Now, this may or may not be worthwhile for games that have high performance requirements. But for games that really aren't based around pushing pixels as fast as possible, writing one game and having it be ready to ship on Facebook, iOS, Android an web browsers is a pretty cool idea.
Time-to-market is an important factor, and one of the most important things about mobile and social games is that they are not a singular event, launched and never touched again. Instead, these games are a process; you put something out there, you gather data from users (harvested automatically by the analytics you embed), you analyze the data, make changes to the game, and see what the effects are. Did it work? Great, keep it. Not so good? Change it.
Given that model, being able to get a game out quickly is important. It establishes your brand, you build an audience, you make some money... and you get the data you need to improve it.