That is a good one. I end up having the thought "damn, a solid implementation of this would really be worth something" when merely skimming half the research papers I come across.