As soon as a game creator says "I'm going to make an MMOG," for instance, more than half the design is locked down purely by the expectations that label carries in the market. We see this in BioWare's Star Wars: The Old Republic. Every time the game is shown, the game's creators say "everything you'd expect from an MMO game" will be present. "Of course it will have crafting," they say as if there was never any doubt. "Of course it will have guilds, of course it will have [insert genre convention]." And while I don't necessarily think that this particular game shouldn't have those particular features, I don't think their inclusion should be justified simply because "it's an MMOG." Just look at the loot system in Champions Online. Super heroes don't beat up villains to take their stuff but because looting is such a powerful part of the MMOG genre, the mechanic is shoehorned into a setting for which it is obviously inappropriate.
When gameplay stops contributing to the overall context of a game's fiction (or, worse yet, actively suppresses it), it has ceased to serve a useful purpose. I'm not saying every game has to be Moby Dick, and I openly acknowledge that there are some games, particularly sports or puzzle games, where the mechanics essentially are the fiction. Still, look what an inventive context was able to do for the simple tower defense game Plants vs. Zombies. I think we wrongly assume that using genre based on the way you interact with a game makes sense because interactivity is the one element that most separates gaming from other forms of media.
Another part of the problem is that genre definitions based on the interaction and presentation of the game are vague enough to be confusing. At the most basic level, for example, almost every game is a simulation because almost every game simulates some kind of reality. From football to fishing to flying to finance, any game that represents a real world system is, by definition, a simulation. Likewise, most games are designed to put the player in a discrete role or responsibility, so most games have some element of roleplaying. You could even argue that all games are strategy games to the extent that they require players to collect and contextualize information to improve the quality of range of their interactions with the game world. The best games blur the lines between distinct genre divisions. Is GTA a driving game? Yes, but not just a driving game. Is Borderlands a shooter? Yes, but not just a shooter. Is Mass Effect an RPG? Yes, but not just an RPG.
So what purpose do genres serve, anyway? Some will say that the purpose of genres is to categorize the wide range of titles available, but that's just begging the question. To paraphrase C.S. Lewis, the description may be true, but it's hardly worth saying. The real point of genres, at least so far as I see it, is to give us a framework of associations that suggest what else we might like and a vocabulary that helps us discuss those associations with other gamers. The purely mechanical definition isn't serving those purposes particularly well.
As the experience of playing has become secondary to the mechanics, we've become like children who can't tell the difference between something that merely looks like what we want and the real thing itself. Sure, the traditional game genres are useful as a means to organizing and discussing gameplay elements, but they shouldn't ever become the primary focus of our understanding or enjoyment of games or limit our expectations of the types of experiences games can deliver. It's time to stop asking, "What kind of game is it?" and start asking, "What is it about?"
Steve Butts still can't figure out what genre his life is supposed to be.