"The “dirtiness” will depend on the extent to which VEGF is needed for normal retinal functioning."
I would argue that the "dirtiness" may extend beyond VEGF-specific effects. I don't think RNAi is necessarily more specific than a good ligand. One case in point:
Beyond this obvious problem, there may be others. Don't get me wrong, I think RNAi is useful; but its specificity is basically assumed (ie not adequately tested). For starters, consider that many promotors are bidirectional, many transcripts are antisense or pseudogene, and the central dogma of DNA=>RNA=>Protein is probably oversimplified. Indeed, the number of mRNA transcripts vastly outnumbers both genes and proteins. Furthermore, even when an si is specific for a mRNA transcript, if that mRNA (or any of its fragments) has multiple functions beyond encoding protein, how can its degradation be specific?
At any rate, there is a developing trend in my environment to question the specificity of RNAi. We just don't yet understand all the cellular functions of RNA.
The obstacle to discovery is the illusion of knowledge.