Eigenmumbles

Our department does not currently offer courses on iPhone programming, or on building web frameworks with Django, on AJAX or HTML5, or on any of a plethora of new programming languages and frameworks that come and go each month. Even if we had the staff available to teach such courses, it’s not clear to what extent we would want to do so. The technology du jour often has a short half-life, and we are not doing anyone a service if we teach skills that are too narrow to continue to be useful for many years after they are learned.

This topic came up in a recent faculty discussion, but it has stayed with me over the past couple days as I think ahead to the coming lectures in my parallel computing class. I would like to think that this course has value because it offers students a framework for understanding serial and parallel performance, along with many examples of how to think about the complex interactions between hardware, software, and numerical algorithms in the process of designing high-performance numerical codes. But at the same time, I do teach technologies, even if many of them have been around for a while. When I’ve taught this material in the past, I’ve always had a few students who were very pleased to tell me that my class is where they learned about Makefiles or gnuplot or some other piece of UNIX lore. And I often think to myself that I would rather here how much they got out of the higher-level ideas in the class.

But… I’m also glad that they feel happy about learning something, and I do think there’s great value to learning some of these tools. The details of a given tool are not so important as the realization that these tools exist, and can be used to do wonderful things in the right hands. Learning UNIX means learning about how to cobble together an endless diversity of useful tools out of small, flexible commands; learning to use a profiler means gaining a deeper understanding of where our mental models of performance can go wrong; and learning about C pointer-chasing provides a valuable lesson in thinking about levels of indirection. Done right, teaching technologies is about teaching ideas. It’s as much about understanding that a thing can be done as about understanding all the grotty details used to do it.

And, of course, some of us enjoy the grotty details far too much ourselves to steer our students away from the pleasure of digging into them.