Based on user's previous answer history and, especially, reputation gains assotiated with each tag, provide an automatic selection of recently asked questions which "you may have an answer for".
Of course the method would be heuristic, but it would help users to discover interesting questions. This is not based on their "interesting" tags, but rather on their past upvoted/accepted answers. (I may be "interested" in learning php, but know almost nothing about it.)
Edit by Ivo Flipse
Sorry for hijacking this post but didn't want to duplicate it
With the new Stack Exchange sites it becomes increasingly harder to keep track of interesting questions to answer.
Yes, tags solve this problem partly, but I believe that with our new Statistics guys should be able to come up with an algorithm that uses the data we already have available to do a better job of suggesting than tags currently do.
As Jeff previously asked what we could learn from Quora, I think we can learn from their front page. It gives me an overview of questions I'm possibly interested in from all the different topics.
Now that the Stack Exchange page has been pimped to show hot questions over all Stack Exchange sites, I think it would be useful to show questions that a logged in user might find interesting. Especially since I don't need hot questions if I already answered half of them or am a regular user.
Any suggestions are welcome!
Oh and I would love a Netflix-like challenge to come up with the best suggestions!
And attribution to Lance, since his question asks for something similar