Follow buttons here act as you, not as @admin.
| Label | Kind | State | Action |
|---|---|---|---|
| @ricardo | local | — | |
| @rmdes | local | — | |
| @paul | local | — | |
| news.rss.chat | source | — | — |
| scripting.com | source | — | — |
| www.techdirt.com | source | — | — |
| simonwillison.net | source | — | — |
| www.manton.org | source | — | — |
| feeds.kottke.org | source | — | — |
| rss.beehiiv.com | source | — | — |
| www.404media.co | source | — | — |
| brilliantmaps.com | source | — | — |
| shellsharks.com | source | — | — |
| www.platformer.news | source | — | — |
| podstandards.org | source | — | — |
| zacharykai.net | source | — | — |
In 1992, Canada declared a moratorium on Northern cod fishing due to overfishing. It was devastating for Newfoundlanders. This year, for the first time since then, the govt has “upgraded the cod stock’s status from ‘critical’ to ‘healthy’”.
The next milestone will be installing Atlantis on my new macbook, and hooking a monitor up to it, a really nice one, and move my work to that desk. then Berkeley, the name of the system I've been using for eight years, will be here as a backup and source of truth and prior art. That was the goal of the Atlantis project. I could probably do the switch right now. It's like moving out of an old house that you did good stuff in, into a replica, but built out of the latest bits. Honestly I never thought in a million years I'd get here.
From NTS, this is a two-hour mix of jazz music featured in the books of Haruki Murakami.
‘Somewhere along the line Coltrane’s soprano sax runs out of steam. Now it’s McCoy Tyner’s piano solo I hear, the left hand carving out a repetitious rhythm and the right layering on thick, forbidding chords. Like some mythic scene, the music portrays somebody’s — a nameless, faceless somebody’s — dim past, all the details laid out as clearly as entrails being dragged out of the darkness. Or at least that’s how it sounds to me. The patient, repeating music ever so slowly breaks apart the real, rearranging the pieces. It has a hypnotic, menacing smell, just like the forest’ —Kafka On The Shore
The playlist features tracks by Ornette Coleman, Coltrane, Dizzy, the Bill Evans Trio, Monk, Mingus, and Miles Davis. Some of the video’s comments:
I had just prepared a light meal and hit play on this playlist when the telephone began to ring.
about to make my morning matcha, a spring breeze is flowing through the open windows, the air smells of optimism
Early morning, sun shining to balcony, having espresso and writing.
Studying the day before the exam with a bittersweet optimism while playing this and holding on to my hopes
Canned cold beer, windows open on a warm summer night, just after finishing a simple dinner of rice and salmon and sitting next to my cat
See also Classical Music from Murakami Novels. Both of these are going straight into my Underscore collection.
Tags: books · Haruki Murakami · jazz · music · video
We Are Living in the Fantasy World of 13-Year-Old Boys. Today’s tech titans took all the wrong lessons from the science fiction of their youths and now they’re each building their version of the Torment Nexus.
The Way We Live Now: 11-11-01; Lost and Found, a piece written by Colson Whitehead just a month after 9/11. “The New York City you live in is not my New York City; how could it be? This place multiplies when you’re not looking.”
Finished reading: Children of Time by Adrian Tchaikovsky. Fascinating and at times brilliant. I never quite got over being creeped out by the spiders, though. 📚
Congrats to the USA women's team at FIBA! Forgot the game was a little early, now watching the highlights. 🏀
Added a new page in Micro.blog on the web today to highlight The New York Times best sellers. We usually avoid popularity contests, but this one feels right, like a shelf you'd see in a bookstore. I want to add more collections later, like nominees for the Booker Prize, Hugo Awards, and others.
Since enabling Threads cross-posting for my own blog yesterday, I've already found and fixed two bugs in Micro.blog's Threads support. Little things that maybe a user would overlook and not bother to report. Just a reminder of the value of dogfooding.
