Listening to This Week in Startups interviewing Medium’s Tony Stubblebine. He makes the wild claim that ChatGPT’s obsession with the em dash is because it’s so commonly used in Medium posts! I’m having trouble accepting this theory, but it would be amazing if true.
Love this story and illustrations on animator James Baker’s blog, from a trip to China in the late 1980s.
Says something about Vimeo’s decline that I heard about them being acquired not from the tech news websites that I read all the time, but from Cartoon Brew in my RSS reader:
Vimeo, once the internet’s most prestigious stage for independent filmmakers and animators, is being acquired by Milan-based app developer Bending Spoons in a $1.38 billion all-cash deal.
There is a narrow space for an indie-focused, YouTube alternative. Hosting video is difficult.
It’s funny to me that Apple made the new battery pack exclusive to the iPhone Air. If I really wanted the thinnest possible phone, no way I’m putting a case or extra battery on it. Even the bumper seems like too much.
From the latest Fediverse Report, about Mastodon quote posts:
Mastodon’s concern regarding the potential for harm with dunking does need some context however, researcher Hilda Bastian has a highly detailed overview of over 30 studies on quote posts on Twitter and their impact. Bastian notes: “There’s conflicting evidence on whether QTs increase or decrease incivility, and whatever effect there is, it doesn’t seem to be major.”
Mastodon’s concern regarding the potential for harm with dunking does need some context however, researcher Hilda Bastian has a highly detailed overview of over 30 studies on quote posts on Twitter and their impact. Bastian notes: “There’s conflicting evidence on whether QTs increase or decrease incivility, and whatever effect there is, it doesn’t seem to be major.”
I think this dunking is real, but there are many other problems with social media that Mastodon does not attempt to address. Not sure quote posts are make or break.
OpenAI policy for teens
OpenAI published a pair of blog posts today about their next steps for teen safety, acknowledging the trade-off on privacy. Sam Altman:
First, we have to separate users who are under 18 from those who aren’t (ChatGPT is intended for people 13 and up). We’re building an age-prediction system to estimate age based on how people use ChatGPT. If there is doubt, we’ll play it safe and default to the under-18 experience.
First, we have to separate users who are under 18 from those who aren’t (ChatGPT is intended for people 13 and up). We’re building an age-prediction system to estimate age based on how people use ChatGPT. If there is doubt, we’ll play it safe and default to the under-18 experience.
The second post goes over plans for parental controls, including a new setting not previously announced:
Set blackout hours when a teen cannot use ChatGPT—a new control we’re adding.
I’m highlighting this because yesterday I linked to a New York law for kids that would ban notifications from being sent at night. Social algorithms, games, and now AI can be so addictive that they keep teens up late into the night. It makes sense to focus on this.
As for whether OpenAI can predict how old people are before requiring an ID, I asked ChatGPT if it could guess my age based on its memory of our past conversations. Here’s part of the answer:
You strike me as someone who’s been coding for quite a while (Ruby, Objective-C, deep infrastructure work, and long-term platform/community building). You also mention personal/family projects, travel, and reflections that suggest mid-career experience. My guess would be around your late 40s to early 50s.
Nailed it. I’m about to turn 50.
Thinking of Robert Redford, what a career. Sneakers is one of my favorite films, and arguably the best film about computers ever made. All the President’s Men is so good. Watched both of these countless times. (Also just learned his mother was born in Austin.) 🍿
Added a help page for upcoming Open Graph improvements in Micro.blog, including a template system to override the default styles. This is rolling out over the next few days. Very flexible, so hopefully plug-ins can be created for various styles.
Finished reading: Small Things Like These by Claire Keegan. A short book, but it’s something special. 📚
A heads-up for Micro.blog theme developers: I’m revamping our experimental Open Graph support with more advanced features. If your theme doesn’t have its own Open Graph image, I recommend adding the og:image
tag based on the page’s .Params.opengraph.image
, which we’ll fill in automatically.
Paul Frazee blogging on Leaflet about whether Bluesky’s AppView should be renamed to more clearly convey what it does:
At this point, it seems better to just call it an App and then explain that the data gets stored in the PDS, like a kind of universal cloud filesystem or datastore.
Fair intro in Jason Snell’s macOS 26 review:
macOS 26 Tahoe is two things at once: It’s the broadest and most productivity-focused update for macOS in years, while also taking collateral damage from Apple’s broader design ambitions on its other platforms.
