Very nice that eightyeightthirty.one is really cool. I wish that the really heavy expensive 3D visualization of the data was opt in because I basically can't run it without lagging out the whole browser lol. But it's pretty cool. I wonder if you could get a local LLM to try to guess how to categorize the pages and get some good results. That kind of data science stuff is something I've never done before tho
@goeshard I'm in the no-AI webring so LLM is a forbidden tool, but I think useful curation truly requires human eyesballs. I attempted an automated directory inspired by Melon's Surf Club, which pulls relevant CSS from sites, but there are so many user quirks (you would not believe how many people title their site index "home") I could not find a meaningful way to automate it, at least not with the data I have.
![auberylis [dot] moe avatar](/site_screenshots/34/10/astrossoundhell/index.html.50x50.webp)
![[ cave of kelpius ] avatar](/site_screenshots/28/07/kelpius/index.html.50x50.webp)
Very nice that eightyeightthirty.one is really cool. I wish that the really heavy expensive 3D visualization of the data was opt in because I basically can't run it without lagging out the whole browser lol. But it's pretty cool. I wonder if you could get a local LLM to try to guess how to categorize the pages and get some good results. That kind of data science stuff is something I've never done before tho
@goeshard I'm in the no-AI webring so LLM is a forbidden tool, but I think useful curation truly requires human eyesballs. I attempted an automated directory inspired by Melon's Surf Club, which pulls relevant CSS from sites, but there are so many user quirks (you would not believe how many people title their site index "home") I could not find a meaningful way to automate it, at least not with the data I have.