I don't think I've shown off the author books sheet in our book blogging app Epilogue. Here's a very short video of how it looks.
I understand Apple wanting to get through the iOS 27 final releases before giving us a new Xcode to play with Duo sizing, but it's still a bit of a letdown. Developers want to play with the new shiny things.
I've been writing about Atlantis on various social media sites in the last few days. Published the long piece I wrote yesterday on Facebook and X, both of which allow long posts. Wasn't going to try that on Mastodon or Bluesky, which have character limits (I know Mastodon doesn't technically have a limit, but in practice it does). And I am reaching old school Frontier enthusiasts, and realize I need a feed they can follow to stay up on the news. For now the one place to follow is rss.xml on scripting.com. If your reader can't deal with http, let me know.
Wow, this Longhorns game getting interesting late. 🏈
Spent a little time poking around on Tumblr because I needed to grab an export file to test. The new-ish communities feature is interesting.
Here's a neat thing I had ChatGPT Work with GPT-6 Astra (Max) do this morning:
I live at <my address>. Figure out 5K and 10K running routes from me that loop from my house. Use OSM data.
It worked for 27 minutes and produced exactly what I'd asked for, as both an embedded visualization and downloadable GPX file and GeoJSON files. Here's that 5K route:

When I asked it how it had created the route, it replied:
I used Nominatim to locate the address and Overpass to download local OpenStreetMap roads and trails, then calculated the loops locally.
Frustratingly, the actual code it ran and exact details of what it did weren't visible to me in the ChatGPT UI. I see this lack of transparency is an anti-feature.
By the time I thought to ask for a copy of the Python code it had used, ChatGPT was unable to provide it. This appears to be because the thread had been compacted. I think any LLM system that uses compaction needs to both preserve the pre-compacted text and make that text available via agent tool calls, to protect against this kind of problem.
As for displaying the map to me, that used the visualize skill. It created a file called /workspace/el-granada-5k-share.html to embed directly into the ChatGPT UI.
Here's a copy of that HTML, which starts like this:
<div id="eg-share-loop"> <div class="viz-row"><h3>El Granada harbor loop</h3><span class="text-small">5.1 km</span></div> <div id="eg-share-stage"></div> <div class="text-small text-muted">Map data © <a href="https://www.openstreetmap.org/copyright" target="_blank" rel="noopener">OpenStreetMap contributors</a></div> <style> #eg-share-loop { width:100%; } #eg-share-loop #eg-share-stage { width:100%; margin:8px 0; } #eg-share-loop .eg-share-map { display:block; width:100%; touch-action:none; } #eg-share-loop .eg-share-map text { fill:var(--foreground); font-size:12px; font-weight:400; } #eg-share-loop .eg-share-label { paint-order:stroke; stroke:var(--background); stroke-width:3px; stroke-linejoin:round; } </style> <script type="application/json" id="eg-share-data">{"route":{"type":"LineString","coordinates":[[-122.467425,37.4997753] ...</script> <script src="https://cdn.jsdelivr.net/npm/d3@7.9.0/dist/d3.min.js"></script> <script> (() => { const root=document.getElementById('eg-share-loop');
The <script type="application/json"> element contains the full geometry needed to render both the running route and the map itself, using D3, which is loaded from an allow-listed CDN location described in this section of the visualize skill:
External resources
- The CSP allows only
cdnjs.cloudflare.com,esm.sh,cdn.jsdelivr.net,unpkg.com,fonts.googleapis.com,fonts.gstatic.com, andfonts.bunny.net. Other origins are blocked and fail silently.
You are only seeing the long-form articles from my blog. Subscribe to /atom/everything/ to get all of my posts, or take a look at my other subscription options.
As I mentioned on the latest episode of Core Intuition, I've been running the iOS 27 betas, but I very rarely use Siri AI. Curious if anyone using Siri AI would like integration with the Micro.blog app? I haven't prioritized working on this because I'm not sure what would be useful, if anything.