While age-gating often creates new problems, I like the fallback in this New York law:
Under the proposed rules for New York’s SAFE For Kids Act, social platforms must serve unverified users or kids under 18 only chronological feeds or posts from people they follow, as well as ban notifications from 12AM to 6AM.
Waiting on hold. “There are currently 146 callers ahead of you…” Seriously?!
Looking through Federico Viticci’s iOS 26 review. He starts on a fairly positive note about Liquid Glass. Or at least, not a “the sky is falling” panic:
I can’t stress this enough: the first thing you need to understand about Liquid Glass is that it’s not a drastic, groundbreaking redesign that changes the look of your iPhone overnight, like iOS 7 did for millions of people in 2013.
Watching the Emmys. 📺
“…and laugh out loud comedies like The Bear.” 🤣
From Mastodon’s blog post introducing quote posts:
Quoting is a powerful tool, and like any tool, it can be misused. That’s why we’ve taken time to introduce quotes in a way that aligns with Mastodon values, focusing on safety and mental health – not just on engagement.
Micro.blog will have limited support for this after it’s rolled out on mastodon.social next week.
Nick Heer blogging about the win Automattic got this week in having some of the charges in the lawsuit from WP Engine dismissed:
Even if this case ends with a complete victory for Mullenweg and Automattic, his actions have shaken my support of — and faith in — the WordPress ecosystem.
While I do think there has been damage to the WordPress community, if Automattic wins most of the case, maybe settles the rest, I don’t know if there will be significant lasting damage. The views from WordCamp US this month didn’t look like a community in decline.
Working in Go this morning! Micro.blog obviously uses Hugo, written in Go, but we also have a little custom Go tool that runs behind the scenes for newsletters, with more to come. Nice to have a common template language.
I think some of the worst-case scenario AI worries are overblown, but I do like this book title: If Anyone Builds It, Everyone Dies.
Good feedback from folks on whether and how to adopt macOS Tahoe menu item icons. After experimenting, menus seem to look better with icons in most places so that major sections of menu items are aligned together. I’ll ship this in a few days.

I’m noticing some new faces and some returning customers. Means a lot to me. From Pedro:
I went elsewhere for a while. I tried other services, they are fine. But they don’t come close to what Micro.blog offer.
Matt Mullenweg blogging on the recently-announced RSL spec for describing how content is licensed for AI:
I have a lot of scars from the web standards wars, so I’m hesitant to dive back in, but this is from a lot of the early Web 2.0 people, as TechCrunch writes about.
For rail fans in the London area, Museum Open Depot days:
Discover rare road and rail vehicles spanning over 100 years, signs, ceramic tiles, original posters, ephemera, ticket machines, and more.
🚂
This post by Nathan Witkin on the case against social media is long but very good. I’m going to have to go through this again in detail.
OpenAI and Microsoft have finally reached an agreement. OpenAI’s Bret Taylor also has a blog post on it, a $50 million grant for other nonprofits, and the OpenAI nonprofit + public benefit structure:
This structure reaffirms that our core mission remains ensuring AGI benefits all of humanity. Our PBC charter and governance will establish that safety decisions must always be guided by this mission.
Great article from New_ Public on indie blogs and websites:
The focus isn’t on personal branding, growth or monetization, or “content” creation, but on freedom from those things. Instead of polished, 10-second snippets optimized for mass-appeal, engagement, and profit, these are largely slow-cooked projects made just for fun.
I’ve never had a coherent answer when asked how big Micro.blog is, so I sometimes stumble and misspeak. The total number of users is inflated with inactive users and spam accounts. Active users are different for weekly, monthly, longer. I care most about paid subscriptions and whether we’re growing.
In that moment just after recording a podcast where I’m second guessing everything I said. I have a lot of respect for people who stop and form their thoughts before answering a question. I tend to just wing it, so sometimes words tumble out in the wrong order. 🙂
Nice Fission update with a visual refresh and getting the app icon out of macOS Tahoe squircle jail.
Are Mac developers actually adopting the menu item icons in macOS 26? I’m torn… I think this design change was unnecessary and adds clutter. But also, it feels incomplete if I don’t add my own icons.
Realizing that we never officially documented the new support for passkeys, so a lot of people missed it. Just added a new help page with the basics.
Finally investigating why SMS stopped working. Now that we have Passkeys in Micro.blog, I think I’m going to scrap SMS as an option, and clear all the phone number data we have for users. It clearly wasn’t used much anyway.