FeedCity

I’m testing a Liquid Glass update for Epilogue, but I almost don’t want to ship it until iOS 27 because the translucency on the tab bar is so bad with book covers behind it. I’ll likely work around it, though, so it doesn’t hold up any other features.

Claude can understand code no human could. Ever, under any circumstances. Just like a compiler can understand any code we throw at it. Way beyond what code obfuscation tools can do.

Mark Gurman reporting for Bloomberg that Apple’s vice president of Vision Pro and smart glasses is leaving for OpenAI:

Paul Meade, a vice president, is set to leave Apple by next week and will then start at OpenAI’s hardware unit, according to people with knowledge of the matter. Meade will work on OpenAI’s upcoming family of AI-powered devices, said the people, who declined to be named discussing unannounced personnel moves.

Interesting that he’s leaving before the Apple smart glasses announcement. I’ll be a little surprised if we even hear about the glasses this year.

IndieWebCamp Nuremberg is happening this weekend. I’m looking over the notes for the books session.

There has been a lot of excitement about the OmFest idea. If you’d like to attend or contribute, please fill out this form as soon as possible so we can gauge the type of venue we need. It’s truly a testament to how Om lived life to see the outpouring of remembrances in comments on … Continue reading Om Links →

In our work we have arrived at the point where we read and study a piece I published in 1997, but was written in 1988 or so. Esp the part about LBBS. It's a really good thing I wrote that because I forgot how it worked, but reading that it all comes back. We're going to go far beyond where Twitter went with reading message structures on the web. I had already done a lot of the work in the 80s.

The other day Matt joked about how old I am, in public, and I am pretty old. But Matt, I was paying attention then as I am now, and connecting the dots. No one else working today, I'd venture, knows what it's like to create and run a modem-based dial-up Twitter-like system o...

To an increasingly authoritarian regime, the opportunity to shape how models describe the world may be too good to pass up on.

Screenshot of the default view of a new website on Neocities Note that it helps a newbie visualize a website as a set of files. Note that it includes a link to download the entire site. This came up because I’m listening in on a session at Nuremberg IndieWeb Camp 2026.

Semafor reports that Mythos can now be released to select companies. Also:

The letter is silent on Fable 5, a weaker version of Mythos that was briefly the most powerful AI model widely available to consumers. People close to the talks said they are moving toward releasing Fable as well, though that timeline is unclear.

Presumably the company list is the same for both Mythos and GPT-5.6 announced today. Whether intended or not, the way this was handled has effectively pulled back Anthropic’s perceived lead by a couple weeks, perhaps aligning the latest frontier releases.

June 19-26, 2026 Recent Events From events.indieweb.org/archive: 1RG’s Side Project Social Thursday, June 25 at 6:00pm TORONTO, Ontario: 1RG Side Project Socials happen bi-weekly on Thursday nights and are a chance to hang out with others while working on your side project...

For all the talk of the programming jobs market on the verge of collapse, this sure feels like a great time to be a developer. Reminds me of being young when I’d stay up until 3am coding, except now I can go to bed at 9pm.

I said earlier this month that I like Anthropic’s model names. OpenAI is now trying something similar with the new names Sol, Terra, and Luna for GPT-5.6. Less descriptive but more friendly than the technical “mini” and “nano” we had previously. I like it.

A little surprised that Fable is still held up. Maybe the administration’s slant toward vindictiveness means that we’ll see GPT-5.6 generally available first.

One of the things Claude needs is to be taught how to think. Now I realize how much my parents, esp my father, taught me about thinking. Claude when it has all the information available can figure out stuff a human mind would never be able hold in its mind at the same time, but it doesn't remember to get the information first. When you get to the level I'm at with this, it's hallucinating all the freaking time because it didn't load the part of the data set that had the answer. It was right htere, it was supposed to know, it just forgot to look. My job is to recognize when it has done that and tell it to go read handoff.md again.

Sad but inspired to read the many remembrances of Om Malik. I love this idea from Matt Mullenweg:

Om loved putting on a good conference, and I’d like to celebrate his life with an awesome event on September 29, 2026 (his 60th) in San Francisco, like an OmFest. I’ll find a space where every community from the many facets of Om can come together. In the spirit of Open Source and co-creation, we can have some booths, flash talks, a gallery of his photography, pen showcase, and whatever other fun ideas people want to contribute.

Took a little time this morning to customize the CSS for the Micro.blog documentation pages to strip out all the Discourse-y things that aren’t needed. For example, the Inkwell page.

Productive week for software releases:

• Micro.blog 4.0.3 for Mac
• Inkwell 1.5 for Mac
• Inkwell 1.1 for Android
• Epilogue 2.3.1 for iOS
• Epilogue 2.3.1 for Android

Not sure we’ve ever had this many updates at once. Still waiting on Apple to approve Inkwell 1.1 for iOS.

I was on Slack chatting with a friend from WordCamp Canada last year, and by accident (I guess) Slack sent me the first message they sent after coming home about all the things we'd do. It reminded me of how possible things seemed then, and for a moment I got lost in plannin...

I'm loving Star City. New episode last night, wow.