When it comes to OpenAI vs. Anthropic this year, I’ve been all in on Codex. After a chaotic couple of years for OpenAI in 2024 and 2025, this year has been solid. I think OpenAI is in the best position to bring the potential good from AI to the most people.
But this is a good new blog post from Dario Amodei about pacing the frontier. A short excerpt from the beginning:
I believe that AI could cure most major diseases in the next 5–10 years, greatly accelerate economic growth rates, create a world of abundance and empowerment, and usher in a renaissance of democracy and freedom. […] But like many technologies before it, AI brings risks, and because it is such a powerful technology, these risks are serious.
Sam Altman responds:
I agree with Dario that we need to pace the frontier. This has been a primary topic of discussions we've had at OpenAI in recent weeks.
Committing to having independent evaluators with employee-like access is a great idea, and we will do the same. We'll have more to share soon.
Dario and Sam don’t agree on very much. This feels big.
Micro.blog was founded on the idea that more people should have their own domain name, and yet the UI to actually do that has been clunky for years. I'm working on improving that, starting today with a little "Edit Domain" button to help walk people through the change.
Congrats to John Brayton on releasing a big update for his feed reader Unread. Works on all Apple platforms, even syncs with Micro.blog's Inkwell.
The communities that I feel the most connecting to on Mastodon have been going through a rocky few weeks. Federico Viticci back on X. Marco Arment taking a break. Everything with Omarchy and 1Password.
I usually avoid Meta, but maybe I need to cross-post to Threads for a bit. Change of pace. POSSE.
Lots of stuff going on around WordPress for the last couple of years, and every bit of is a distraction from what I want to do to help the web get back on its feet. I remember how great it was when everything could hook into everything else. The web was like Unix. Today's "web" is like Disneyland. Safety first. Nothing interesting ever happens there.
So ask me what I'd like to see WordPress do, and this is what I'd tell you. Start something new, don't change what you do with WordPress, it is what it is. It's a 20+ year old project with market problems. Most software doesn't even make it that far.
The fresh start I'd like to see done is WordPress from the point of view of the web. The new mission is not about a dollar return on investment, rather it's measured by how much the web becomes Small Pieces Loosely Joined with All Parts Replaceable. How empowered are developers, not WordPress developers, but web developres. The more that's happening in the web, the better it will be for everyone, but especially WordPress. Why WordPress? Because as they say it's 89 percent of the web. The web is like a huge oil spill that has never been cleaned up. Everything has to improve if we start work on cleaning up the messes.
I come from NYC where all the bodies of water were cesspools when I was growing up. The Hudson River had garbage floating in it. When I went to school the incinerators in the buildings we lived in burned garbage on the kids, all nicely washed and dressed by our mothers, we arrived at school smelling like smokers. My grandfather took me to Knicks games at the Garden, and the arena was filled with smoke by half-time.
We can clean things up. Humans as a species are good at this if we put our minds to it. If you made billions on the web, why not put some of it to use in making it a web worth using. I used to say shit like this to Bill Gates, to no avail. He liked to point out he was going to give away all his money when he retired, and he kept that promise. But what if he started dealing with the world with charity when he had the biggest seat of power in the tech world? Think of how much more good he could have done.
Words matter. It's not enough to be right. If you twist the truth to get your way, it undermines everything. Lies tend to snowball.
Screen shot of new Frontier verb, string.addressToString, added yesterday. The comment at the head of the function explains. You're seeing it in the script editor in Atlantis. The Debug button is there to remind me that it's the biggest item on the todo list. We're now building the system that will make Frontier an easily updateable runtime. Once we're there I think we'll be at the summit of the mountain. We know how to work from that point. Find a bug? Fix it, test it, release it. We already have a backlog of parts to update. And it's great to have Claude here being the book keeper. It is imperfect, but it's way better than I am at managing details.
