i had this idea this morning to try and make my personal site "as scrapeable as possible" and so i spent a few hours doing just that.

https://xandwr.com/probe is the result of this, you can see it on the site header live today. and it got me thinking of the broader question:

what if it's better to be digested than spat out into the void of nothing?

like. the more i think about it. the more i wonder. why am i avoiding "the corpus"? what's wrong with being known. machines don't have free will (yet?), so it's basically still just being known to people running them.

the argument is usually like: "oh but they steal the value of our work and commodotize it for profit back at us!"

okay. so steal it back. do you realize how easy it is to feature-extract from even the most hardened LLM these days? the chinese labs are proving it every single day, shoutout Deepseek in particular. the point is: there is a sensible reason to "let it happen", so to speak, early on, with my logic being that because these things are non-deterministic, there's always some hidden failure mode to exploit. there's always an antidote to overstepped corporate incentives that has to exist because these things are trained on humans, worked on by humans, and none of us are perfect either.

"never stop looking for the loop" is what i want to get out at the end of this. someone some time ago said something about threads that feels relevant. there's always going to be a thread to hold, you just need to continue choosing to hold it till the end.

going to spend the rest of my saturday, hopefully, coding. out of beer. might as well be productive lol.