The origin of viruses is actually somewhat mysterious, because they are extremely limited in their evolvability. Viruses can only speciate through cross-species infection and wandering around a little bit in genome-space, and they are not evolvable enough to cross types, so an RNA virus can never become a dsRNA virus, or a ssDNA virus, or a retrovirus, or anything other than a slightly different RNA virus. This is in stark contrast to eukaryotes, where there is a smooth evolvability path linking an oak tree to a baboon (through a single celled common ancestor).
This is the limit of evolution of self-replicating entities with errors. It is very different from entities which are truly alive.
Since viruses cannot really evolve, if new ones can't be formed, you would think they should all be extinct. As soon as one type is gone for good, nothing can replace it. One can make two hypotheses here:
I prefer hypothesis b, because I can't imagine that dinosaurs got influenza, or hepatitis. There must be a source.
The source then is in self-packaging genetic material which can leave and enter cell bodies, which has nothing to do with virus infections. Such entities do exist (surprisingly), the endo-retroviruses are an example.
Endoretroviruses form a significant fraction of the human genome, and they produce a coat, a reverse transcriptase, and they can even become virulent under certain conditions, such as cancer. There is a Koala endoretrovirus event underway right now, as Koalas are all getting a retroviral insertion in their genome.
The retroviral properties make it natural to assume that they are serving an important role in the body, perhaps to communicate RNA from one cell to another, so as to link the different computations in different nuclei, and perhaps incorporate into the DNA. They might also be used to transfer RNA between different individuals for all we know. This is speculative, but it is a very easily tested hypothesis--- look for endoretroviral particles in the bloodstream of a healthy animal, and see if they insert themselves into genomes of distant cells.
If the retroviruses are used for intercellular communication, this case, one can postulate that new retroviruses form when endo-retroviruses are accidently virulent and replicating. This is not hard to imagine, and it allows new types of viruses to emerge, with no relation to previous viruses. This claim, as far as I know, is original, it is not present in the virology literature I have read. It might be false, but at least it's original (no, actually, it's not--- I just didn't know the relevant literature--- see Adriana Heguy's answer on this page)
The source of RNA viruses is more difficult to see. It is possible that these evolve in special places, perhaps in bird genomes. If the viruses come from the complex animals, each virus, one must find the originating animal, and it must have homologous packaging proteins.
I should point out that if this is true, AIDS did not have to result from somebody screwing a chimp (or eating a chimp). The chimps could have independently generated the same infection from their own endoretroviruses. I don't know if this is viable, although I could test this by myself, just by comparing the sequence of SIV and HIV to see if they differ by more or less than the ERVs of humans and chimps. I didn't test yet, I am just asking.
Viruses actually evolve quit effeciently and rapidly, so your assertion that they don't is incorrect, infact viral evolution is mostly why we cannot cure HIV infections.
Bullshit. You don't know what it means to evolve "efficiently". They search genome space for little alterations in their proteins. They can't do anything more than this steepest descent. When you give 3 drugs, that's it, they're past the error threshhold and you control the infection. This is not evolution, it's miniature search.
Wait, do people actually think we got an altered version of SIV by screwing chimps? I always thought people eating chimps was accepted..
Either way, it sounds fishy to me. Eating chimps is rare.
I remember reading that mountain gorillas were actually killed and eaten in Congo a couple of years ago. That stuff happens fo sho, but having sex with a chimpansee? I'm quite sure most chimpansees would rip you apart (they're strong as hell) if you tried to do anything similar.
Of course, one could kill a chimpansee first and proceed to have sex with it, but I'm seriously hoping that one of the deadliest diseases of the 21st century was not the result of some dude into necrophilic bestiality.