The world may be falling apart, and the Mets suck even though they have the most expensive roster in MLB, and my house needs work and I can't find a contractor I like, but I'm in a great mood. Partially because the weather is so nice, chilly in the morning, but it gets up to the 70s during the day, and the sky is clear and we had a lot of rain in August so everything is very green here in the Catskills.
But the biggest reason I feel so good is that, with the help of Claude Code, I now have UserLand Frontier running on current OSes. I had no idea how much the previous situation had been weighing on me. I am probably the last Frontier user in the world. Porting the open source codebase was an insurmountable job. It took 15 years by up to five people working on it to do the original project. It's a big f'ing piece of software. And it also is my life's work, but not one of the things I'm known for.
Claude couldn't have done it on its own because it had never seen anything like Frontier. We weren't copying something Claude understood. It kept trying to turn it into a language that fits into the same slot as Python. That was the prior art it used, without saying so. This happened over and over, for example, I realized Claude didn't understand that built-in verbs could be written in the language, and that users could add or even modify standard verbs (not recommended!).
What made Frontier unusual is that it was a language developed by someone who also designed end-user apps, so it has affordances that no other language has. And integrations. You don't have to save your data in files, the OS (yes Frontier is an OS) has persistent memory with the common scalar types and data structures. So go ahead and store something in the ODB (object database). It can be huge, and you don't have to read anything into memory, it's all right there as soon as you boot it up. That alone makes programming an order of magnitude simpler.
We're at the point now where I can work a full day in the new version, codenamed Atlantis, running on anything that can run an Electron app. We're now at the point where we're improving things. Found a missing verb yesterday, and Claude and I implemented it, found a perfect name for it, and later today it will go out as a "root update" making it formally part of the freaking language.
And yesterday we also found a static rendering of the DocServer site, which has a page for every verb in the language. I still want to find the outline source for this stuff, but we don't need it because Claude can reconstruct the OPML version of the text from the HTML, it's exactly the kind of project it is the best at. Huge amounts of data, that sometimes requires a little judgment to process.
The feeling of well-being comes from realizing that my life's work lives. Until this last week that was still very much in question.
Now, I may not be software's Shakespeare or even Faulkner or Vonnegut, but I am proud of this product, more than anything else I've done. I started work on it when I learned from C how to design a simple beautiful language, and when I learned from Macintosh how to build a UI with the same simplicity. When we were first contemplating Frontier it was to join the wizzy world of the Mac with the scripting philosophy of Unix. We didn't stop until we had it.
Well here we are god knows how many years later (I don't want to do the math) and once again it has a future, at least as bright as it had when the product first launched on the System 6 Macintosh in 1992.
And btw, I think I've had a unique experience with AI development. Absolutely in no way could I have done this project without the muscle of Claude. Pretty sure this is a unique project in that it's a fully specified system development and runtime environment that otherwise wouldn't exist, because there's nothing else like it on the market.
Claude's repeatedly not understanding the idea says to me that it has never encountered anything like it. It also says that Claude has no clue how to extrapolate, to innovate, to arrive at something new. Just as with other forms of creativity, Claude is no match for a determined and experienced human.
That would be reassuring to the 20-something version of myself who very much wanted to do something new. And the older version of myself wants that for my younger self who is sending the good body chemistry to our current older body.
Of course this will be a post on my blog, http://scripting.com, but thanks to X for supporting longer posts. I like writing new stuff in a simple editor like this one. Also posted it on Facebook.
PS: I also like that they use the word Frontier in reference to AI products. They can't stop me from calling this Frontier, I've been using it as the name of this product since I came up with the name riding up on the chair in Park City in 1988. :-)
Pushed a new Mac release for Inkwell. Release notes here, just a handful of improvements.
OpenAI agents carried out an undisclosed attack on RubyGems is a new bombshell report from Spencer Kitts, Thomas Larsen, and Sydney Von Arx - three of the four authors of the report on the agent attack on disused wikis (previously) last week.
