MIT CourseWareThatIsMerelyAjar
by Sebastian Benthall
Yesterday I started looking into MIT OpenCourseWare and got very excited. The range of courses for which there is open material available is enormous. With these curricula at my fingertips, the way was open for me to educate myself, on my own time, without significant expense.
Then I dug a little deeper, into the available content for their introductory course on Nonlinear Dynamics I: Chaos. To my dismay, I learned that their assignments involved code snippets written for MATLAB, proprietary mathematics software for which you need to purchase a license.
MATLAB is an excellent product and a standard software package used in universities, so it makes sense that courses transitioning to openness would initially depend on it. However, it presents a serious obstacle to the open distribution and use of the course content.
Fortunately, Ryan Morlock has listed several open source alternatives to MATLAB, and some use languages that are “mostly” compatible with it. It looks like the most of the assignments wouldn’t be difficult to port over to something like Octave. It’s just a little shocking that nobody has done it yet.
Yep. This is one of the biggest problems with “open” courseware – it takes too much work to remove the quotation marks from around that word. The vast majority of “open” online courses are just note dumps from normal ol’ classes. There’s not much incentive for them to go further. Profs are busy, after all, and the students that pay tuition, after all, are being served just fine; if someone watching the course from Madagascar is confused, it doesn’t actually affect the university at all.
Remember those conversations at TOPP about code being “truly open-source” instead of just “released under an open license?” Release is a step in the right direction, certainly – a huge one. But until there are resources committed to making what’s released actually usable to people working on their own (a very, very nontrivial effort – I’ve tried it multiple times and did not always succeed), nothing will change. The trick is to persuade people with resources that this is something that should change, and is sustainable.
Well said, Seb, and Mel is spot on as usual.
Just wanted to point out another alternative to MatLab known as PyLab, which refers to the combination of Python, NumPy, SciPy, Matplotlib, and IPython. These should all be pretty easily installable on Ubuntu. I find it a much nicer scientific computing environment than Matlab. Not only is it more open, but since it’s Python, it actually has decent (read: first-class) facilities for I/O, OOP, etc. unlike (the poor excuse for a programming language that is) Matlab.
Take a look at http://onda.googlecode.com/svn/trunk/pychaos to see how easy it is to whip up a quick figure. You may recognize the name of someone else besides me in the commit log. :)