The first approach, and the one which is most common to games, is to start "from the gameplay first" or top-down. By this, I mean that when approaching a game, a developer will try to decide what their gameplay mechanics, themes, and general player experiences are before they go into constructing the universe it takes place in, the characters that are involved, the storyline specific to that game, etc. This approach is the one that the "classic" console game developers, such as Nintendo, Konami, Capcom, Sega etc. tend to take, with the aesthetic side of games built after the mechanics are made solid. It's also how Valve Software, one of the premier Western developers, handles their own games. I'd like to think that this might be, at least for those who want to make "just games", the more flexible approach, and I'm sure that a lot of the developers who follow it will agree with that statement.
Zelda's world is expansive and engrossing, but it's also
just a fancy coat of paint for its real focus - gameplay.
As an example, take The Legend of Zelda series of games. While it's true that the series has a fairly robust cast of characters and fiction, the truth is that Zelda has very little to do with these bare aesthetic elements. "Link", "Ganon" and "Zelda" are only integral to the Zelda series insofar as they are icons we associate with the games; essentially, they're good for marketing and brand recognition purposes, since they are familiar to players. Strip these away, though, and you'll find that those elements of the games are little more than surface details. The real meat of the series is in the combat, puzzles and exploration, all perfectly balanced with each other and honed to a point. A lot of the effort that goes into the Zelda games has very little to do with "Zelda" at all. While this is a testament to the quality of Nintendo's internal development teams, it also demonstrates just how superficial Zelda can be. That's not to discount the quality of the Zelda fiction, but rather, to draw attention to the way that Zelda's world is built around its gameplay, and not vice-versa.
There's little doubt that taking this approach to game design has its major strengths. Not having what may be seen as arbitrary limits on what you can do from a development perspective is extremely freeing - you can have any gameplay element you want, and you don't have to worry too much about it contradicting your fiction or your world. If there is a contradiction players are likely to notice, you can usually come up with an explanation of some sort, and most will be satisfied (such explanations can be as simple as adding a small bit of plain text, maybe in the form of a book, or an off-hand line of dialogue, but can be much deeper if the developer wishes). It also means that, at least early on, your gameplay can determine what direction your story, characters and universe go. Sci-fi setting not working for you? You don't have to worry about scrapping hundreds or thousands of hours of work put into building a world if you haven't built it yet.
Of course, there are some downsides to this as well. Namely, you risk sacrificing a cohesive world, as I mentioned earlier, something which I think is highly integral to the success of a game. This can be circumvented, of course, if a developer is mindful, but too much disregard for consistency in story, premise, world, characters and all those elements will leave a game feeling disconnected and out of sync. Sometimes, players will overlook it, especially if it's a case of them being "just there for the multiplayer". Modern Warfare 2 may well be one of the best recent examples of world building done wrong, with vague conflicts and character motivations all around. Infinity Ward's attempt to create cool and sometimes shocking scenes and scenarios largely backfired by focusing too much on the spectacle and too little on the reasons for events happening. Players might like pretty explosions, but it's important that they care about, much less understand, what is happening in the game and why.
The second approach I'd like to mention is the opposite: starting world design first and building your game on top of it, which I'll refer to as bottom-up. Rather than taking what you think might be a good idea and then trying to shoehorn it into a larger framework, instead, it's the framework that you start with. This is often a huge amount of work, but the attention to detail and care that goes into this is also something that really shows through in your game. Not only is the player exploring a ruin, for instance, but it's a ruin with strong history and situated within a larger world, which isn't just implied through vague imagery, but that can be explained clearly by the writers of the game. When it comes time for a developer to really hunker down and build a game, chances are they'll have a very strong idea of what it can and should be if they opt to build their world this way.
This is Dragon Age's world, Thedas. It's big. You can bet on
the fact that the majority of those locations all have strong stories to tell, too.
