news (external)

When React came on to the scene it was heralded as a way to create instantly responsible web pages. It did this with a virtual document object model (VDOM) which mirrored the content of the web page, ran API operations in the background, and synchronized it with the actual page display. I found it interesting but hesitated to jump in because it was developed by Facebook. It is currently by far and away the most popular framework for interactive site. This article suggests that the age of React may be ending, not because of any association with Facebook, but to make speed and interaction improvements. Via John Allsopp, who notes "finding developers who know vanilla JavaScript and not just the frameworks was an 'unexpected difficulty.''

I'm a bit overdue in posting this - I've been behind in posting archives. This is a conversation I had a couple weeks ago with Jim Groom and Taylor Jadin of Reclaim Hosting. The purpose was to talk about my experiences moving my website from the CPanel site to Reclaim's cloud hosting, though we covered a wide range of related topics.

This is the recording of a conversation I had with Matt Stranach on a wide range of topics related to educational technology and learning theory. We looked at (to paraphrase) some of my ongoing contributions to the practice, research, and outreach of e-learning along with the concepts or associated practices going into connectivism.

It has been often said that if somebody tried to create a library today, publishers would stop them. This case is proof of that. The Internet Archive established a virtual library - they would buy a book, then loan it out to one person at a time (I've made use of it and on occasion have had to wait until the copy was available). During the pandemic they loaned out more copies, to make up for those copies locked in inaccessible physical libraries. Publishers sued, fearful that they weren't able to gain maximum profits from people during the worldwide crisis. This is the result of that case. " In an open letter to publishers signed by nearly 19,000 supporters, IA fans begged publishers to reconsider forcing takedowns and quickly restore access to the lost books."

This is an in-depth look at the three major contenders in the fediverse: the ActivityPub protocol used by Mastodon and others; Bluesky, founded by Twitter founder Jack Dorsey; and Threads, Facebook's alternative. The article looks not only at the technical details but also at how the three networks handle moderation and censorship. It's a good overview written from an open perspective.

This is a textbook case of applying good critical thinking technique to show the fhe flaws in a causal study. The target in question is Jonathan Haidt's The Anxious Generation which tells us "how the great rewiring of childhood Is causing an epidemic of mental illness." Given the recent wave of mobile phone bans in classrooms, a sober rethink is sorely needed. According to the authors, Haidt cherry-picks research, infers cause from correlation, dismisses alternative explanations, generalized above the data, and more. It's as though Haidt's book was blended with a critical thinking textbook as a case study.

I've been enjoying Matt Tower's ETCH newsletter for the last year or so (the domain was registered in 2022). It offered a business focus on edtech, tracking mergers and acquisitions and such, things I don't normally cover here. Anyhow, it was acquired by Whiteboard Media group this week. "The EdSheet will join a growing roster of Whiteboard Media publications." I mention it here to reassure readers that there's no possibility of OLDaily being acquired - the data is open, the list of email subscribers is never shared, and most of the readership (via social media, RSS and web) can't even be counted, let alone tracked.

Interesting look at a copy editor whose job changed from editing the work of 60 human writers to editing the work of the AI that replaced them. "Mostly, it was just about cleaning things up and making the writing sound less awkward, cutting out weirdly formal or over-enthusiastic language," Miller says. "The real problem was it was just so repetitive and boring. It started to feel like I was the robot." This, of course, is just a temporary stage, and the editor was soon replaced by another AI. In this particular case the articles were park of the pink slime internet. But, eventually, they will be good.

The problems discussed have in common the need to serve hundreds of thousands of clients. This introduces numerous sources of error, as the one-in-a-million event is an even bet to happen. The recommendations in this article are woefully inadequate to the task. Two of them are focused on deputy ministers, the people least likely to have the necessary skills to address the issues. Another recommends training procurement staff in IT - though it would be far easier to train IT staff in procurement. The last offers the most hope, focusing on a minim viable product (MVP) methodology. But how do you test incremental development with hundreds of thousands of people? I would personally lean on more open source development, using already hardened components, and offering incentives directly to developers to create improvements.

At last November's ICDE conference, writes Junhong Xiao, "the president of an open university (OU) outlined the ambition of building a global digital university in his keynote speech." But is this the right ambition? Xiao makes the comparison with the parkway bridges built extra-low "to keep buses from the city away from Jones Beach – buses presumably filled with the poor blacks and Puerto Ricans Moses despised." In the same way, "if improperly or blindly adopted, AI can turn discriminatory and, in the case of OU education in Asia, may lead to more harm than good." AI is expensive. The cost of deploying AI may create a barrier against lower income institutions and students. He has a point. I have zero interest in AI - or any educational technology - if all it does is help the already-advantaged. Quality, to my mind, is meaningless without access.

GPT Builder was a nifty tool for people building custom GPTs. "Builders can use a conversational interface to create their GPT without having to manually fill out the required fields." It is being shut down in Microsoft's CoPilot. Microsoft reports, "we are shifting our focus on GPTs to Commercial and Enterprise scenarios and are stopping GPT efforts in consumer Copilot." Anything you built in GPT Builder will be deleted (I was feeling badly about not building anything but I feel better now). There's a ton of coverage online and people gloating about how there's no business model for AI, all of which seems overblown to me.

