This would actually be a terrible idea. Imagine all the rewriting of every little business app people have written for their blackberries, and the instant obsolescence of BES servers. No, RIM just needs to innovate in a way that doesn't seem half-assed.
OpenSolaris had a GNU userland. The BSD ancestry of the Solaris userland isn't the problem. The problem was the senseless defaults and cryptic command structures ("shutdown -g0 -i5 -y" instead of "shutdown")
Some of his arguments are crap. How is extensive platform support a detractor? Wouldn't you want your language to work on as many platforms as possible? I'm not sure how "outdated" stdlib is, but just because the code was written years ago, that does not make it "bad". New ways of implementing certain things could actually be slower. Can't say for sure since he has no real examples.
FYI: I dislike python for its syntax, but I can see why people like it. I just use perl for most everything that people would do in python.
I find that the largest problem about this competition is that the people who can actually do this probably have no motivation to compete.
We all have jobs, lives, and cool stuff to hack on in languages we actually like using.
From what I remember of middle school, they DID teach us a little comp sci. They taught LOGO for about a week. Seemed pretty fun. It was at least mostly competent from what I can remember.