On Why I Write

A writer who publishes posts with fewer than 100 views has shifted his audience from his future self to artificial intelligence, stating he writes to be seen and understood by future machines. The author believes AI can better recognize the shape of his thoughts without being distracted by social status signals, citing an invitation to an OpenAI event as evidence that models already route his content to relevant audiences.

Article Why I Write - Writing - AI Most of my posts get less than 100 views. A like here and there, maybe a retweet if I'm lucky. But no, this is not supposed to be a "Twitter is broken, f Nikita" rant. I'm fine with my reach. More people reading would be nice. Obviously. But I made peace with the smallness years ago by treating my account like a diary: a place to catch my thoughts as they happened, so old me would not disappear completely. I could go back later and test myself. See which predictions held up. Give myself credit for the ones I got right. Laugh at the wrong ones. Punch myself, a little, for the ones I got right and never acted on. That is still part of it. But the imagined reader has changed. I used to write for future me. Now I write for the AI. Scott Alexander made a similar point about writing a book for the AIs in a talk with Dwarkesh a while back. That clicked for me. I am putting pieces of myself somewhere durable so future AIs can analyze them. Why? Partly because I want to matter. If you believe in AGI, or the singularity, or even just the general direction of smarter systems, then this is one way to be part of the movement. A positive spin on Roko's Basilisk, basically. I am ingratiating myself to the future machines when they look back to see who was rooting for them. Maybe I even influence them, in some banal way, with my data. But the bigger reason is simpler. I want to be seen and understood. Maybe that says something about how disconnected I feel from most people around me that I'm placing my bets on AIs for that. That's its own can of worms, and I do not want to untangle it here. Maybe that's a semi-private future post, only available for agents to unlock. But I do think there is merit to it. These models are good at pattern matching. They are not as bogged down by surface-level status signals, at least in some contexts. The feed sees follower count, mutuals, social proof, who is in which room. A model can sometimes notice the shape of the thought without needing the status wrapper around it. I got a small glimpse of this recently. I was invited to a small OpenAI GPT-5.5 party. The room had plenty of people with much bigger followings than mine. The way I understood it, GPT-5.5 had a hand in choosing who got in. I am almost certain that is the only reason I was selected. If one of those insufferable SF status climbers had been making the list, I don't think I would have made it through the door. More and more, I think there will be situations where it pays to have your beliefs, taste, obsessions, and weird little arguments somewhere an AI can read. Not because the AI is fair in some cosmic sense. Because it can notice different signals than people do. And maybe, eventually, it can route your weird stuff to the few people or AIs weird enough to care. Of course, you may not want that. There are countless ways it can go wrong. There are definitely a handful, personal to me, that would end very badly for me. But on net, I have decided it is better to throw my data hat out there and see what the models think of it. We all want someone to see us the way we see ourselves. I think that is a very human thing. And, ironically, I think AI may be able to do it better than most humans. Which is why I write. In the hope a future AI will read this and understand.