Stephen Downes

Hollis Robbins gives us a great article comparing Friedrich Hayek's 1945 essay The Use of Knowledge in Society with his 1949 The Intellectuals and Socialism as analyses of the centralization and standardization of university curricula. "Read together, the 1945 Hayek explains uniformity as the product of a bureaucratic system that destroys local knowledge by demanding interchangeable units. The 1949 Hayek explains uniformity as the product of an intellectual class converging on fashionable ideas." The problem with the 1949 Hayek is that it "cannot explain why universities are doing the same thing everywhere, when there's no ideology involved." And that's what we're seeing today. It's not ideological conformity. It's "a system that has destroyed its own capacity to adapt. This is the world we're living in, until critics start noticing the problem is not ideology but centralized planning." It reflects the convergence of a university system around standardized curricula at the expense of local knowledge. Obviously I don't think this is the whole analysis, but this is a big part of it.

I've mentioned my own experiences working with AI to navigate the intricacies of Git. Here Jon Udell relays his own experiences. He writes, "What I brought to the table was the knowledge that git bisect was the right tool for the job. Claude Code brought the ability to wield the tool effectively. And as it did so, I watched and learned. This aspect of LLM use is not a black box. When agents run commands on your behalf you can see and approve them. 'I should probably take an online course,' my friend said, 'or watch some videos.' You can, I said, but there's no better learning experience than to be guided through the use of a tool in a situation where you need it to solve a problem in the work you're actually doing." People who create and teach classes for a living should ponder this. The issues of 'cheating' or 'cognitive offloading' are merely distractions obscuring the real impact of what's happening here.

This is a longish and wide-ranging article in response to an OpenAI Education 'summit' for education leaders held last week in San Francisco. Marc Watkin's primary take is that "there was no roadmap or vision for shifting students from the free version of ChatGPT to the educational version of the tool" where, presumably, they would be required to pay for it. Personally, I think the answer is pretty easy: all ChatGPT has to do is charge money for access to upgrades and new services. The free access students enjoy now will be a toy compared to what will be available in a year. Anyhow, the article is a good overview of the summit, but most of Watkins's criticism consists of questions he had that were not answered at the seminar. 

Cloudflare's announcement caused a stir in the socials today. "You can now crawl an entire website with a single API call using Browser Rendering's new /crawl endpoint, available in open beta. Submit a starting URL, and pages are automatically discovered, rendered in a headless browser, and returned in multiple formats, including HTML, Markdown, and structured JSON." To be clear, web scraping applications have been around for a long time (it was the first thing anyone wrote back in the 1990s!) with services like Firecrawl and Elastic (try). The issue is that Cloudflare has marketed itself as a service that prevents web scraping. Still, as Ian Kerins argues, the Cloudflare scraper "doesn't bypass anti-bot systems." However, "What this '/crawl' endpoint really is though, is another step in Cloudflare's broader push toward signed agents and a 'pay-to-crawl' internet."

I spent yesterday working with Claude to transfer my newsletter functions to Amazon SES, something that would have taken a week in the past, if it were possible at all (there were some very tricky bits involving signature validation and depreciated Perl crypto modules). For the record, it was a lot of fun, and I learned a lot (though Amazon's user interface is still an impenetrable wall of confusion to me). So when Ethan Mollick talks about how much AI models have improved over the last few weeks, I believe him. He outlines AI's performance on various tests and standards and describes The Software Factory, a system that takes human specifications and outputs functioning code, no human code review required. "AI companies are telling us, fairly explicitly, what comes next: recursive self-improvement, or RSI. This is the idea that AI systems are increasingly being used to build better AI systems."

This is really neat: a version of WordPress that runs entirely in your browser on your computer. See also Planet WordPress on this. "There's no sign-up, no hosting plan, and no domain decision standing between you and getting started... This isn't a temporary environment meant to be discarded. It's a WordPress that stays with you... sites on my.WordPress.net are private by default and not accessible from the public internet... This turns WordPress into a personal workspace." I've tried it and it works. It uses local storage for files and an SQLite database. My CList app is built the same way, but this is much more sophisticated. "Thanks to incredible advances in WebAssembly (WASM), we can spin up a web server, a database (SQLite or MariaDB), and a full WordPress installation inside your browser in about 30 seconds. Instantly. No server needed." Via Alan Levine.