"We've found GPT-4 class models particularly effective in creating role-play scenarios" (76 page PDF), write the authors. In this article they describe some of their techniques: crafting prompts to the AI knows its role, creating a variety of scenario options, and having it provide a positinve and supportive experience for students (you don't want it to turn toward the dark side). There's a really nice 'negotiation role-play prompt' provided as an example. What I like about this is that it plays to the strength of the AI, where it doesn't matter if it hallucinates (that actually makes the scenario better) and where it doesn't depend on factual knowledge.

I've given tons of talks, some of them short, and while I see the appeal in this approach ('giving a talk is like selling yourself') and while there is some truth in it, I think the overall approach is misguided. Forget about selling; it's more like entertaining than selling. For one thing, I think readers should ignore the 'anatomy of a short talk' offered in this paper. That's a recipe for a snooze-fest. No, the main rule is this: start with the demo. In other words, show something right off the top. Present the main idea right away. Go straight to the most interesting thing. Take any questions the audience may have. Only then do you explain what you were up to: what problem you were trying to solve, maybe, or what theory you think this shows.

People like to say that "what's on the internet is forever" but older hands know that stuff disappears all the time. This is known as 'link rot' and this paper (14 page PDF) studies the rate of link rot in a university website and open data service. "A surprising 13.4% of shared URL links pointed to a website homepage rather than a specific record on a website. After testing, 5.4% the 2166 supplemental data links were found to be no longer available... Links from older publications were more likely to be unavailable, with a data disappearance rate estimated at 2.6% per year."

I did not enjoy this paper at all. In a nutshell: universities are dealing with 'supercomplexity' in and among the various disciplines, which pushes researching in one discipline across boundaries into other disciplines, creating what might be thought of as 'border wars' among them. To resist this, practitioners at the University of Hong Kong have implemented what they call a 'Common Core' to negotiate and facilitate these cross-boundary influences. We don't learn anything about the Common Core itself, though we read a bit about the 'de-territorialism' process involved in creating it. "Such an assemblage," write the authors, "creates a transversal curricular construction of refrains rather than one that relies on the tired dichotomies between the 'specialised' and the 'general' or the 'horizontal' and the 'vertical.'" The relatively simple story I relate here is presented in this paper with heavy and unnecessary layers of theory. I suppose that's what's needed to get published in the journal, but it otherwise serves no useful purpose. Image: Turner et al., Creativity and Innovative Processes: Assemblages and Lines of Flight.

I'm sure this metaphor appeals to a lot of people. "Learning's 'prickliness' is the inherent discomfort it brings. As workers, parents, and friends, we are often required to step out of our comfort zone, confront challenging situations and grapple with complex concepts." But the thing with metaphors is that they don't work for everyone. The challenges of learning are right where I feel at home, for example. What some people find prickly I find soothing and engaging. Your cactus is my aloe.

This is a long detailed post documenting the rise and (mostly) fall Lambda School, aka Bloom Institute of Technology. To be clear, "Lambda School targeted single mothers, the disabled, reformed convicts, and people struggling with serious medical problems. They lost tens of thousands of dollars, some lost years of their lives, on a broken, predatory program." This is billed as a tech story - the founder, Austen Allred, is compared to Elon Musk and the company to Uber. But it isn't, it's just another case of someone hyping a service using 'tech' and venture capitalists falling for the hype. There's a lot of this in EdTech, just as there is a lot of this in industry generally. Via Marco Rogers.

George Station calls it clickbait, and maybe it is (though unlike real clickbait it took a lot of time to create and delivers real value). This wide diagram allows readers to "explore how technical and social structures co-evolved over five centuries in this large-scale research visualization." I would have my quibbles (I think they leave out a lot in the period 1500-1800) but it's certainly something to pause and ruminate over. And yes, there's a column for education. Read more about it here.

Thisis kind of a neat design exercise: "a new app called Butterflies is aiming to create a social network where humans and AIs interact with each other through posts and DMs." Now not surprisingly the reaction is that it "sounds downright sociopathic and great for narcissists". I personally don't see why narcissists shouldn't have their own app, and to be honest I'm beginning to question the barrage of moralistic criticism of this and related technologies (this comes right after I finished reading a downright offensive criticism (via Ben Werdmuller) of AI by someone who pretends to be the world's only authoritative voice on the subject. I get that people don't trust the corporations behind AI nor the ideals of the people doing the developing. Me neither. But the language is getting judgemental and the responses reactionary, and I'm not a fan.

This paper (14 page PDF) offers an outline of how academic credentials could be secured using blockchain. Eventually something like this will be put into place - the main challenge here isn't the technology but whether institutions will cooperate enough with each other to support such a system. I would also add that most people will never see the workings of such a system - instead, they'll just have 'verified credentials' on their digital wallet or some such thing.