By Stephen Downes - Stephen's Web. A Gentle Introduction to Deep Learning for Graphs
Davide Bacciu, Federico Errica, Alessio Micheli, Marco Podda, arXiv, 2020/01/29
By the second paragraph of the 'high-level overview' (the paragraph titles 'Mathematical Notation') the reader may be forgiven for thinking that this introduction is not so gentle. More...

By Stephen Downes - Stephen's Web. Why the Brain Is So Noisy
Michael Segal, POCKET, Nautilus, 2020/01/29
This sounds like my brain: "Kaschube and his colleagues... have found a host of features that stand in stark contrast to the circuits that engineers build: spontaneous activity and correlation, dynamic context generation, unreliable transmission, and straight-up noise." It's never quiet in my brain. I suspect it's never quiet in most people's brains. More...

By Stephen Downes - Stephen's Web. Top 5 Myths to Teaching a Choice Curriculum
Janet Taylor, Art of Education University Magazine, 2020/01/15
To be clear, the article is marketing for the Art of Education University, which describes itself as "a university built by art teachers, exclusively for art teachers." It offers access to online professional development resources for a monthly fee. I'm not endorsing them, just noting that they exist. Anyhow, the article specifically discusses offering "choice" in classes, which is what caught my eye. It addresses arguments against choice, for example, "(students) just don’t know what to do when they have full choice", and "you don’t teach any technical skills!". More...

By Stephen Downes - Stephen's Web. From context collapse to content collapse
Nicholas Carr, Rough Type, 2020/01/14
When social media first came into being, pundits like Michael Wesch and dana boyd talked of "context collapse" - you would no longer have a work identity, home identity, party identity, whatever; they would all collapse into a single public identity. But eventually people rebelled, and social media began to help us respect boundaries. But now we're entering the real on "content collapse". All content is the same. "A presidential candidate's policy announcement is given equal weight to a snapshot of your niece's hamster and a video of the latest Kardashian contouring." The danger here - as with context collapse - is that "content collapse consolidates power over information. More...

By Stephen Downes - Stephen's Web. Three ways to combat peer review bias
Verena Weigert, Jisc Blog, 2020/01/13
I'm not sure exactly how the headline writer got 'three ways' out of this article. By my count it could be four, or more likely, one. Or maybe one thing that is three things, like the Trinity. The article centers around an open-access scholarly publishing platform called F1000Research and cites it as an example of full open peer review. (One?) It references a project to "to test if reviewers would judge research papers differently if they can see who else has reviewed the paper." (Two?) It also describes a tool that does sentiment analysis of peer reviews. (Three?) Finally, it describes an open peer review prediction tool. More...

By Stephen Downes - Stephen's Web. Connecting Learning Analytics and Problem-Based Learning – Potentials and Challenges
Daria Kili ńska, Thomas Ryberg, Journal of Problem Based Learning in Higher Education, 2020/01/08
According to this article (24 page PDF), " Learning analytics (LA) have not yet gained much interest among the Problem-Based Learning (PBL) practitioners and researchers and the possible connections between PBL and LA have not yet been properly explored." The authors intent to address that by looking for possible connections. They do find two examples; one, a PBL workshop enriched with a LA dashboard, was of limited use; the other, representing existing surveys from social sciences in a visual form, was more useful. They also find potential in self-directed learning, reflective writing, and discussion support. More...

By Stephen Downes - Stephen's Web. Education and Employment Should Be One System, Not Two
Doug Irving, RAND, 2020/01/08
According to this RAND report (56 page PDF), a systems approach should be undertaken to mesh education and employment such that individuals can "acquire skills that are in demand in the labor market" and "ensuring timely and appropriate matching and rematching of skilled workers with jobs." It's an oft-expressed view that I think misses the mark in a crucial respect: an individual's educational experience serves aspirational objectives (which may or may not include getting a job) while a person's employment experience serves existential objectives, recognizing today's economic reality that if you don't work, you don't eat. More...

By Stephen Downes - Stephen's Web. What’s the Problem with Learning Analytics?
Neil Selwyn, Journal of Learning Analytics, 2020/01/07
This article (9 page PDF) sets out a number of major concerns about the use of learning analytics and sets the stage for a discussion of the subject in an invited dialogue in the SOLAR Journal of Learning Analytics. Selwyn argues that he looks "beyond the assertion that learning analytics is essentially apolitical and objective" and suggests that "While you might not agree with every one of these contentions, they all illustrate the inherently political nature of learning analytics." Responses follow from Paul Prinsloo, Rebecca Ferguson, Carolyn Penstein Rose  and Alfred Essa. Image via Debbie Morrison. More...

By Stephen Downes - Stephen's Web. Instructure’s Proposed Acquisition is a Bad Risk for Everyone
Michael Feldstein, e-Literate, 2020/01/06
"I have become increasingly worried," writes Michael Feldstein, "that, by going through with this particular sale at this particular time, there is a very high risk of destroying the company's value to customers and shareholders alike." Instructure doesn't have any viable products apart from its core LMS, and it seems to be stalled, writes Feldstein. More...