Orwell wrote:
The thing is, though, that the phage will adapt to the bacteria on its own, so you can pretty much let nature do the work of creating new drugs for you.
Ah, yes: but different populations of bacteria will co-evolve with different strains of virus, leading to balanced populations (parasites, including disease parasites, generally evolve into non-lethal symbiosis with their hosts). Even this would be helpful in many infections, because it could slow a bacterial population down long enough for a human immune system to figure out a good antibody to fight it with. But a disease parasite that totally wipes out its host population usually needs to be an accidental crossover from another species (a la ebola) or moving from a co-evolved population to a naive one (a la smallpox and native Americans). So mixing and matching different strains of virus would provide a higher bacterial kill rate than allowing the viruses to evolve into less virulence.
Also, it would probably make sense to re-harvest viral stock from infections that are clearing up more rapidly than normal, as that particular viral sub-population might have a higher lethal rate than normal. I wonder how patients would feel about having their viruses harvested for posterity?