Deep dive: What are platforms, and why do they have so much power?
Question for parents: How are you teaching your teens to stay safe and think about safety and security while they surf, shop, chat and learn on their digital apps and devices? Are you sharing books, articles and videos with one another? If so, which ones? I’m asking for a friend; the friend is me.
Question for micro.bloggers: Can you separate keywords by comma in this category filter field, or is it one word per filter? 
Some thoughts on linklogs and link libraries
A linklog is a dated list of outbound links, maybe a tag, and a note about why it’s worthwhile. But that little note does two things at once, and together they make for a better web.
First, it’s a way of organizing the web. The web doesn’t sort itself, so finding good information depends on someone curating and tracking interesting content, preferably on the open, free internet. It acts as a customizable bookmark library: newest at the top, tags to group items, and a line explaining why the link was worthwhile. This is similar to what del.icio.us did, where people grouped and tagged links, creating a rough filing system.
Second, it’s a social act. An outbound link points to someone else’s work, so a linklog is essentially a renewable engine for sending traffic to other people and places. It stays renewable because the content is easy to sort and remix: a URL’s relevance no longer depends on where an algorithm or activity feed happens to place it, so older links can keep resurfacing instead of scrolling out of reach. That’s the precise opposite of what big platforms design for audiences; platforms such as Twitter, Instagram and Facebook quietly bury posts with outbound links to keep your attention (and dollars) on their site and with their advertising customers, losing relevance after one interaction. A linklog explores the opposing dynamic: it willingly hands your audience and attention away while crediting your forebears, comrades, mentors and co-conspirators. It turns the rough digital filing system into a “folksonomy,” if you will.
Compare these dynamics to an algorithmic feed on a platform. An algo also sorts the flood of content, but computer magic does the sorting, and someone else monetizes your attention and relationships and hoards your data. “Engagement” for these companies is primarily a measure of how long you stay on their platform with their content. The goal is to keep you scrolling and the ad revenue rolling. But over here on the indie internet, our goal is to build human communities online and connect people to quality information, tools, ideas and processes.
Back in the day, we had blogrolls, and blogrolls came with a lot of politics around representation and belonging, and some of those choices had real-world impacts. Who you listed in your sidebar was a public statement about your community and your allegiances; however, though smaller blogs linked to bigger ones, the big blogs rarely returned the attention. That contentious dynamic faded as social platforms became the main sites for online interaction, and the follow list, or the Twitter list, took over the blogroll’s old job of reinforcing networked hierarchies of writers and audiences. Because a blogroll is a static list of people, it hardens into a status game and is kind of stale the moment someone stops babysitting it; meanwhile, a linklog points at individual pieces of writing one at a time, so it stays current and rewards the work rather than the social and professional standing of the writer.
In this sense, a linklog is an easy way to be an opinionated curator and a decent neighbor at once, by hosting and sharing with thoughtfulness and deliberation. Out here, my data is also likely being scraped, but I have far more control over what data is findable, shared, and shareable, and over what is investigable and degradable over time.
Posting a lot while I get through the inbox today. Among other stories, WordPress is offering free hosting to students, marketing WP as a way to change how people relate to their own writing. Does a self-hosted public blog make students think differently about audience than a school LMS post does?
Chester argues that universities should govern AI by classifying data by sensitivity rather than approving specific tools. Public data goes anywhere –> regulated data requires an institutional contract –> sensitive data requires a local system. He also makes a renewed case for MS Copilot.
Community Notes was meant to democratize moderation power (and spare platforms the burden of thinking through moderation). Instead, a new paper shows that 20% of users cast 80% of ratings, and removing just ten active note writers from the pool changes what happens to nearly 5,000 community notes.
Gruber argues that AI text watermarking likely degrades prose quality and, even if it doesn’t, remains objectionable because it operates as an embargoed, unverifiable tag or signal controlled by providers like Anthropic and Google.
A person was found to be hiding an AI prompt injection in a legal filing that tells any AI that reads it to side with his perspective. The judge in this case likened it to “an automated agent communicat[ing] covertly with a juror during trial.”
