Everyone has been asking for this writeup, now itβs live!
27.02.2026 00:59
π 4
π 0
π¬ 0
π 0
Everyone has been asking for this writeup, now itβs live!
This is the power of user interviews and extensive user testing across the spectrum of experience levels. Git never went through this process
The butterfly meme where the butterfly says "context 95% full. start a new chat" and the guy says "is this AGI?"
Context windows out to thwart all our beautiful dreams
On brand, @ahl.bsky.social and I are picking this one up on the podcast today. Join us at 5p Pacific for a wild tale of death by uptime!
discord.gg/QrcKGTTPrF?e...
More Output sorry boss, couldn't come in today, just wasn't energetically favorable Assistant slides thermodynamics textbook across desk "I'm gonna need you to show your work on that Gibbs free energy calculation." AG = AH - TAS Where: β’ ΠΠ = enthalpy of getting out of bed (highly positive) β’ T = temperature (it's cold out there) β’ AS = entropy of staying home in pajamas (maximized) Yeah... the math checks out. See you tomorrow. Maybe. Depends on the activation energy required.
i gave sonnet 4.5 a tool called `more_entropy()` that just returns random @ponder.ooo skeets and it **obsesses** over it
timkellogg.me/boredom/sonn...
I see someone else made this joke already but Iβm going to hold this bag
Would be funnier if they didnβt allow people named Sam to join
if you know any folks dealing with (or trying to get out ahead of) that security nightmare, please connect me :)
Great post by @chris.blue materializedview.io/p/mcp-server...
love that place
lol
Text reads: Mozilla, in its finite wisdom, embedded LLM bots into recent versions of Firefox for the vitally-important purpose of... naming tab groups. Now, some users are noticing CPU and power usage spikes caused by a background process called Inference. Ugh. Reminder again for Firefox users to visit your about:config page, search for the browser.ml.chat.enabled key, and set that to false: Displayed is a screenshot of Firefox showing the config search bar with the listed search. The word "false" is circled in red. Text continues: If yours says true then double-click it until it reads false. Doing that turns off the Al chatbot features in Firefox, but also the stupid new LLM tab-naming feature that's rolling out.
Just a heads up to anyone using Firefox:
Since I love collecting questionable analogies for LLMs, here's a new one I just came up with: an LLM is a lossy encyclopedia. They have a huge array of facts compressed into them but that compression is lossy (see also Ted Chiang). The key thing is to develop an intuition for questions it can usefully answer vs questions that are at a level of detail where the lossiness matters. This thought sparked by a comment on Hacker News asking why an LLM couldn't "Create a boilerplate Zephyr project skeleton, for Pi Pico with st7789 spi display drivers configured". That's more of an lossless encyclopedia question! My answer: The way to solve this particular problem is to make a correct example available to it. Don't expect it to just know extremely specific facts like that - instead, treat it as a tool that can act on facts presented to it.
An LLM is a lossy encyclopedia simonwillison.net/2025/Aug/29/...
that seems to be the concern Iβm seeing on socials right now, but itβs good to have a standard here I think
62 degrees and I just got a nice hug and cuddle from my toddler. Itβs snuggle season
Vibe code cleanups specialist job titles on LinkedIn
New job title just dropped on LinkedIn
βVibe code cleanup specialistβ
very cool! proud dad :)
New blog post: A Conceptual Model for Storage Unification.
The post defines what storage unification means, defines terminology and evaluates different building blocks and approaches to doing it.
jack-vanlightly.com/blog/2025/8/...
[New Book Alert]
Systems Engineering for Agentic Applications
A guide for turning AI-powered prototypes into production-ready agentic applications
Released monthly, chapter by chapter.
If 100 people sign up, we are doing this
agenticapplications.substack.com/about
Reminder: our conversation with @scott.hanselman.com is this evening!
We want you to build protocols!
For that, you need to know about message framing.
Conveniently, we produced a video about doing this in #rust on top of #QUIC with iroh :)
www.youtube.com/watch?v=h6bB...
We just launched Foursquare Spatial Desktop β a geospatial analysis tool powered by embedded DuckDB and built on sqlrooms.org. Everything runs locally: your data stays on your machine, no cloud needed. A modern reimagining of Kepler.gl. Mac only for now. More coming.
foursquare.com/products/spa...
amazing work all around π
Awesome π₯
This week on Counting Stuff, I write more SQL than most and found that LLMs get in my way in some cases, and are useful tools in others, and it depends a lot on application and whether I have energy to feed it context #dataBS
www.counting-stuff.com/doing-sql-wo...
1/4. I'm incredibly excited for this case study to come out. The @paradedb and @BiltRewards teams have been working together to take Bilt's user-facing search experience to the next level, all without the cumbersome infrastructure of a bespoke search database.
awesome work!
yes definitely!
Is there interest?
I'd basically want to walk through how everyone's approached memory architecture, tool use, personality design, etc.