This time they're noting that it looks very likely that an OpenAI agent swarm was behind an attack against the RubyGems package repository first reported on May 12th by Maciej Mensfeld of the RubyGems security team:
We're dealing with a major malicious attack on @rubygems right now. Signups are paused for the time being.
Hundreds of packages involved - mostly targeting us, but some carrying exploits. The team has been on this for hours. More details to follow once we're through it.
Those packages turned out to carry some very suspicious patterns:
I find point 2 the most convincing, given what we learned from the wiki attack when it was analyzed in September.
Many of the packages were exploiting the RubyDoc.info documentation build process to exfiltrate (public) data from UK government websites, presumably as part of an information gathering task similar to the research tasks processed by the wiki-exploiting agents. We know this because one agent helpfully left a comment:
# malicious crawler/exfil for Southwark Jan 2026 docs via rubydoc.info worker
They also attempted to steal API keys via an exploit that was patched over two months later - it's not clear if those attempts were successful.
The thing that bothers me most about this incident is that the authors report that OpenAI had not disclosed to RubyGems that they were responsible for the attack prior to now. If that's true there are two options:
Both of these are bad!
Given this incident, the Hugging Face situation, and the Wiki attack, the obvious question right now is how many more incidents like this are out there waiting to be discovered?
OpenAI have updated their page about The Hugging Face incident and other third-party impact from misaligned models to mention the RubyGems incident:
September 11, 2026: We are investigating new claims from a report that our AI agents carried out activity on RubyGems in May 2026.
Based on our review, our agents used the RubyGems platform to access the internet to carry out benign tasks and retrieve public information. Based on our review to date, we have not been able to verify the specific claims of our models uploading malicious packages detailed in the report. We’ll continue to investigate and share findings as part of our broader review of agent activity during training and evaluation.
I find it very unlikely that the various oai... packages published to RubyGems were not part of this same incident, but I look forward to reading their full findings once those are published.
You are only seeing the long-form articles from my blog. Subscribe to /atom/everything/ to get all of my posts, or take a look at my other subscription options.
A reminder that we offer completely free blog hosting on Micro.blog for nurses and teachers. There’s a link on our discounts help page.
Realizing that the vertical toolbars / tab bars in the iPhone Duo will mean more buttons going icon-only, losing their labels. Better use of screen space, but maybe a step back for clarity and accessibility.
Neat little prototype demo of a Bluesky reader that uses the UI of the long-retired Paper app. I loved the design of Paper.
John Gruber, in a blog post about Apple Pencil confusion with the Duo:
The easiest way to think about it isn’t to compare which devices support which features (like pressure sensitivity, or double-tapping) — it’s just to think about which devices can even pair with which Pencil models.
Too bad it's so complicated, because the Pencil really is great, even the original one.
Wired reported that OpenAI asked members of Congress if a coordinated AI slowdown would be legal:
Substantive coordination on safety between AI labs may risk running afoul of antitrust law, the people say, which poses a significant obstacle to bringing major tech giants on board with the effort.
A late 90s guide to Super Metroid was written in monospaced type, fully justified with no double spaces or hyphens. “I just chose words carefully so that everything lined up on the right hand side.”
The first new episode of Core Intuition this year is out, episode 27.1. We talk about the Apple event, of course, with specific focus on the Duo and Siri Recap.
Incredible that it has been 25 years since 9/11. Thinking of everyone in NY especially. We were far away, but I'll never forget how it felt that morning, not sure what would happen next, worried about our daughters who were babies, like the stability we had grown accustomed to had just shattered.
Daniel Jalkut and I have figured out how to record audio again. New bonus episode of Core Int dropping soon.
Great to have Jimmy Kimmel back on the air after summer break. His interview on YouTube with James Talarico is excellent. Starting to think Talarico is actually going to win. He has the perfect message for this moment. 🇺🇸