OLDaily

Welcome to Online Learning Daily, your best source for news and commentary about learning technology, new media, and related topics.

POSSE is in the air
Ludovic Chabant, The Stochastic Game, 2024/09/30

---

As Ludovic Chabant writes, "POSSE is a pretty bad acronym (standing for "Publish (on your) Own Site, Syndicate Elsewhere") describing a pretty good approach where you publish your writings, status updates and cute pet photos on your own website before some sort of (preferably automated) system posts a copy to various 3rd party proprietary services (aka 'cross-posting')." It's what I do with OLDaily. Chabant reports, "I've written my own ad-hoc tools to implement it, and for a bit more than 6 years this tool has been SiloRider, available on Sourcehut (using Mercurial) and Github (using Git)." Cool beans.

 

Web: [Direct Link] [This Post][Share]

Two-Sigma Tutoring: Separating Science Fiction from Science Fact
Paul T. von Hippel, Education Next, 2024/09/30

---

I'm not often convinced by articles in Education Next but this one by Paul von Hippel has me nodding in agreement. I'd need a pretty convincing counterargument. In a nutshell, von Hippel discusses Benjamin Bloom 1984 essay that asserts that tutoring offers "the best learning conditions we can devise" and can "raise student achievement by two full standard deviations—or, in statistical parlance, two 'sigmas.'" Hence, we have the 'two sigma' problem posed against any innovation in pedagogy or learning technology. But this level of improvement proves to be a chimera; it doesn't exist. There's an analysis of the history of the argument and von Hippel takes a deep dive into the student work that gave Bloom his original data. The work was good, but the accomplishments were based on a lot more than just tutoring, and most tutoring isn't of the same quality the students provided. So, as a coda, von Hippel remarks, "It seems rash, though, to promise two-sigma effects from AI when human tutoring has rarely produced such large effects." Via Baldur Bjarnason.

Web: [Direct Link] [This Post][Share]

DSA compliance handbook for Fediverse administrators
ILP Lab, 2024/09/30

---

This handbook (25 page PDF) was written to help European social network hosts comply with the Digital Services Act (DSA) "aimed at increasing transparency and fundamental rights protection for social media platforms." It's of particular importance to people who host Fediverse instances (like Mastodon, say).  "Server administrators hold primary responsibility for setting and enforcing content moderation policies specific to their instances, but not every server administrator has the tools or the know-how to do so." What's important is that "the DSA exempts online platform providers that qualify as micro or small enterprises from the extra obligations established by the DSA for online platforms." So mostly you won't need to worry about the DSA, unless you're running it as an enterprise, in which case it should not be surprising that you're regulated.

Web: [Direct Link] [This Post][Share]

Web Components Are Not the Future
Ryan Carniato, DEV Community, 2024/09/30

---

What I find fascinating about this article is how closely it parallels the whole discussion the education technology community went through with learning objects. There's a lot of tech and jargon in the article, which discusses Javascript front-end programming, but it boils down to this: " Instead of making a million variations on the same thing you try to re-use one thing... (but) When primitives overstep their desired usage, when they over abstract, you don't get to come back from that." You have to either rewrite the primitive (that's a problem) or write another abstraction to deal with the primitive (that's also a problem). If you're interested in the discussion, here's the counterargument.

 

 

Web: [Direct Link] [This Post][Share]

The Zombocom Problem
Gordon Brander, Squishy, 2024/09/30

---

This is a problem I've wrestled with myself though my career, which is what prompts me to pass it on. The problem is this: I have always wanted to code platforms, that expand the affordances available to a user, but users "want something specific. To succeed, you need to solve a specific problem for a specific user and find specific product-market fit." Now this seems to apply for various definitions of 'succeed'. But I don't want to solve some existing problem because these are almost always rooted in trying to do the wrong thing (ie., the old thing) in the first place. I guess my approach is a good lesson in how not to succeed. But (from my perspective) success isn't the goal.

Web: [Direct Link] [This Post][Share]

Pronunciation Assessment with Multi-modal Large Language Models
Kaiqi Fu, Linkai Peng, Nan Yang, Shuran Zhou, arXiv.org, 2024/09/30

---

The authors conclude (5 page PDF), "the proposed scoring systems achieve competitive results compared to the baselines on the Speechocean762 datasets." It's not a surprise to me that an AI could be used to score participants on their pronunciation. What I wonder is how consistent their assessments are when compared with human assessors.

Web: [Direct Link] [This Post][Share]

Realistic dialogue is here....
Donald Clark, Donald Clark Plan B, 2024/09/30

---

Donald Clark notes that "advanced Voice Mode on the OpenAI app is now avaialble in the UK" and gushes over it. I notice that the AI audio performances are being referred to as 'parlour ricks' on social media, but it seems to me that if this was so easy to do it would have been done before now. No, I think there's a genuine capacity being demonstrated there, and the real questions are "do we want to interact with an AI via voice conversations?" (answer: sometimes) and "how real a voice interaction is too real?" (answer: it depends).

Web: [Direct Link] [This Post][Share]

We publish six to eight or so short posts every weekday linking to the best, most interesting and most important pieces of content in the field. Read more about what we cover. We also list papers and articles by Stephen Downes and his presentations from around the world.

There are many ways to read OLDaily; pick whatever works best for you:

• Read in your web browser on this web page, updated weekdays
• Receive a daily or weekly email newsletter by subscribing here
• Subscribe to the RSS feed using your favourite feed reader
• Follow @OLDaily@mastodon.social on Mastodon or elsewhere in the fediverse
• Follow @OLDaily on Twitter
• Integrate the open API using the JSON feed