Stephen Downes points to this announcement from MIT and Fujitsu.
He (at DLORN) and I (at the CMR site) have both proposed something similar back in the first decade of the millennium, but this second decade team have added some interesting new wrinkles. I don’t know if there’s anything new in a “navigation technology, which can organize massive online learning materials into multi-layer topics” “with multi-layer topics having different granularity based on a probabilistic topic model (Latent Dirichlet Allocation) framework”, but their “students’ learning behavior simulation based on an advanced probabilistic learner model”,via a “stochastic, Bayesian Knowledge Tracing algorithm” may be something new. And the “implicit rating system for learning materials, in which learning nuggets are not rated by learners directly, and instead their ratings are calculated based on learning outcome of learners” may also be a good idea so long as there is some user input into the identification and prioritization of learner outcomes.(Otherwise it seems little different from the choosing one side in the perennial debate about whether instructors and their materials should be rated more on the basis of student opinion surveys or on exam performance.)
Anyhow, it will be interesting to see what the product actually looks like when it finally comes out of the box.