So what is your (note irony) "objective" definition then?
When I say "defined" I'm not talking about a single-sentence dictionary definition. I'm talking about a game's success in terms of how well received it is by a given user group or similar (the market in general). And if you want a job at Blizzard or Bioware, then they may focus on your success in terms of game design, use of algorithms or otherwise that tells something about you as a developer. Just doing an opportunistic money scheme like what Zynga seems to currently be doing, may not cut it.
You don't have to look at all players, or even most players. All you have to do is decide that if a company of X size spends Y dollars and T time on development, and makes Z profit, that they're "successful."
Yes, and where do you suppose that money is coming from if there's nobody buying the game? Then you need to figure out why. What do you think defines a game's (monetary) success to begin with, if not for the players actually buying the game? And the reason why people buy the game isn't just random, they do it because they found the game interesting enough for them.
This is not threatening to some of us. However, you have a vested interest in defining things "objectively," so maybe it is very threatening to you. How are you personally going to make profitable games, if you don't have an objective criterion for how to proceed?
I'm sorry, I seriously don't get what you're talking about. You're now asking me essentially the same question that I've essentially been asking you. And what is this talk about threats? What threats are you talking about? I'm talking about whether you actually know, using sober reasoning and measureds facts, if a given game or game feature is popular or if you're just making guesses (however educated they may be).
In order to figure out what makes a game successful, you need to look at what is actually interesting to a group of people and not just individuals. E.g. simple questions like "How many players would like my game if it featured so and so?". Sometimes it's hard, but sometimes it's easy. But to actually learn it (rather than insisting that it cannot be known), it's not enough to ask individuals in the group. You need to test the group as a whole, to the extent that it's possible. If a high percentage of players like a given feature, you might keep it. If they don't, you might choose not to prioritize it.
Experimenting with unique features isn't about making the entire game super-unique and resting solely on the chance that it could potentially be a great hit. That's a sure way for it not to be. The reason why new and innovative games have been huge hits in the past, is because they manage to balance an ordinary core with extraordinary elements. That way, the game won't flop completely if those new features turn out to be terrible.
Trying to design a completely unique game in all respect and succeeding with it, is like trying to win the lottery. No serious game developer would ever do that, but that doesn't mean that you're catering to Silicon Valley either. It's a balance.