Certain games have the addictive quality of influencing your thoughts when you're not even playing them. Scribblenauts does this amazingly; having begun to play it, I've often been falling asleep, shaving in the morning, or doing any other mundane thing while running through problematic levels in my mind, trying different options. Sometimes I listen to a song and hear a noun that I think can be useful in the game, or think of using an object I walk by in real life.
There are two elements at play here I think: first, of course, the unique innovation in Scribblenauts, the ability to call up vast amounts of objects, and honestly - it's easier to be surprised by what it has than to stump it. But there's also something else, the fact that your tool-set is so clear and intuitive. The logic governing the AI is quite simple, pretty much [object] - fears/consumes/attacks/protects/follows/steals - [object]. Then there are a few other simple interactions like climb, ride, shoot, use and the ability to fly. Thanks to the brilliant hierarchical Objectnaut engine, the 5th Cell team was able to efficiently populate and interlink all these objects and actors so that adding a vampire could be as simple as adding a humanoid, giving it some art, flight, aggression towards all and fear of garlic. For instance.
Now, an immediate counterpoint to what I just said is that such an approach decreases the uniqueness and depth of individual elements, since they are all merely different combinations of the same base components. Well, that's true. Of course programming and animating unique jumps for each of your animal races will be more flashy than using the same jump logic/animation on different rigs. But while this approach can provide a very polished experience, detailed and expressive, I think its also a conscious choice to guide the player, rather than offer them freedom. A fine option, but for the point I'm making I want to focus on the latter option, creative choice, and consequent emergence.
I touched on emergence in my previous article about trigger-whoa mechanics and interacting with systems like physics and AI. Returning to Scribblenauts, there's a lot of that going on. A passable physics engine ensures things fall and roll well enough, and the aforementioned AI rules can cause real chaos when you fill the screen with living things. On top of all this, there's your vast amount of choice, and that's the kicker: for any given scenario, I'd wager there are objects that you can write which would interact with each and every different parameter of your situation.
And that's the core, really - the fact that everything is interconnected. Bad game design will leave loose ends, cool but one-off mechanics, unbalanced events requiring more fuel than exists, things like that. Good design brings it all together, makes the tokens you want advance your progression, enabling you to fuel your abilities and expand them to get more tokens and repeat the cycle, and so on. It's about creating positive feedback loops, and never letting the player down through frustrating inability on the game's part to deliver.
This isn't anything new, just basic game design. It's easily forgotten though, especially as you make a game bigger an deeper. Which is understandable, really - the more systems, bits and pieces you've got in your game, the harder it is to make them interlock neatly. What happens when the cool fire gun is perfect for gameplay with its spread and damage over time but the ability to set things on fire in any realistic way would break the balance completely in your realistic ice world? Ok, so make "magic purple non-fire". Sure, but then what's your ammo? Can't be oil or anything realistic, so what?
See my point? You either accept just the enemy burns but nothing else, or that the fire becomes overpowered, or that you make immersion a little harder by breaking realism. So admittedly, my example is somewhat exaggerated, and I can think of a few options around it, but that's the essence of it - the fact that the more there is, the more you need to piece together.
So I think this makes a strong case for the type of design and development approach that aims low with the system count and high with the options to interact with it. Physics and simple predator/prey dynamics, with tons of physics and manipulation abilities, or AI and motion control, with lots of interesting ways to influence AI with your gestures. And these are those games that, like I was saying, keep you running their options. RTS games can do this a lot, although they have become increasingly complex in their number of subsystems, and therefore harder to simulate in your mind. Other games that share ideas with Scribblenauts like Little Big Planet, Crayon Physics with their editors (software and drawing) and variety of choices to interact with simple physics. More traditional block and maze type puzzle games, of course, but those can often get boring to think about with their typical handful of ways to interact - more fun to play.
The converse of this is of course the linear mega-game with every kind of system and gameplay in it, giving you key moments and very specific ways of interacting with things. Very often these games will have far too many systems, too much happening too fast, for you to even hope to simulate them mentally. Bioshock comes to mind as an exception, a game with an eco-system that you understood somewhat once you learned it, and became able to manipulate quite well. I at least relished my convoluted traps for those Big Daddies and Splicers, exploiting both their alliances and animosities as well as their various weaknesses, but I thought on my feet, as events happened. Planning skating runs in Tony Hawk as well, for example, or running through well-practiced tracks in Burnout with all its shortcuts. For those latter examples, I think the depth of your interactions (from speed to tricks and drifting) counts as a strong diversity of ways to interact with the few game systems at play.
One particular development implication that comes to mind is that diverse and balanced interactions can often require more pre-production in the underlying infra-structure than the more immediate one-off counterparts - these can just be built sort of in order. One option has the risk of forcing you to spend too much time and effort before you've even got a product and with the other you may end up with a product that's a bloated chain of disconnected elements; so it's not like there's a better one. But I do think, like everything in life, that projects that that longer to really start are always less appealing than those that throw you right in.
And above all, I commend those people who would try something new, something that at first glance may be brushed off as too expensive/difficult, too simple or without enough blockbuster value. Those that try to build something new, and focus on richness and depth, rather than size and breadth. It's a lesson I've tried to incorporate in all my designs, no matter how commercial, that principle of balanced interconnected dynamics rather than lists of moves/events, and hope to see more of in the future. The empowerment of the player, the ability to make them feel, well, godlike, while playing will always be of the finest goals in game design, because with empowerment comes the ability for great challenge, and the ability to overcome this can lead to inner strength, confidence and self-improvement.
And I mean, really: what other game allows you to summon up God, the Devil and Death, tame a dragon with Cupid's arrow and ride it, wearing a full set of armor and lance, to attack the Grim Reaper himself? Oh, and Longcat along with alter-ego Tacgnol. :D