The TeraGrid is an NSF-funded grid computing endeavor in support of science and engineering research. This week, TeraGrid ’09 took place in Arlington, Virginia and the organizers added on a one day workshop for high school teachers. I was hoping to get some ideas on how to introduce parallel computing concepts early and maybe even get access to some TeraGrid cycles for education but that’s not quite how it panned out. More than anything, it reminded me of CS4HS and provided me with some ideas to pass on to this year’s organizers.

The day started with a quick mention of some videos modeling scientific phenomena from CMIST. We then got a long presentation from Shodor founder Bob Panoff on using their computational modeling tools for teaching math and science topics. Shodor (allegedly named for “Short Dork”) is a really neat non-profit based in North Carolina that provides all kinds of great free tools as well as workshops and internships in technology for high school students. For example, teachers could use their function grapher to get students to explore the effect of constants on trigonometric functions by having them graph 1 * cos(x + 0) + 0 and use the sliders to change the values of the constants. The rabbits and wolves simulation is a great tool for showing population dynamics complete with updating graphs of each species count and parameters to modify the type of world.

These kinds of materials are great “hooks” for math and science teachers to use in support of discovery learning (more on that in a future post). They can also be used to get students to think about building and using abstract models to represent real-world situations, a key part of computer science. A teacher could, Bob suggested, use the histogram activity to make a point about the importance of knowing how to interpret models and the assumptions they make. For example, looking at NBA salary data, a coach might slice the data into three ranges and use the resulting graph to claim that nearly all players are paid the same. By slicing the data into smaller ranges, though, players might claim that salaries are not equitable. The Shodor system makes that kind of “zooming” in and out on data really simple.

Given that we were nearly all computer science teachers, I felt this was the wrong crowd. These are cool, but aren’t there 101 different (and most likely inferior, I’ll admit) simulations like that online? Besides, this seems like a use of technology as a tool rather than an example of building computational thinking abilities. But Bob started his presentation with an interesting statement — he rather casually said something like “in CS, we’re starting with modeling rather than programming because there’s very strong evidence that starting with programming creates barriers that wouldn’t otherwise exist.” No one really blinked and the tone of the day was set. I’m uncomfortable with that position but have to think about it more before I can craft a coherent response. I do believe that using tools like Vensim to describe models in a lightweight way is a good first step but I’m not sure that manipulating a ready-made simulation such as those Shodor provides really is actively building computational thinking abilities. In the best-case scenario, strong students will build some sense of what makes a good model or abstraction by playing with many existing ones. Not all students are strong students. I think that using scientific computation as a hook for algorithm design and computer science is unrealistic for most kids in our urban schools.

Another theme of the day was discovery and exploration as the best teaching/learning strategy — “let’s get out of the way of students’ learning.” Certainly, for self-learners with strong family support and excellent role-models, this is an excellent strategy but I don’t think it works for the majority. We saw an interesting presentation on a great collaboration between a community college, a liberal-arts college and a couple of universities. Charlie Peck and his collaborators had brought their LittleFe cluster — a home-grown machine made of off-the-self parts for under $3000. They run a Linux distribution called Bootable Cluster CD on it and use it to teach parallel computing. This is where they lost me a little bit: from the presentation, they made it sound like their strategy was letting students loose on the machines and having them build cool stuff. Note that this could be a misrepresentation as I haven’t done much exploration to see whether they have more structured content available. Still, I think that for most people, this seems like a wonderful idea! Indeed, when I mentioned the need for structured instructional materials it kind of fell on deaf ears.

All this sounds pretty negative, but I really enjoyed the day. It was a good opportunity to meet some interesting people and to continue to figure out my own ideas on education and where computer science fits in. I’m definitely a minority in a lot of my stances (structured instruction over discovery-based learning, structured assignments rather than open-ended projects, deliberately ordered and linear curriculum pacing, early emphasis on basics, etc) so I have to be particularly careful in what I do to either be open to change when things don’t work or to have data to point to when challenged.