A lot of little theories run around about the fundamental nature of game design. Games as sequences of meaningful choices, or as skill training, or whatever your favorite is. I don't like these, because to me game design is artistic, and therefore any rigid mold you try to put it in is something that should be broken as a matter of artistic expression. When you say "game design is about presenting meaningful choices", someone will counter with a game like Bioshock Infinite, which basically had no meaningful choices but was still highly regarded.
With that in mind, let's talk about other theories of game design that might shake the status quo a little bit. Here's one:
Game design is about giving the players striking moments.
When you think about your favorite games, in general the things you remember are the most striking moments in the game. Sometimes these scenes are parts of the story - for example, the early Final Fantasies were built around having half a dozen fundamental striking scenes and then just sticking game between them. Final Fantasy VI is the best example of this. The story itself was not terribly well executed, the graphics were primitive, the gameplay was so flawed that nobody noticed they'd turned off several stats and left them turned off... but there were foundational scenes which punched you in the face. The epic opening, the choice of which chapter to play next, the burning of Figaro followed by the sinking and rising of Figaro, the quiet and inexplicable western music when you walk into a bar and suddenly see Shadow, the beserk espers, the rending of the earth, Celes' grinding, hopeless island quest, and so on.
These don't really craft a meaningful "story" as such. Even if you removed half of the pieces, you could still have exactly the same story. Figaro's ability to submerge and tunnel through the sand adds nothing significant to the plot... but it does add an amazing, striking set of scenes and some conceits for some minor missions.
Later Final Fantasies remembered this basic idea, but have put the player on rails. We'll talk a bit later about how pacing is dreadfully important, and how they drop the ball.
The same can be said of basically every game you remember loving that had a significant story... but lots of games we loved had little to no story. For example, XCom, Street Fighter, Mario, and so on. So many of these games did not have many (or any) plot-related striking moments.
But they still have striking moments.
Programmed right into the game progression.
The first terror mission you face. The first time you encounter M. Bison as he floats down. The time you manage to stay alive despite getting 6 Z-shaped bricks in a row at the highest Tetris difficulty. The time when you managed to jump across eight flying turtles, or found the warp gates. The first time you send a star fleet of your own construction into enemy territory.
Some games also have multiplayer moments. The time you managed to get all those headshots. The time you and your friend held out against twelve enemies. The time you wouldn't stop hitting each other in the back of the head at the worst possible moments. The time when you dressed up as ridiculously as possible in a contest to see whose avatar could be made more outrageous. And, of course, the time you managed to jump a Ferrari over the river while your buddy did handstands on the roof.
I've been thinking a lot about these striking moments, and the nature of games as a vessel for striking moments. I've been thinking about them in part because - and this is something I can't stress enough - I think we're not using them well enough. There are a lot of kinds of moments we're not using because we're not thinking of them in this way. Not all striking moments have to be about murder or subverting the game's implicit seriousness.
So, if games are simply a delivery vehicle for striking moments, why aren't they just sequences of striking moments?
Well, for the same reason movies aren't. The gameplay and pacing of a game allows us to establish the striking moment as striking. Taking on a single demon in God Hand is an impressive moment, but in most other games it's barely even notable. Similarly, a woman being killed by an evil villain happens in virtually every M-rated game - but it's only when that woman is established as a major character that you get something with the emotional punch of Aerith's death.
This is something that modern spectacle fighters are skirting the edge of. Every time you press X to rip a gorgon's head off, you're supposed to feel it is a striking moment. But it isn't very striking to most people because the game hasn't set it up with any weight. Some people will still find it striking, but only those whose heads are in the exact right spot at that moment. Most players aren't going to feel exactly like you want them to feel, or expect the things you think they are expecting.
This is where games really shine, because you can really guide the player into the exact spot you want them. By crafting the characters, plot, and gameplay, you can teach the players exactly what to expect... and, therefore, you can strike them perfectly with your striking moments. A good example of this is the poisoning of Doma in FFVI - that's the scene where Kefka poisons a whole city. Not only is poisoning of that sort something that is obviously villainous, but when Kefka originally orders it, he is countermanded by someone who is (at the time) in the running for main villain. That's right - one of the major villains thinks it's beyond the pale. So when Kefka does it anyway, it's not just a villain event horizon moment. It's so beyond what is acceptable that the villains turn against him, throw him in prison, and make peace with you because they are so horrified by what he did. That context really sets Kefka up as a massive, once-in-a-game-generation memorable villain.
Of course, you can also do it with gameplay. The 20-kill streak you're so proud of in Team Fortress 2 is a good example: the rules are set up such that it is difficult to pull off, and everyone knows it is difficult to pull off. So pulling it off is a hell of an accomplishment, even though in many shooters a 20-kill streak is something you achieve five minutes into the first stage.
