The book "The Lean Startup" by Eric Ries helped me understand what an MVP is about. It is about learning. So 'viable' doesn't mean someone can barely use it to do X. It is more like an experiment that should answer a question (an hypothesis).
So an MVP can be crappy, as long as it helps to answer the question it is supposed to answer. The lovable part should come, once you understand your environment and can build something that people actually need.