I frequently describe openness and analytics as chocolate and peanut butter – both are tasty individually, but together their synergy is truly remarkable. Until recently we only had one example – CMU’s OLI – where this synergy was really running at full steam: openness providing permission to make improvements to curriculum and analytics providing empirical evidence about what changes are needed. (Note that neither the permission nor the evidence alone are nearly as powerful as the two together.) CMU OLI also leverages openness to increase the number of students using their material, which in turn generates more data, which in turn enables more powerful analytics, which in turn leads to better material, etc. CMU OLI’s openly available research shows the progress they’re making on using openness and analytics to improve student learning.
David Hu’s recent and awesome post How Khan Academy is using Machine Learning to Assess Student Mastery adds Khan Academy to the very short list of organizations really working hard at doing this well. Kudos to David, Sal, and everyone involved. They appear to be squarely in the openness/analytics feedback loop.
Next generation OER, or whatever you want to call it, is not just about publication. It’s about continuous improvement – that little bundle of philosophies and approaches that has revolutionized just about every large-scale field of endeavor besides education.
The next generation has a few problems to solve before they grow up completely, though. First, there is currently no meaningful way to reuse, revise, remix, or redistribute the assessments used by CMU OLI or Khan Academy. (I’ve addressed open assessment previously.) One first step that could be taken down this path is to make assessments embeddable like YouTube videos, with full analytics about use of the embedded instance available to the embedder. Even that tiny step would be huge headway, but would not address 2 of the 4Rs (revise and remix).
Second, both these initiatives are generating huge amounts of data which could be deidentified, aggregated, and shared with the community under open terms. Have you ever tried to teach a course on learning analytics? When you do, you’ll suddenly realize that there are precious few places you can go to get access to an education-related dataset of the size you need to really practice analytics techniques in a meaningful way. Contrary to popular belief, this can be done legally today under the terms of FERPA.
CMU OLI and Khan Academy are clearly out of the egg, but won’t be out of the nest until they make meaningful headway on these problems. But having said all that, it’s wonderful to see some innovation and forward progress in the OER world.