| Website, blog, etc. | https://www.marginalia.nu/ |
| Search Engine | https://search.marginalia.nu/ |
| Encyclopedia | https://encyclopedia.marginalia.nu/ |
| Git | https://git.marginalia.nu/ |
Did some local LLM-based labeling of the HN comment corpus, tasking the model to classify how likely it is that each comment is:
* Using AI-like grammar (e.g. it's not X, it's Y)
* Using AI-like markup (e.g. em-dashes)
* Trying to shill something
* Trying to influence public opinion
Still a bit work in progress, but here are some preview data.
Did some statistics, and it seems posts from newly registered accounts on HN are nearly 10x more likely to use EM-dashes, arrows, and similar typography than established accounts.
Started putting some more information in the site viewer tool. Needs more polish, but it's accessible now.
I've found the best way of ensuring data quality is to make the data visible. More eyeballs = more bug reports = fewer bugs.
New blog post, about index compression, and taming tail latencies.
Turns out the search engine wasn't correctly counting how many additional results are available per domain, showing a much smaller number than what was accurate.
Easy fix.
Published a short writeup about the new ranking changes from yesterday.
