Please keep your suggestions programmatically very simple. There are all sorts of bright ideas we could be trying, but the actual strong filter by which only a very few are implemented is that programming resources for LW are very scarce and very expensive.

(This filter is so strong that it's the main reason why discussion of potential LW features didn't in-advance-to-me seem very publicky - most suggestions are too complicated, and the critical discussion is the one where SIAI/CFAR decides what we can actually afford to pay for. It hadn't occurred to me that anyone would dislike this particular measure, and I'll try to update more in that direction in the future.)