Context plumbing
Context appears at disparate sources, by user activity or changes in the user’s environment: what they’re working on changes, emails appear, documents are edited, it’s no longer sunny outside, the available tools have been updated.
This context is not always where the AI runs (and the AI runs as closer as possible to the point of user intent).
So the job of making an agent run really well is to move the context to where it needs to be. [...]
So I’ve been thinking of AI system technical architecture as plumbing the sources and sinks of context.
Tags: definitions, matt-webb, ai, generative-ai, llms, ai-agents, context-engineering
Quoting Wikipedia content guideline
Large language models (LLMs) can be useful tools, but they are not good at creating entirely new Wikipedia articles. Large language models should not be used to generate new Wikipedia articles from scratch.
— Wikipedia content guideline, promoted to a guideline on 24th November 2025
A ChatGPT prompt equals about 5.1 seconds of Netflix
Bluesky Thread Viewer thread by @simonwillison.net
Bluesky Thread Viewer thread by @simonwillison.net
I've been having a lot of fun hacking on my Bluesky Thread Viewer JavaScript tool with Claude Code recently. Here it renders a thread (complete with demo video) talking about the latest improvements to the tool itself.
I've been mostly vibe-coding this thing since April, now spanning 15 commits with contributions from ChatGPT, Claude, Claude Code for Web and Claude Code on my laptop. Each of those commits links to the transcript that created the changes in the commit.
Bluesky is a lot of fun to build tools like this against because the API supports CORS (so you can talk to it from an HTML+JavaScript page hosted anywhere) and doesn't require authentication.
Tags: projects, tools, ai, generative-ai, llms, cors, bluesky, vibe-coding, coding-agents, claude-code
Bridge Clearance
Macaulay Culkin Watches Home Alone With His Kids Who Don’t Realize He’s Kevin
Rebecca Solnit: A Year on From Trump’s Victory, Resistance Is Everywhere. “There...
Rebecca Solnit: A Year on From Trump’s Victory, Resistance Is Everywhere. “There has, in fact, been a tremendous amount and variety of resistance and opposition and it’s mattered tremendously.”
Atul Gawande of Harvard’s School of Public Health: The Trump regime’s destruction...
Atul Gawande of Harvard’s School of Public Health: The Trump regime’s destruction of USAID “has already caused the deaths of six hundred thousand people, two-thirds of them children”.
A long profile of comics legend Alan Moore. “For the first time...
I’m still working on the KDO gift guide for this year. In...
I’m still working on the KDO gift guide for this year. In the meantime, here’s the 2024 edition, which has aged well and includes these popular Japanese nail clippers.
“Australia is on course to meet a target of eliminating cervical cancer...
“Australia is on course to meet a target of eliminating cervical cancer by 2035, which, if achieved, will make it the first country to do so.” This is due in large part to their HPV vaccination campaign.
Radiolab
• WNYC Studios
Fela Kuti: Enter the Shrine
‘Fifteen Years’
A masterpiece from Randall Munroe, perfect for Thanksgiving.
Link: xkcd.com/3172/
Quoting Qwen3-VL Technical Report
What Are You Thankful For?
It’s been a really tough year for many of us, the citizens of the world. But on this day of thanksgiving in the US, I wanted to ask you all: what are you thankful for today?
I’m thankful that I was able to travel to say goodbye to a friend, thankful for the time I’ve spent with my kids over their holiday break, and thankful for all of you, especially those who support the site with a membership, a small front in what feels like the final stand of the open web. Thank you.
Tags: holidays
deepseek-ai/DeepSeek-Math-V2
We now have an open weights (Apache 2 licensed) 685B, 689GB model that can achieve the same. From the accompanying paper:
DeepSeekMath-V2 demonstrates strong performance on competition mathematics. With scaled test-time compute, it achieved gold-medal scores in high-school competitions including IMO 2025 and CMO 2024, and a near-perfect score on the undergraduate Putnam 2024 competition.
Tags: mathematics, ai, generative-ai, llms, llm-reasoning, deepseek, llm-release, ai-in-china
I just added “Polaroid” to the list of film formats to my focal length equivalents site.
David Lerner, Co-Founder of Tekserve, Dies at 72
nytimes.com/2025/11/26/technology/personaltech/david-lerner-dead.html
Sam Roberts, reporting for The New York Times:
David Lerner, a high school dropout and self-taught computer geek whose funky foothold in New York’s Flatiron district, Tekserve, was for decades a beloved discount mecca for Apple customers desperate to retrieve lost data and repair frozen hard drives, died on Nov. 12 at a hospital in Manhattan. He was 72. [...]
Tekserve specialized in finding the cures for sick computers — including insect infestations — and recovering first novels and other priceless data, which the company said it was able to do about 85 percent of the time.
“We only charged for success,” Mr. Lerner said.
There were many great independent Apple resellers from the pre-Apple-Store era. There was only one that was legendary: Tekserve.
Link: nytimes.com/2025/11/26/technology/personaltech/david-lerner…
Running to the Press
daringfireball.net/linked/2025/11/26/november-update-to-the-app-store-review-guidelines
Regarding my earlier post on similarities between the 2010 App Store Guidelines and today’s: Notably absent from the current guidelines (I think for a very long time) is the specious but very Jobsian claim that “If you run to the press and trash us, it never helps.” Getting the press on your side is one of the best ways for a developer to get an unjust App Store review decision overturned. Apple loathes negative publicity.
Link: daringfireball.net/linked/2025/11/26/november-update-to-the…