EDIT: In the interest of transparency, I did have to edit this post. I originally mistyped “2025-05-17” as “202-05-17” in the post metadata which broke the timestamp, mea culpa.

Just write, that’s the thesis. I’m writing this in the github web IDE because I can’t be bothered to reclone my blog repo. I’m probably not going to edit this or even spell check it, but it really doesn’t matter. You’re smart and you’ll know what I mean even if I have a few mispellings. Like 10 years ago, I thought that blogging was super cool, people would publish these treatises like Martin Luther, shot out into the ether, destined either to never be read again, or go viral and get torn apart.

I used to write like my posts would be scrutinized by every internet armchair commenter. I couldn’t have been more wrong. People don’t really care anymore, we’re post quality, everything is slop and it’s been trending this direction for years. I have so many half baked drafts and projects that will never see the light of day because I can’t commit to finishing them, largely for fear of them being critizied. It’s not a rational feeling, but it’s one that ultimatley prevents me from following through and publishing anything.

Even now, I’m getting that feeling that maybe I’ve hyped myself up for a thoughtful post called “just write” and I should really just shelve it.

But I wont.

I’ll be honest, Generative AI completely killed my will power to do generative art (the pre-AI, cool stuff, you know). LLMs have almost tempted me to do give up writing, especially in a professional context, where some of my coworkers have started using chatGPT to reply to emails. Almost, just.

I write my notes in such a scatter shot fashion, that it’s a miracle that I can look back a few months and be able to reconstruct what I was working on at the time.

This reveals to me a very important truth about writing, it’s just lossy compression. I’m compressing my thoughts into a format that you can understand, you read, uncompress and magically high level concepts have been communicated. That’s the thing that bothers me most about LLM text, there’s so much fluff that gets needlessly inserted. Just fire off an email with a few bullet points, your brain is already doing the heavy lifting.

I’m not claiming to be an amazing technical writer or anything, half of the RFCs I’ve written feel super messy, but they get the point across. I haven’t had a single point of feedback that was related to the style in which I wrote it, everyon got the gist and the idea was communicated. Communication and language are messy, I don’t think we should standardize on “standard ChatGPT english”. We should write frequently and with less concern for accuracy, you can always go back and edit your work, no one will notice if you edit a blog post from 3 years ago because you had a typo in it. It should be like sanding and shaping a piece of wood, first rough out the general shape of the thing, then if it needs it, go back and sand down the rough edges.

Anyways, I’m not even going to edit this one, I’m hitting “commit” and not looking back. Just write.

⧠