Fun reading. As an aside, I've grown wary of data visualization tools tied too closely to a particular language. Each one is a little different, and while cranking out simple plots never takes much effort, making them look just so for presentation always involvers learning yet another low-level syntax. I've come back full circle to Gnuplot (http://www.gnuplot.info/), which I originally learned nearly twenty years ago while working on my Ph.D. It forces you to learn a shitty DSL, but you can get at Gnuplot from any environment that supports writing to text files and calling a sub-process. Plots are tweakable to your heart's content, and it does a fair job with 3-D graphics. I've reused a surprising amount of code originally written to plot error-control code statistics from Octave into a Ruby project that analyzes wireless network performance. Pretty cool!

But if you need interactive 3-D plots (e.g., a wire frame that you can rotate), look elsewhere.