Teaching the systems by letting students touch them

You can lecture about a cloud testbed for a semester and end up with a room full of students who can describe one but have never used one. That gap is the part of teaching computing that has bothered me for a long time. The technology moves quickly, the syllabus moves slowly, and the distance between the two is mostly closed by hand-waving at slides.

The way out, as far as I can tell, is to put the actual systems in front of students. Not a screenshot of the system, the system. Cloud testbeds and notebooks make that more possible than it used to be. A notebook can hold the explanation and the runnable thing in the same artifact, so the student stops being an audience and starts being an operator. A testbed like Chameleon, which Kate Keahey runs out of Argonne and the University of Chicago, gives them real infrastructure to operate on, not a sandbox that pretends.

That’s the bet underneath the new NSF CyberTraining award we picked up this summer, where I’m a co-PI on the NIU side with Dave Koop, in collaboration with Kate’s group. The funded work is about course material that runs on Chameleon. The thing I actually care about is the pedagogical shift the funded work makes possible: less lecturing about data-centric computing, more doing it.

I don’t think this fully solves the gap. Pure lecture is hard to dislodge, and good hands-on material takes more work to build than slides do. But the direction is right. If we want students to be ready for the systems they’ll actually meet after graduation, those systems need to be close enough for them to touch.

Award details are at the NSF award listing.