Here's an interesting case study about a class that was used to launch the entrepreneurial careers of several folks at Stanford. It is very useful for thinking about educational entrepreneurship since it shows a strategy with very little investment, but one that can have a huge impact in terms of spurring innovation. The key looks to be making sure the right mix/type of folks are taking the class.
Imagine K12 is another startup launch initiative being pushed by some veteran Silicon Valley folks. The model is "unabashedly inspired by Y Combinator" and seeks to give guidance and support to new startups looking to reshape education. Here is a link to the basic structure of their program:
This could be a great experience for Edlabbers or a useful model to look at for our own Launchpad 39B initiative.
Interesting choice by the Media Lab for a new director. My experience with that lab, however, has shown that it values mixing traditional academics with some unconventional characters (at least in the realm of higher ed) ...
We've been talking a lot about moving to HTML5 for Vialogues and eventually having more robust support for vialogging using mobile devices. One challenge we've often run in to is making a choice about a web application (run inside of the mobile browser) vs. native applications. Native apps are inherently more powerful and have more features available, but web apps are much quicker to develop and can be ported across multiple environments.
The New York Times is rolling out its new digital subscription policy. There's a whole article by its publisher about the shift explaining the different options. What struck me is that they are struggling through many of the issues we are encountering and it will be an interesting experiment to see how users react. Some highlights that I thought echoed our experiences thus far with vialogues:
- They want to let people access some limited content for free (so people don't just turn away from the site), but want heavy users to pay for it.
- They don't want to penalize people for coming to the site from other places. If you follow a facebook link to a site, then apparently you will be able to view the article even if you've reached your unsubscribed limit of articles. This is just like our struggle to let content producers give away their content for free to encourage the production of content, while also realizing that when people access that content it will cost some money.
I've recently read several articles discussing how YouTube is trying to expand beyond just a collection of short, user-generated videos to more professional content. For example, this article discusses how YouTube is looking for web production companies that can produce content for their site. Other efforts have included expanding the limit on how long uploaded videos on YouTube can be and other efforts to attract better content.
This really points to the maturation of Internet video from just novelty videos that are typically short and poorly produced to more sophisticated content that can attract actual dollars from users. I'm sure as Apple, Netflix, Hulu, and other players continue to battle for content online, the pressure will only grow for YouTube to post content that people will actually pay for rather than being just a collection of free videos (there's already much talk about how YouTube doesn't generate as much money as it does buzz for Google).
Thought this new effort was very relevant, given Gary's recent post on educational data mining.
Right now it's primarily ad-driven businesses that are complaining about this, but given the trends around data collection and adaptive learning, this could become a major issue for non-traditional learning services that may have a harder time convincing users to let them track their activity.
Google has a "manifesto" on the social web. It's definitely an interesting read and helps frame some of the things that may influence connectivity, privacy, and collaboration in learning-oriented social environments.
This was a presentation by researchers at the University of Memphis who mined a variety of data about user behavior to detect emotions while individuals are engaged in online learning or assessments. What I found really interesting about their approach is that it is multi-modal: they take into account facial features, bodily postures and even physiological parameters (eg. heart waves). They take this data and try to predict different emotions more accurately using machine learning techniques.
Here is a link to a related paper.
And here is a link to their project page.
I'm also attending the EDM conference with Pranav and Ankit. We are currently sitting in on a session about mining student data. One of the important observations I've made so far is that several of the papers rely strongly on defining a clear outcome of interest. In most cases, this involves test scores, which are then subsequently mined.
One question that comes up is whether such outcomes are always appropriate and if there are other observable and quantifiable outcomes that we could use to measure the progress of some adaptive learning tool. In particular, are there ways to plug in more "qualitative" outcomes into the machinery of artificial intelligence to predict outcomes, cluster similar students, and serve appropriate content to students.
This seems to be similar in many ways to the problems that arise in econometric studies of learning as well. Both data mining/AI fields and economics rely heavily on statistical methods that use some measure that "behaves well" in the various statistical algorithms that are available to the researchers.