How much programming here?
Consider the two questions below. I'm not sure why one was closed and other not. I think both are fine; I liked Suresh's answer to the closed question, and was hoping to see more. I think this is relevant because there seems to be some consensus that Machine Learning is within bounds, and computational statistics can't be completely out of bounds, so we should expect questions on algorithms related to the field.
So what rules, loosely speaking, are in play when deciding to close such a question? I think this would be nice to advertise in the FAQ, so people can decide whether they should be headed to MetaOptimize or StackOverflow before posting such questions here.