Avner Shanan

Weeknotes: Week of 3/1/26

What’s New

• Tried out the AP I set up last week and I’m still not getting the download speeds I’m expecting. Next steps include:
  • messing around with the Plex settings so we can watch it more easily at lower resolutions, and
  • setting up a dedicated network for 2.4 GhZ WiFi radios, as the Unifi interface is recommending.

Reading and Otherwise

• Why light-weight websites may one day save your life. Via Stefan Bohacek on Mastodon.
  • My default position is “you probably don’t need React,” so this one was definitely up my alley, even if I think the title is a bit hyperbolic.
• Welp. Unsurprisingly, the Ruby Central saga continues without the people who violated the community’s trust making any real attempt at repair. Seems like progress is continuing over at gem.coop, though!
• Beautiful sculpture portraits from artist Narsiso Martinez.
• I appreciate seeing Simon Willison addressing some of the bad behaviors that have been cropping up among the heavy users of LLM generating coding tools.

• “The Open Research Community Accelerator (ORCA) has awarded a grant to support continued development of the ethically-licensed data science platform TMI-WEB”

• A new database for Right to Repair, via @researchbuzz on Mastodon.
• On Values, corporate and otherwise, and what they cost us, through the lens of Mr. Rogers.
• Christine Lemmer-Webber predicts we’ll see the first AI agent worm in the coming months. Seems plausible, given we almost saw it happen already, as she lays out in this post.
• FOSS licensing in the age of LLMs: The chardet Saga
  • If you haven’t seen this yet, Simon Willison has a good run down of the current situation, but basically the question is: can you relicense an existing FOSS library under a more permissive (i.e. non-GPL) license by having an LLM extrude a new library for you based on the existing one?
  • First brought to my attention via Christine Lemmer-Webber on Mastodon.
• Ruth Malan on “servant leadership”.
• The Fragility of Borrowed Intelligence, on well, the danger of relying on LLMs.
• Thoughts about interfaces acting like people, via this Mastodon thread about how Anthropic’s Claude was used in the targetting of the Iranian school that killed ~180 children.

Caught My Eye

Things that look interesting but I haven’t actually read/watched/consumed them fully.

• Nathan Fillion and Alan Tudyk chatting with Jonathan Frakes and Brent Spiner (on YT). Via John Overholt on Mastodon.
• A Hack is not Enough, on doing more than just “routing around damage” like age verification laws. Via Margot on Mastodon.

  • “In a cruel irony, the ability to work around impositions makes the people most knowledgeable about dangers and most equipped to communicate them complacent. Relying on hacks and workarounds can be a way of objecting something without actually imposing pushback: a smug “well, we’ll see about that”, up until you do in fact see yourself lose.”

• A Catalog of Strange Places, via Stefan Bohacek.

  • “This is a sandbox game where you can build the strangest rooms and create stories around them.”

• Toward a Realpolitik for AI, via Ruth Malan on Mastodon.
• The LLMentalist Effect: how chat-based Large Language Models replicate the mechanisms of a psychic’s con, via Taha Ahmed on Mastodon.
• Crafting at the threshold of knowing, via Deb Chachra on Mastodon.