As with any modeling venture, it's hard to build synthetic models of real networks that are simple enough to yield to enough analysis to produce insights, and yet are close enough to the true process for the insights to carry over to practice. I will present a few short vignettes about real Internet phenomena which I think have properties that are robust enough to allow for meaningful synthetic modeling, and yet have not been studied with theoretical tools. Along the way, I'll delineate my personal view of the strengths and limitations of synthetic network models, with more emphasis on the limitations than most of us are willing to write down in submitted papers. I promise to present no actual technical results, and to not say anything that doesn't sound obvious or over-simplified to at least a few in the audience. I'll do my best to make sure some of my open questions can feasibly be cracked by the end of the workshop (or the coffee break), and that some others can inspire a dissertation or two's worth of time investment.