OLDaily

Roadmap for Open Science
Office of the Chief Science Advisor, Government of Canada, 2020/02/27

The Canadian government's Office of the Chief Science Advisor has just released a roadmap (12 page PDF) to open science with a set of ten recommendations, something I view as a substantial step in the right direction. In particular, the document embraces "Open Access by default without an embargo period", as well as FAIR (“Findable, Accessible, Interoperable and Reusable”) data principles. Departments and agencies, including, I hope, my own, "should develop action plans for Open Science by October 2020." Via Richard Ackerman.

Web: [Direct Link] [This Post]

(Re-)decentralize the web with MoodleNet!
Doug Belshaw, MoodleNet, 2020/02/27

Doug Belshaw reacts to Society Desperately Needs An Alternative Web, in which Hessie Jones reports on the recent Canada-United Kingdom Symposium On Ethics In Artificial Intelligence (EIAI-2019), noting that "The article covers a lot of ground, but in particular adds to the growing chorus to re-decentralize the web." As Jones writes, "From Diaspora to Bit Torrent, a growing list of technologies continue to develop alternatives for the DWeb: for storage, social nets, communication and collaboration apps, database, cryptocurrencies, etc."

Web: [Direct Link] [This Post]

The Moral Problem of Grading: An Extended Analysis
John Danaher, Philosophical Disquisitions, 2020/02/27

My least favourite thing about teaching is grading. Unsurprisingly, I am not alone. And it's not just because grading is a lot of work (though it is), it's because it's so ambiguous. This article looks at the ethical issue of grading, examining what sort of communication a grade is intended to be, what it represents (is it an absolute measure or a relative measure?) and what good or (mostly) harm it causes. In the end, the conclusion is that while we may have to live with grades, we should recognize that they are a poor solution to the problem they are trying to solve, and so we should not take them as an absolute indicator of anything.

Web: [Direct Link] [This Post]

Clearview AI: Face-collecting company database hacked
BBC News, 2020/02/27

You may recall that Clearview AI is the company that has been scraping websites in order to obtain a gigantic collection of portraits which it is using to train facial recognition software. So this is an object lesson teaching us (again) that such data is never secure. While hacking is obviously illegal, though, I can impagine the hackers justifying themselves by saying they were just scraping Clearview's database.

Web: [Direct Link] [This Post]

Data-driven JAMstack with Sourcebit
Brian Rinaldi, Stackbit, 2020/02/27

I haven't had the chance to follow the instructions in this post (I'm still playing with Fastpages, which is now working for me - I'll be posting a follow-up soon) but what's happening here is a really good indication of how modern web design is progressing. A JAMstack, recall, is a site built with Javascript, APIs and Markup. Typically this build process is automated using a static site generator like Jekyll (which is what the Fastpages site uses). What Sourcebit does is to pull the data from the source, trsnaform it as needed, and make it available for use by the static site generator. In the demo example, the data source is a Contentful site, and the static site generator is Hugo.

Web: [Direct Link] [This Post]