That sounds an awful lot like you're saying it goes unmaintained and is therefore useless.
The truth is that these project view getting students to stick around and continue contributing to be more important than the actual projects they work on over the summer, and those retention rates are fairly low.
However, 20% retention rate is better than the 0% it would otherwise have been.
And for those who choose not to continue, if the project is healthy and the code useful then it will continue to be maintained by the project. Remember that this is something a student comes up with in a few months time. We aren't talking about massively complicated projects here. If someone else doesn't pick it up, then either the project is in trouble more generally, or else no one really viewed the contribution as useful anyway.
Other two big things are:
- video acceleration
- improving the Xorg state tracker for it to be on par with, say, xf86-video-ati (we've proved that the Xorg state tracker can outperform a "classic" DDX, but some features are still missing)
Yes there are many ambitious projects on offer, but there's often alot of easier tasks available aswell. If the student bites off more than he can chew it's not really the fault of the projects themselves. Still, I'd say that if this is a repeated pattern then it would probably be wise for the projects to lower the requirements of their propositions.
But when we are talking about stuff like gcc, llvm, kernels etc, there's not going to be a whole lot of 'easy' tasks, so here it's pretty much up to the student and the mentor evaluating her/him to make a correct assessment, which is obviously hard.
But overall, Google Summer of Code is a great initiative which has not only contributed a great deal of good code to numerous open source projects, but more importantly introduced all these students to these projects, which they may not even have bothered looking at had it not been for GSOC. I certainly wish there had been something like this back when I was a student, sure would have beaten working in electronics retail and all the other stuff I did during summer break to earn extra cash.
By that metric, more than 90% of ALL open source projects are useless.
No argument there. One only has to visit sourceforge to see that. (They really should do a huge purge there, at least on the projects that have no code after 90 days of the project page being setup.) They should also have a active/inactive setting.