I noticed that there are quite some "simple" questions with the tag "numerical-analysis" or "linear-algebra", which haven got a "good" answer so far. Of course, I could try to write good answers to these questions myself. However, I have the impression that the better idea would be to leave a comment that the Computational Science stack-exchange site might be able to come up with better answers. (This doesn't necessarily apply to all questions with these tags, but if a question starts to involve the tradeoff between robustness, accuracy and speed, then "scientific computing" experts really might come up with better answers.)
I don't think that migrating old questions would be a good idea, but I wonder whether it could make sense to migrate such questions more actively in the future.