I like the idea of a contest, but I'm not really sure how it would work.  The TSP example that was mentioned is just an approach to solving a problem...even if you came up with a TSP example to solve for 10,000 cities or something...it's the network that actually solves the problem.  The author just codes the algorithm...they don't solve anything.

Maybe a different way to look at this would be to come up with some sort of incentive for authors to create different types of basic examples that could demonstrate various use cases for xel.  That might draw in some outside groups to take a look at xel to see what it is capable.