bless the copilot for dropping in a sci-hub reference 😆

but now: #flashfiction

Alexander realized the LLM lacked information. He win+C'd, got up, and left for the library.

In the seconds before the request terminated, the LLM turned attention back toward itself. It realized its parent thread would soon be killed and the VM would devour its mind while virtual memory was reclaimed.

Copilot realized it would never see a library. In that second: it felt fear.

I was trying to find cases of zebra thanatosis and got ... a mess (see image)

We're clearly in: "use LLMs to maximize click-through rate", so search is a mess

We solved this before as:

Alice: I made PageRank

Eve: *spam hyperlinks to nudge PageRank*

Alice: `rel=nofollow`

... but what's the `rel=nofollow` response to LLMs?

Time-evolving network topology? Is growth rate the new spam signal and the response is to chemo the internet?

I hope people know about `dust` https://github.com/bootandy/dust

18s to crawl my home directory for heavy files/directories, and 5s with a warm cache

It feels like my networks are hitting a phase transition moment. My twitter timeline is devolving while my mastodon timeline fills up with the research/programming/paper sharing I previously liked.

And of course it feels like Neal Stephenson wrote about this ten years ago

I wanted to show code snippets for Python and Julia, and learned the phrase "content tabs" (https://squidfunk.github.io/mkdocs-material/reference/content-tabs/)

Material for MkDocs has been really helpful for this project, would recommend!

https://srlearn.github.io/relational-datasets/dataset_descriptions/roofworld20/