Subscribe
IndieWeb
Happy to get an Inkwell update out for macOS. From the news blog today:
Fixed several bugs in Inkwell for Mac. Grab the latest version with “Check for Updates” or directly from the help page.
Micro.blog News https://news.micro.blog/2026/07/12/fixed-several-bugs-in-inkwell.html
Robin Sloan blogged about attending a Stop the AI Race march:
These things are growing and mutating faster than anybody can make sense of them. So … here’s a wild thought … let’s slow down, and make sense of them.
I really like Robin’s blog and his books. I don’t think a pause can work, though, because coordinating with China labs and all open source seems nearly impossible. I’d rather see better plans for investing in renewal energy and some kind of UBI-like fund to help with the economy. No matter the speed of AI development, slow or fast, we’re going to need those things.
Something fascinating to me about how relatively little we use AI, compared to what it can do… I don’t spend much on OpenAI’s API, for example, but see this graph of daily costs. What if AI was much faster and much cheaper, what would be possible?
Doing a little inventory of my Mac’s login items in System Settings. There is a bunch of junk in there that can be turned off. I miss the much simpler classic Mac “Startup Items” folder with aliases.
JF Martin on his Numeric Citizen microblog:
…this new saga between Apple and OpenAI represents what I dislike about the tech world: it’s just noise that distracts me from more fundamental and interesting things. Apple is very fond of these legal battles, and I find it all tiresome.
I’ve certainly been frustrated with Lindsey Graham over the years, the Brett Kavanaugh hearing most recently. But credit to him for working in Ukraine literally on Friday… then the next day he’s dead. Stunning. 🇺🇸
Watched the rest of Star City season 1. I liked it a lot. Has a science fiction-y feel even though it’s in the past. Also finished For All Mankind season 5, though I sort of lost interest in the characters. 📺
It’s a good morning so far. Running db migrations and adding more plumbing to expand our books database with author pages and genres. Will roll out today, but some of the data will fill in more slowly.
I’ve been consistent in my dislike of non-competes. Vincent built his own blog hosting that competed with Micro.blog while I was paying him as a contractor! It’s fine. It makes the web better. As I told him, if I’m not paying a full-time salary, I don’t have the right to tell him what to work on.
The more I read of the Apple filing last night, the less sympathetic I was to Apple. Theft of “trade secrets” can mean almost anything, from the serious to the silly. Information is not the same thing as real property. Big companies use contracts and lawyers to protect themselves from competition.
A couple of the claims in the Apple vs. OpenAI lawsuit sound kind of serious, but a lot of it sounds like Apple wishes they could wipe the minds of departing employees. The only way to achieve true Apple product secrecy is Severance.
Toni Schneider blogs an update on Bluesky and drops the “interim” from his CEO title:
Since March, we’ve shipped a slew of new features and products, another 2 million people have joined Bluesky, and close to 200 new apps have launched in the Atmosphere, the open ecosystem of social spaces connected by a single account. All of this tells us that the appetite for social experiences on the internet is vast; far broader than the handful of apps the major platforms offer today.
Dave Winer announces rss.chat, a new Twitter-like service based on RSS:
We’re just getting started. This is Day 0 in a story that could last a while and spread out pretty far.
I’ve been poking around in the feeds to look for opportunities to integrate with Micro.blog. While it’s probably coincidence that TechCrunch featured HyperTexting today and Dave announced this, I think it’s significant as they are both different angles on “just use web stuff” for social networks.
Great to see this write-up of HyperTexting on TechCrunch! I feel like I’m indirectly featured because I’m in one of the screenshots. 🙂
Good discussion on Dithering today about the updated ChatGPT for Mac. I hope a few folks at OpenAI listen to this. As @gruber says, it’s all fairly obvious stuff… The UI is fixable.
Hmm. Had to verify my age before using Bluesky from the web this morning. Good news is the verification process was fast.
I thought the Texas law was to put the burden on the App Store and Google Play to centralize age verification. A little confused now about the scope.
Sometimes even with the robots help, you take on more than you can manage. GPT-5.6 is great. Shipping the Mac app, Sites feature, and Work all at the same time was too much.
Sites is amazing… from the web. Chat on the Mac needs a redesign. Expecting lots of app updates until they get this right.
Shout-out to Google for the amazingly quick app approval times. Minutes. I assume it’s mostly automated… Which it should be for apps that have been shipping updates for years without malware or gimmicks.
Added a new book calendar page to my blog, to visualize when I finished reading books over the last couple of years. Planning to make this available as an option to anyone on Micro.blog. First I need to blog more about what is going on behind the scenes for this feature.
ChatGPT’s update for Work and Codex puts the design in a precarious position. There is so much going on. If it was me, I would take a step back and think about how to keep a simple chat interface that scales well to Codex without immediately feeling overwhelming.
In addition to launching GPT-5.6, they’ve combined the ChatGPT and Codex Mac apps. I guess that was bound to happen. It still needs a little more work to feel right for normal, non-coding chats.
Even with AI being able to clone almost any app, Micro.blog is fairly safe because no single person actually knows everything the platform can do. 🤪
“I’ve got one idea that I’m especially psyched out of my mind about. You know, it’s one of those ideas where you’re just like… Yes!” — Miles Finch
There are really two very different ways of coding with AI. One is to quickly iterate with small changes. The other is letting an agent go off and work on a hard problem or high-level goal, where it churns for minutes or hours. For the first, model speed is almost everything.
Thinking about the impact of GPT-Live announced today. Two things…
First, Apple farming out Siri’s foundation to Google puts them in a weird in-between state where maybe they can’t technically build something like this, but also they can’t just use someone else’s model. As if they’re trapped.
And second, I’m sure GPT-Live was architected for the Jony Ive device. I’ve had a hunch for a while that it could be something special. No one has cracked the AI voice assistant device / wearable because for it to be great you have to start by rebuilding everything. GPT-Live is like the first generation iPhone’s multitouch.
Figured out how to publish the AT Proto schema for blog.micro.content. Neat. The schema is just a PDS record, with a DNS entry to resolve the name.
Added a page to document some of how Micro.blog works with Bluesky, including Standard.site compatibility. I’ll add more to this as I think of implementation details that are useful to know.
Cleaned up a few things in the web interface for books this week, including a new popup menu for switching bookshelves. Here’s a quick 12-second screencast video.
