Fascinating! So, how much of the genome do you think is made up by these? And are there start & stop codons?
Unknown. This is a very new area of research, especially as it relates to mammals. The Scientist just published an article on a handful of scientists who now believe that much gene regulation is actually effected through untranslated RNAs transcribed from intragenic regions. This is, of course, a highly controversial idea. If so, short, interfering RNAs would be one of perhaps a number of RNA-based mechanisms for gene regulation. It's not controversial that RNAs do have certain very specific regulatory roles. For example, it's pretty clear that there are RNAs which play a role in X-linked dosage compensation in female mammals -- that is, where one of the two X chromosomes is effectively put to sleep in every cell to produce the same gene dosage as in males (XYs). However, it's generally thought that proteins are far more important than RNAs in regulating transcription in general -- but these scientists are suggesting there's a whole world of RNA-based regulation that we've only scraped the surface of to date.
In general, there is a general trend in molecular biology towards discovering more and more things that RNAs can do that we had previously only attributed to proteins. This is not that surprising: RNAs are just another type of polymer, and most biologists think that the first forms of life were based entirely on RNA, with RNA performing the catalytic role now largely taken over by peptides, and the genetic or information storage role now performed by DNA, in addition to serving its present primary role of messenger. However, no one knows for sure if the "RNA world" idea is correct, and there are some interesting alternative hypotheses out ther.
Getting back to RNAi, I don't think that is a legacy of RNA world, or anything quite as exciting as that. RNAi is based on a set of proteins which recognize double-stranded RNA -- something that isn't normally used in the cell -- and attacks it. So, it probably evolved as a defense mechanism against foreign nucleic acids. However, once the system was in place, it was co-opted to a gene regulation role, at least in a few instances that we have recently discovered, and possibly in a bunch of others that we don't yet know about. But that's something you see all the time in molecular biology -- a new module or function evolves, and then gets co-opted to other roles over geologic time.
My primary interest in RNAi is as a controllable method of knocking out specific genes, rather than in the phenomenon itself. We didn't know how to do this in mammalian cells until very recently, and it could turn out to be a powerful tool for reverse-engineering genetic circuits. Having said that, all the results aren't in yet. Check back in a a couple of years, and it may still turn out to be a lot of hype.