Saw “The Odyssey” with the fam. 🍿
Computah: Make it a link
I spent a little time with the blog last night and pulled together two new site features using Claude Cowork. The last time I experimented significantly with Claude like this was to use Claude chat to build the link log from scratch, walking it through my thinking in plain language, then copying and pasting its suggestions into the backend of the site and hitting publish. This time I used Cowork, the tool that runs in the browser, and it clicked through the screens itself, fully taking on the execution of tasks. I have some coding skills, but not the kind these changes required. If I were taking this on, I would need YouTube, Hugo for Dummies, and my own personal IT guy, and still probably couldn’t pull it together.
Last night I asked a few things of Claude: I asked Claude Cowork to get into the backend of the micro.blog site and change my theme to link each line in the linklog back to my original post. The date field in the right column now links back to the original post for each link logged. No problem, easy request with easy execution. I made the request, confirmed the plan, and went about my business while Claude made the edits.
I also asked it to analyze my post content and suggest category tags for groups of content, then to label that content correctly in the backend. Claude reviewed about 500 posts and suggested I add a few new categories to my blog: AI in Practice, Books & Reading and Writing & Language. Then I set up some auto-filters to run at publish time to automatically categorize posts based on keywords moving forward.
Finally, I manually added the archive page, which lets you sort posts by category or year. This means I now have a functional archive here. Enjoy my anodyne thoughts, dear reader.
Adjacent to my day job, I’ve been toying with Claude Pro now for about a year; in my experience, it has improved significantly within the last six months. It’s not perfect: my requested edits were completed, but it also changed the CSS on the linklog so some of the text is too light to read, which I didn’t ask for and don’t want. But it has arguably extended my ability to execute on work that requires skills I don’t otherwise have (such as design and coding). What I do bring to the table is an expansive practical background in publishing and production, and all the language to describe it.
“When you gaze long enough into Instagram, Instagram gazes back into you.”
A friend of the blog shared the Embroidered Computer, a programmable 8-bit computer created by using a combination of magnetic and glass beads, gold wire, and traditional gold embroidery techniques to fix material elements to cotton cloth. When complete, it has the processing capacity of a 1950s mainframe computer without the modern silicon microchip. Something about this reminds me of Ellen Ullman’s observation that computers are made of boxes and wires, written on chips and tape. (via)
Claude is launching an AI watermark, one that users can’t opt out of (but can easily outsmart). As internet god Notopoulos wrote, “Claude’s watermarking poses the question: Are you willing to proudly own that you write with AI?” More: What is an AI watermark and how do they work?
I can’t believe it’s mid-August. I’m terribly behind on my reading list. Still, I did want to draw attention to this entertaining and extremely shareable article on the glory of Record Head, a legendary record store in Milwaukee. Make sure to watch their public access TV commercials at the link.
The knitting community is one of the great deliberative communities of the internet, and as they’ll happily tell you, knitting and fiber arts are closely tied to industrial and computing history. It’s really hard to pull one over on knitters online, because they have a deep and different relationship to the technical side of their hobby and all of the political and tech history behind it, and especially how it relates to computer engineering. I was thinking about that dynamic while reading about the latest fracas around YouTuber Hank Green, who is taking time away from public life after being called out by his audience, when his AI use was perceived as being in contradiction with his public persona. If you want to go down the rabbit hole on this, here’s a good video summary.
Ultimately viewers are reacting negatively to Green’s admission of using AI to quickly aggregate research to drive his content creation, suggesting that he lost his own creative discretion amid a growing demand for online content. But in the meantime, his technical inaccuracies were paired with completely avoidable gender bias alongside the AI use, which drew additional scrutiny from technical experts. It wasn’t his first go-round with the knitting community, it turns out. Was the bias Green’s or facilitated by his use of the tools? Debate ensues.
Winer on the purpose of web standards: “I find that most standards that come out of the standards bodies aren’t developed by implementors or with them in mind. They avoid trying to solve any specific problem, rather focusing on the elegance and flexibility of the format. This makes interop much less likely. I saw the differences several times as the web standards were growing up.”
I found this American Affairs article on the insidiousness of ad tech quite relevant. It focuses on foreign propagandists’ exploitation of social media, but it’s just as instructive for thinking about ad tech as tech, channel and product, and its entwinement with social media as infrastructure.
Garbage Day uses Jimothy to trace how virality works on the internet of 2026. Related: this story on how the algorithm has prevented us from producing a single “song of the summer.”
UW–Madison communication professor Catalina Toma argues that technology-mediated communication more often supports human connection and well-being than harms it, provided people use it intentionally.