by Stephen Downes
Sept 22, 2016
One of the biggest disappointments I had with the commercialization of the MOOC through the Stanford and MIT products was the idea that the MOOC would have to be "sustainable" through some user pay mechanism. In 2009 the average tuition was around $4500; this accounted for between 30-50% of the total cost of an education. And of course it was paid only by wealthier families; low income earners need not apply. If you multiply that by the 2 million people enrolled in post-secondary institutions in Canada, you get $18 billion. If MOOCs could have reduced this number, they would have been a substantial success, and not cost students a dime. I think we could have made a dent in that. But too many people sound like Alex Usher, saying "The problem is there’s no revenue model here." No, that's not the problem.
In this long (25 page PDF) article and interview with Clifford Lynch, Riochard Poynder looks at the state of affairs of open journal repositories (for example, the Open Archives Initiative Protocols for Metadata Harvesting (OAI-PMH)) and does not appear hopeful. "As with all issues concerning scholarly communication and open access, no one appears to have the necessary authority (or even perhaps the capability) to oversee strategic decision-making at this level effectively. And that is why it seems to me most likely that the academic publishing oligopoly will succeed in appropriating both OA and the institutional repository." From where I sit, reading this article, the main culprit (beyond the publishers themselves) seems to be the indifference of university professors. Publishing openly still seems to be a "minority sport".
There are several things to say about this report. First, the headline is wildly overstated. 'Not finding' a correlation is very different from 'finding no correlation'. Second, it's a metastudy. The authors took a number of previously published studies, copied their data, cleaned it up and ran a new analysis on it. Third, only in-class student evaluations were used, not the popular online teacher evaluations. Fourth, we are given utterly no definition of what counts as 'learning'. Does it mean test scores? If so, it's old news that students don't base their evaluations on test scores. Finally fifth, the original study, still in preprint, is locked behind a paywall, and I just don't think I could bear spending $41 only to find that it's test scores. If the authors of this study have anything to say, let them say it openly where it can be scrutinized and criticized.
It's just an infographic and won't give you a need knowledge of neural network configurations. But it's still useful. "Though all of these architectures are presented as novel and unique, when I drew the node structures… their underlying relations started to make more sense.mOne problem with drawing them as node maps: it doesn’t really show how they’re used. For example, variational autoencoders (VAE) may look just like autoencoders (AE), but the training process is actually quite different."
This newsletter is sent only at the request of subscribers. If you would like to unsubscribe, Click here.
Know a friend who might enjoy this newsletter? Feel free to forward OLDaily to your colleagues. If you received this issue from a friend and would like a free subscription of your own, you can join our mailing list. Click here to subscribe.