The key to setting these striking events up is to shepherd the player so that their expectations are aligned in just the right way for the moment to hit hard. Self-directed pacing helps a lot, here. For example, that gameplay moment where you get a super kill streak isn't one where the designers tell you "okay, now aim for a 20-kill streak". No, they simply allow it to be possible, and then, when you are ready, you will move into it on your own and quickly realize that a hell of a moment is developing. Similarly, the developers don't say "dress as silly as possible and show your friends!" They simply make it possible to dress silly, and possible to show your friends, and you quickly realize that there is potential to be had.
This is a little more difficult with story elements, but the same philosophy applies. The gameplay between scripted moments should be crafted such that playing the gameplay sort of tugs the player into alignment. Players that are out of alignment will generally continue to play the gameplay instead of progress through a story beat, and therefore be brought into alignment steadily.
The RPG is the great example in this. An RPG's play serves two purposes. It is an outlet so that a player can let off statistical steam... but, simultaneously, it takes place in a place and for a reason which aligns the player with the next story beat. When the player is eager to level and tweak stats, they can put off the story and do so... simultaneously, they are primed for the next story beat, because every moment they spend wandering the town or screwing around in the forest is another jot of investment into the town and the forest and the people within it. At least, when well-designed.
More recent Final Fantasies show that you can fail to do this. These Final Fantasies should, by all rights, be full of striking moments. But I literally can't remember any of them. None of them were primed properly, because the gameplay was too static and linear. I wasn't allowed the time to be brought into alignment - the gameplay either had too little vectoring, or was too rigid to allow me to let off my statistical steam.
Now, with all that said, I mentioned that I think we're not using moments to their fullest.
By that I simply mean that our emergent events are limited to violence, skill challenge, and subversion. I think we can allow for more kinds of moments than that. Think about the plot moments that strike you. These often involved impressive places or things. Sometimes it involved people dying, but usually not because you were killing them. These moments are scripted... but couldn't we create similar moments through gameplay?
The basic concept is that the gameplay establishes a vector, and allows the player to get invested in the fundamental concept... and then we throw in a twist or a challenge option which makes their head tilt. As a basic example: teach the player how to fight, then offer a pacifist route reward. However, in our case, we want to talk about concepts rather than skill challenges.
We could talk about places. Teach the player to cut trees and make houses... then show him houses built right into the trees, made without harming them.
We could also talk about people. Teach the player to treat NPCs as people... then let them actually act like people, and confound him.
The first is easier, because we can rely on multiplayer for fuel. By allowing players to build things and then share them, you can introduce players to massively impressive works of art built with the same resources they themselves have. However, you do need to take care with pacing. Striking moments are things which require you to be on the right vector - vectors established through gameplay. Simply throwing content at players will drown them in content, not create striking moments. The ultimate example of this is Spore, which drowned in its own content overdelivery. Recalibrated so that the gameplay could pull the players into the proper vector over time, introducing new alien species could have made each species intensely interesting and feel unique. But delivered willy-nilly, they all blend together into mush.
The second option - people - is harder. There are several reasons. 1) We don't have any established frameworks for allowing emotional interactions. 2) People are going to want to turn it into porn faster than they'll want to turn it into meaningful relationships. 3) Creating unique content may require a lot of effort.
One option is to simulate it using multiplayer, just as before. You can create a framework where NPCs are controlled asynchronously by polling several players for responses with mirror techniques. For example, the NPCs they interact with are built out of the massed actions of other player characters, while their own responses to that NPC will form the NPC responses to those other players. This could theoretically have a lot of potential, but you'd need to do a lot of innovation because there's no standard practice.
Another option is to embrace the porn in the same way an RPG embraces level grinding, and assume that when the player is getting tired of porn they'll move into the next personal arc moment. I dunno whether or not this would work, but it'd be difficult to convince anyone to take it seriously.
Another option is to have an algorithm which controls the NPC responses. This doesn't require a very smart algorithm, any more than the algorithm for putting bounties on kill streaks does. All it requires is carefully planned out algorithms that pull players into the vector properly. However, because the uniqueness of the algorithmic responses is low, you need to make the gameplay highly variable, again like TF2. No two games are the same, so your 20-kill streak might be extremely impressive and striking... or a bland reminder that there are a lot of newbies floating around. By making the gameplay vary wildly, you can make the context vary, and therefore make the moment vary as to how striking it is. Good gameplay will constantly tug the player into alignment with at least one of the possible striking moments, so if you embrace the chaos you can create a sort of "radial" moment system where the players might aim for different kinds of moments depending on the situation they find themselves in.
In the same way as someone in a fighting game might aim for a kill-streak... or they might aim for a perfect, or a handicapped run, or a carebear tutorial run, or a hat-achievement run, or...
These are still fundamentally skill-type challenges, but I think a little baby step could be taken with this concept. See where it goes before running off at full steam.
Anyway, those are my thoughts on the matter. Sorry it ended up so long.