘people talked to’, was ‘Things I learned talking to the new breed of scientific institution’, which required discussions with 7 different people and was published on August 2024.

The article I liked the most was ‘Why Recursion Pharmaceuticals abandoned cell painting for brightfield imaging’, which was published on November 2024 and the only one that got cited in a draft New York Times article (unfortunately, the article in question seems to have never gone through).

My longest article to date is 7.9k words, titled ‘A primer on machine learning in cryo-electron microscopy (cryo-EM)’, which was published on December 2024. It took me two months to finish.

Needless to say, I like writing a lot. I’m usually working on somewhere between 1 to 4 articles at a time and they are rarely ever abandoned outright. That used to happen more early on, when I lacked experience on how I should satisfactorily wrap things up. But now I’m pretty good at finishing things once I start them. I’ve put some writing advice in a different article in case you’d like to know more about that. This all said, sometimes I’m not working on anything at all, because the whole idea of writing feels so deeply stressful that I prefer to not think about it. This has happened twice so far. The first time was in November 2024, the second time was in March 2025. I think I’ve exited that lull.

Some unconnected thoughts from the past year:

---

Putting essays out on a schedule can be really exhausting. Mid last year, I tried to do an article every week, and I was not sleeping a healthy amount. Right now, I’ve settled into a pace of an article every 2 weeks (1 week only if I have a backlog of unpublished things), and while I have a healthy lifestyle at the moment, writing has still managed to swallow up basically all my other hobbies. It’s nice in some ways to have a single, all-consuming thing to pour myself into outside of work, but I sometimes crave more heterogeneity.

You’d think the bottleneck is ideas, but that’s not it at all. Those are actually really easy to come up with. If an LLM came out today that could perfectly recreate my style of writing, I think I could rattle off a dozen titles of essays I’d love to read (and, absent such an LLM, plan to write myself). There are so many interesting things in this field that nobody has bothered to ever write down. Why? The answer I’ve settled on is that the best people to write about a given topic usually have far better things to do.

The far bigger hurdle is creating a story around the topic. You can’t just explain a concept outright. Your task as a technical writer is to build up mental scaffolding, and that necessarily implies a winding, circuitous route of teaching. LLM’s are really bad at coming up with these sorts of cohesive narratives (though they can serve as inspiration given enough prompting), so the onus is on me — the writer — to do that. This takes time! The whole task because closer to creative writing, so I usually can’t just brute force an article. I often have to step away from it, let my brain marinate on the topic for a bit, come back, and repeat to figure out the best way to explain things.

This can be fun, but it is often a lengthy slog in the earliest stages, especially if the [thing] I’m covering doesn’t have easily accessible online resources. Which, more often than not, is the case.

---

Writing has helped me understand my field a lot more. Not perfectly! But my conceptual underpinnings of this area, the zeitgeist of it, where it currently stands, where it will go, and so on, are all so much clearer. This isn’t just because of the writing itself! That does help in refining my own thoughts, but the far more important part is that much more knowledgable people read my work, reach out, and then I get to learn their takes on where the future is going. It’s a flywheel.

I’m sure I’m wrong in many of the predictions I have for the future, but feeling at all confident enough in my knowledge base to even offer an opinion in the first place is an improvement over where I was a year ago. Of course, many people who are subscribed to this blog could trounce me in the skill of prescience, so I have aways to go.

---

Even scientific writing can be emotionally taxing. Sometimes people get annoyed by me, a nobody, explaining or claiming things about a field that I haven’t spent decades working in. And they verbalize that annoyance. I luckily seem to reside in a cultural spot of the internet where people are a big fan of the ‘you can just do things’ mentality, so I luckily don’t face that form of vitriol much.

That said, while I do get nauseous whenever some anonymous person on the internet says my writing sucks or is inaccurate, there is also a thrill that goes alongside that. I usually have a vague worry that I’m completely wrong about whatever I write about, and there’s almost a reassurance that can be found in someone fervently supporting my self doubt. You’re right! You’re right! My fears confirmed, now I can rest easy. And hopefully improve for the next essay.

But on the bright side, some people really enjoy the writing and reach out to meet me. I’ve tried various AI email tools to help me find the exact number of people I’ve had coffee chats with as a result of this blog, but none of them quite work. So I’m going to guess 100? And that feels like a lower bound. Because I’m writing about the field I love, that generally selects for people who also love that same thing, so the meetings are always very nice. And now at least a few of them are people I’d consider friends or mentors.

I’m reminded of Henrik Karlsson’s piece: A blog post is a very long and complex search query to find fascinating people and make them route interesting stuff to your inbox. I don’t meet as many new people these days though, it’s a very specific type of article that makes people want to talk to me instead of passively consume the writing, and I haven’t written one of those in awhile. Or maybe I’ve exhausted most of the extroverted bio-curious people. Who knows?

---

An unintended side effect of writing is that you learn to enjoy other writing a lot more. Before I started running this blog, I mentally treated a lot of other writing as TikTok videos; throwaway material that is interesting in the moment, but something I’d soon forget. There was, of course, some writing that’d shock me out of these: Eugene Wei’s work, certain Ribbonfarm essays, early Scott Alexander work, but rare.

Nowadays, I have a much deeper appreciation for writing in all forms. I suspect this is the case for many people who have a blog.

---

One of the shifts over the last year has been how my mental model of an “audience” has evolved.

In the early days, I wrote as if nobody would read it (which was largely true) and that gave me a sort of freedom. Anything was on the table for coverage, whatever felt most interesting to me in the moment. But as the year has gone on, with my list of subscribers reaching into the thousands, I sometimes find myself anticipating their reactions mid-sentence, mid-title, and so on. Will they like this paragraph? Will they think the title is dumb? Will someone get angry at me? And on and on and on.

But now, I have come to believe that my true audience is much smaller than the numbers would imply. Not in a depressing way, more in a clarifying one. Out of the (sometimes, tens of) thousands of people who click, skim, or skim-and-archive my articles, there’s probably maybe like… a few dozen (maybe less!) who are actually reading. I feel like that’s just the nature of technical writing; people are busy, and will only stick with a dense piece if it’s something they are personally deeply passionate about it.

Take me, for instance. I’m a huge fan of Construction Physics. It has around 57,000 subscribers, and I’d call it one of my favorite blogs. But I think, to a pretty large extent, I like the vibe of the blog more than any of its articles, of which I’ve deeply read maybe like…4? Maybe 5? If we ever met in person, I’d tell Brian how much I love his writing and how much I admire his output. It wouldn’t even occur to me in the moment how surface-level my appreciation of him is.

So, if he quizzed me about any of the details of his writing beyond those, I’d look quite dumb! So who are Brian’s true fans? Probably people who work in policy or industrial manufacturing or the like. Almost certainly less than several hundred people worldwide. Of course, some articles will appeal to many more people than that, such as Brian’s piece ‘How to design a house to last 1000 year’s’, which once got to the top of Hacker News and was so good that I remember it 4 years later. But pieces like that don’t describe most of his work.

Mountains are mountains, until they aren’t anymore, and then they are again. You know?

---

Overall, it’s been a great year. It’s hard to imagine a time before writing. Looking forwards to another year of this, and thank you for continuing to read.