Ian Schreiber recently posted a summary of the Southern Interactive Entertainment & Game Expo (SIEGE). One bullet point in particular seems a brilliant encapsulation of things I've tried to enunciate in the past:
Instead of saying "I'm going to create a game that causes a particular emotional response in the player," start the other way around: find existing game moments that produce your desired emotions, then extrapolate to figure out what mechanics can cause those emotions.
Basically, build emotional experiences from the bottom up, not the top down.
I find designs (both for games and software in general) often get mired with excess structure and stumble when trying to find robust solutions without understanding the full space of the problem. There are simply too many details, too many subtle complexities. Spending a lot of time and effort on the initial solution, only to discover some aspect of it doesn't work, means even more time and effort will be needed for revisions.
I realize this is nothing new to anyone familiar with the ol' waterfall vs. agile chestnut. And even though agile has become more prevalent, at least on the software side of things, it still seems that too many experiences are built from the top down. Designers imagine a profound moment that will provoke an emotional response in the player and then steam toward that. If it's later discovered to be less poignant than intended, various components need to change to move closer to the desired response. But by the time this is discovered, there's already a ton of content (models, environments, dialog) that has already been created and probably other design dependencies. Only so much can change and likely that desired response will never be reached.
The net result of this is evident in countless games. There's a character the player is supposed to empathize with and care about, but instead of organically building rapport, the game just beats the player over the head. For all intents and purposes, there's a giant arrow and flashing neon sign saying "Care about this person!" Of course, we usually don't. And we end up with things like the discovery of Dom's wife in Gears of War 2.
The solution, or at least a step in the right direction, is what Ian's summary captures- find something that works and build upon it. Ideally, those seeds will be in your own game already. But if you have to bootstrap off a similar moment in another game, that's fine. The point is to get something, anything, moving you in the right direction.
Once you have something that works, build upon it. Figure out why it works and how similar moments can be created. In attempting to evoke emotion, there are so many variables it's difficult to predict what will and will not be successful. Rather than (somewhat arrogantly, in my opinion) assume you know what will create the desired response, there will be more success in remaining dynamic and humble.
I'll keep harping on the importance of iteration, prototyping and playtesting, and it's because it has dynamism and humility built in. Realizing you're going to get lots wrong and end up throwing a bunch of stuff away is fine if you're building cheap prototypes and learning from each one. Constantly having those prototypes evaluated and discovering what comes up short will definitely come with a hearty dose of humility.
Obviously you want to have enough foresight to not paint yourself into a corner. A design that's just a constellation of provocative moments without any latticework holding it together will feel disconnected and shallow. Perhaps a few people do possess the vision and intuition to get it right from the onset. But that's a very small number of people and many more think they belong to this cadre than those that actually do. I know I do not possess those necessary qualities, and thus, I cleave to iterative, bottom up design.
Editing this, I realize it's pretty vague and hand-wavey. Rather that transmute this post into Bigby's Unsurmountable Wall of Text, I'll follow this up next week with at least a pair of examples of where this technique worked, and where avoiding it did not. In the mean time, if you have any thoughts on bottom up emotional design and where it has/hasn't worked, I'd be curious to hear it.