OLDaily

Gist
Delz Erinle, Gist, 2020/08/13

I've been waiting for this. Essentially it's a really simply way to upload short podcasts ( it needs to be less than 5 minutes long, in MP3 format and less than 8mb), combined with a way to listen to them one after another in a stream. There's a search available, so it will be possible to listen to a series of audio clips on a specific topic (search doesn't seem to be working yet). Delz Erinle writes, "We believe there's a large and untapped market for audio as a social media, mainly because people have tried to use Twitter as a north star for a voice-based platform when the true north star for a platform that really works is YouTube."

Web: [Direct Link] [This Post]

Machines can spot mental health issues—if you hand over your personal data
David Adam, MIT Technology Review, 2020/08/13

I've been saying for some time that AI analysis of our online data (and especially the things we create and say) will replace tests and assignments as a means of assessing competence. This isn't exactly that, but it's following the same line of reasoning. For example, "People who are prone to hearing voices, it turns out, tend to talk about them." Also, "Low semantic density is a telltale sign that a patient might be at risk of psychosis." Now none of this is ready for deployment. And of course neural network based AI will provide much more fine-grained analysis (akin to a psychologist simply 'recognizing' that a person has an affliction). Still...

Web: [Direct Link] [This Post]

100 learning theorists... 2500 years of learning theory...
Donald Clark, Donald Clark Plan B, 2020/08/13

None of the articles is particularly deep, but when you take one hundred of them, you end up with a fairly substantial work. At the very least, as Donald Clark says, it's "written as quick, readable introductions to the many theorists who have shaped the world of learning."

Web: [Direct Link] [This Post]

Hear My Train A Comin'
Rob Abel, IMS Global, 2020/08/13

Rob Abel writes, " as I mentioned in my last post and the introduction to the recent IMS annual report, we have begun driving toward more specificity in the terms 'personalized learning' and 'student success.' Our starting point is the more specific goals of equity, agency, and mastery." These are good starting points, but of course a lot depends on the details. For example, George Moore "details some challenges when the modality switches to entirely online... such as the ability to support identity, privacy, and security."

Web: [Direct Link] [This Post]

TikTok and the Sorting Hat
Eugene Wei, Remains of the Day, 2020/08/13

The question considered here is "How did an app designed by two guys in Shanghai managed to run circles around U.S. video apps?" The answer, at least according to Eugene Wei, "in some categories, a machine learning algorithm significantly responsive and accurate can pierce the veil of cultural ignorance. Today, sometimes culture can be abstracted." But even more importantly, "The algorithm allows this to happen without an explicit follower graph." I think there's a lot more than that, and the detailed history of TikToK offered in this article supports that. But the algorithm is a big deal, a huge improvement over the "horrifying" algorithms of other services. See also Donald Clark, who explores some of TikTok's more interesting features.

Also, a quick news update from Metafilter: "Amid Microsoft's attempt to buy TikTok before a September 20th ban (which TikTok says shows "no adherence to the law"), Instagram launched its knock-off feature Reels, just the latest in a long line of Facebook's clones of competing products. Sarah Jeong says the "only question worth thinking about is why this matters to ordinary Americans — more specifically, should we be afraid of Chinese apps like TikTok?"

Web: [Direct Link] [This Post]

The Weird Toy Space of Scavenged Public Domain Images (A Bit Slimy, Too)
Alan Levine, CogDogBlog, 2020/08/13

In typical CogDog fashion Alan Levine takes an ordinary task and turns it into a deep dive into some aspect of online learning and design. In this case, he looks into where you can find reusable images online. Two things stand out: first, that there are numerous services that simply scrape and repost images from other sites; these sites are ad-supported or include links to for-pay services. The worst of these Levine (quite rightly) classifies as 'slimy'. The second thing that stands out is how these slimy sites can rank at the top of Google search results. "Google tilts the scale away from creators toward slimers," he writes.

Web: [Direct Link] [This Post]