A more recent example of a game taking this approach is Dragon Age: Origins. The game took about five years to make, and it shows. The depth and detail in the world is stellar, from the world map looking like a real piece of topography and not something built for gameplay convenience, to the locations and even architecture of cities and towns, to the various competing in-game religions, to the racial tensions between different characters. Despite the small number of Qunari characters in the game, there is a long and storied history chronicled within in-game texts providing a logical explanation for their rarity. Even the premise for the story itself is something grounded in thousands of years of fiction, and story events of Dragon Age itself are just small pieces of that fiction. Although there are a few rough spots here and there, like the isolated village of Haven, overall Dragon Age can be seen as a modern testament to what having a cohesive world can bring.
The benefits of this approach are pretty obvious. Not only does having a strong universe add a lot of colour and flavour to a game, it also means that characters, story, locations and situations are all situated as part of a larger whole, and thus are likely to make more sense. As players, we get a real sense of immersion from being able to see that, say, a certain community was built because of a particular reason, which we can then investigate, and so on. Sometimes it can provide shorthands for storytelling - if the player is informed in advance of certain circumstances, designers can rely on them to recall that knowledge later on in order to cut down on what might be flow-damaging exposition. On the gameplay side of things, it means that developers can find a style that's right for them, that fits in well with their fiction. If a world is a place with large amounts of history, a huge world, and dozens of important characters, for instance, maybe a slower-paced game would work, while a fast-paced, glitzy, high technology science fiction world may be more suited to a shooter.
Unfortunately, this approach to building a game world also has a number of major downsides to it. Creating such an expansive place can be a big problem for artists and designers because it means having a large, authoritative document that needs to be updated on a regular basis as details change. Things need to be hammered out in stone as early as possible so that time isn't wasted on creating gameplay, plots and environments that have no place in the world. Using the world to fuel your story and characters might well be limiting, and could even lead to use or abuse of stereotypes or obvious situations, for fear of more exotic choices being "out of place". And, most importantly, a huge and consistent game world means that a whole, whole lot of work is spent building things that may not even appear in the game at all. While this is true of pretty much everyone involved in creating a game (artists make thousands of sketches, level designers scrap things they have put hundreds of hours into, programmers work on features that never get used or seen by the player, etc.), for a writer, this isn't necessarily "part of the job" in the same integral way it is for those other roles.
Grand Theft Auto is a fairly rare example of a game whose world
and gameplay are intertwined so strongly that they are nearly inseparable.
I don't want to argue that one is inherently better than the other. I think that both have their place depending on the goals of a particular game and the type of market it's intended to reach. If you're building a role-playing game aimed at hardcore fans, for instance, it's probably best to start with a world and work your way up, because it's those reasons that will keep players interested, invested and coming back for more. For a shooter, it might not be as important to justify what's happening; if your gameplay is strong enough, people are willing to overlook a lot of flaws, although visual appeal is also highly important to that genre. My own preference leans towards building a world first, gameplay and story second, but that's largely due to the sorts of games that I enjoy. I acknowledge there's room for both.
Another thing that I don't want to suggest is that these two approaches can't coexist, or that developers have to make a binary choice between one or the other. Shifting back and forth at different points in development might prove useful depending on a game's needs, and having different parts of a team focus on one or the other is something that's going to happen simply due to the nature of their roles in development. Level designers might still be more focused on creating fun things to do and see, while writers are inherently geared towards building an encompassing fiction. That still doesn't mean it's not a good idea to give level designers a strong sense of the fiction and world, though, and writers should be involved in creating gameplay so that they have a better sense of what is interesting and fun for players to experience. Of course, if you're making a sequel, there's going to be some limits on what you can and can't do regardless of your approach.
Most importantly, though, is that this isn't a formula for making good games. It helps, most definitely, to have a solid idea of what you're doing either from a gameplay or world-building perspective before jumping into things, but that doesn't mean that a developer can ignore one or the other and hope to get things right. A game world can be extremely compelling and interesting, but if the game is no fun to play and doesn't feature any good ideas or mechanics, it's still a failure. This even extends to the story: a good universe with nobody interesting in it, no meaningful plots or conflicts, no emotions there for the player to experience, etc. is just going to fall flat. A bad game is a bad game regardless of how strong one or two individual elements might be. Players can, and usually will, forgive a few mistakes, but in the end, a game has to be able to stand up on its own as a complete experience for any of this to matter.
[Image credit 1]
[Image credit 2]
[Image credit 3]