"Level" has to be the most overloaded word in the game development lexicon. We've got character levels, difficulty levels, level design, tech-tree levels, recursion levels in programming, and so on. Computer games are full of numbers, so almost anything with a quantitative value can be said to have a "level" of some kind or another.
I'm going to further complicate matters in this column by talking about multi-level gameplay - that is, games with different gameplay modes, one of which is conceptually inside another. In war games, this idea is usually implemented as a global mode, in which the player makes strategic decisions about an entire campaign, and a local mode, in which the player fights a battle in a limited region of the overall game world. It doesn't have to be about war, though. In my game design workshops I often give people the challenge of designing a game about running the CIA. Many teams decide that running the CIA is a lot of boring political desk work, so they include a second mode as well that involves actually going out on missions.
I need to be clear that I'm not talking about a global mode that only records the success or failure of local-mode missions, but doesn't actually have any gameplay itself. The tournament tree in a sports game, or the list of missions fought in a combat flight simulator, don't count as true global gameplay modes. They're just shell menus around the actual game. I'm talking about gameplay modes that are interrelated. To make a multi-level game, the decisions the player makes, and the successes or failures that she has in each mode must have an effect on one or more of the other modes.
So why do we create multi-level games, and what are they good for? First, let's look at board games. Board games usually have only one level. Everything takes place on the same board. In Risk, for example, you're trying to conquer the whole world with hundreds of armies, one army at a time. The entire campaign is fought at the tactical level, and it requires endless die-rolling and takes hours. More complicated board games have more interesting tactics, but they can easily overwhelm the player with data. The original tabletop version of Civilization (which inspired the later computer games) required as many as 14 stages to get through a single round of play. In the Avalon Hill's old hex-grid war games, the players had to keep track of a vast amount of information. You didn't get out your copy of Squad Leader for an evening's light family entertainment.
The benefit of multi-level games is that they separate this material into convenient chunks, and while you're in one gameplay mode, you don't have to worry too much about the other one. Fantasy role-playing games like Dungeons & Dragons can be pretty intricate, but they're played at two levels. The overall quest takes place on a big map of the world, but when the party gets into a fight, the players set up pieces on a local map and battle it out at that scale. When the battle is over, the party can go back to its larger-scale mission.
With computers, we can move back and forth between gameplay modes rapidly, and the players don't have to worry about keeping track of all the data. However, they do have to learn a new user interface and get used to a new set of challenges. Multi-level play definitely complicates your game and makes it less accessible to the casual player; it's not for people to whom "computer game" means Super Collapse! The tradeoff is that multi-level play makes your game richer and gives it more lasting value, while not overwhelming the player with too many things at once. It's also a way of providing different kinds of gameplay in the same game.
In a lot of multi-level games, there's not much to do at the global level, and the all the critical gameplay occurs at the local level. This can make the global level feel a bit pointless. One of the earliest multi-level computer games was Archon, created by Anne Westfall, Jon Freeman, and Paul Reiche III of Star Control fame. In Archon, you moved pieces on a chess-like board (the strategic level), but instead of capturing an enemy piece by simple replacement, the two pieces fought in an arcade-style gameplay mode. This is an old idea - what if chess pieces actually fought each other? - and Archon put it to the test. In chess, strategy is everything. In Archon, the importance of tactical play strongly outweighed the strategic. Your strategic decisions didn't give you that much of a tactical advantage, and it was only by tactical victories that you could win the game. If you weren't a good action gamer, you were doomed.
The key to creating a game that balances strategic and tactical play has to do with how much each one contributes to overall victory, and how they are interrelated. In Archon, tactical play was essential, but it doesn't have to be that way. Right now I'm playing an excellent example of balanced strategic and tactical play. The game is Weird Worlds: Return to Infinite Space, Strange Adventures in Infinite Space that I wrote about in September of 2004. Weird Worlds operates on much the same principles as SAIS, but the relationship between strategic and tactical play has been tightened up. In the game's global mode, you're cruising around the universe in a starship, picking up valuable objects and trading them, and trying to collect as much loot as you can within a certain amount of time. But about half of the star systems you visit are occupied by hostile aliens. They have to be defeated in tactical combat (or in some cases, wooed with diplomacy) before they'll yield up their booty. If they're too tough to beat in combat, you can retreat and avoid them - unlike in Archon, where every battle was to the death . It's OK to end a game of Weird Worlds without exploring every star or defeating every alien, but you'll earn a lower overall score.
So if you want to make a multi-level game, ask yourself how the gameplay from each mode affects the other. In Archon, only two strategic data affected the tactical play: the types of the pieces that were going into combat, and the color of the square over which they were fighting. Only one piece of information came back from the tactical play to affect the strategic: you either won or you lost your piece and control of the square. In Weird Worlds, a huge amount of information from the strategic play affects the tactical play. In strategic play you can customize your ships with new weapons and systems that you find or buy, which helps you in battle. You can also set a formation for your flotilla of ships. Different aliens have different characteristics, so part of the strategy lies in deciding when you are strong enough to take on a particular group of them. In the other direction, a tactical victory adds points to your final score, plus you get hold of whatever item the aliens were guarding. As the game is randomized each time, this can be anything from the ultra-valuable Particle Vortex Cannon to the nearly worthless Three-Headed Threep. Some of the items you can find are improved star drives and deep-space scanners, which benefit your strategic play. But you can also get points from strategic gameplay alone: exploration and diplomacy. In short, good global play directly benefits local play (you build strong ships to fight with), and good local play directly benefits global play (you get more points and useful gear if you win battles).
Another difference between Archon and Weird Worlds is that Archon is a two-player game (even if one side is controlled by the computer) in which one player is certain to win, while Weird Worlds is a single-player game that just produces a score at the end. It even produces a score if you get killed in a battle, albeit a low one. But I don't think this distinction is really important. What really matters is the inter-relatedness of the gameplay at the different levels.
As with so much in computer gaming, it comes down to a question of balance; how much time you want the player to spend in each mode. As I said before, in a lot of games the global mode is just a sort of resting point between local-mode battles. If you want your global mode to be really meaningful to the game, you need to do two things: make sure that the global mode contributes directly to victory in some way, and that it also contributes to local-mode play as well.
Games aren't limited to just two modes, of course. You could design a gigantic game about the world economy, for instance, that moves from the global picture down to national, regional, city, and neighborhood levels. The problem with going too far down is that any one neighborhood is unlikely to have an effect that makes it all the way back up to the global level again. You might get lucky -- a particular neighborhood could turn out to be sitting on top of a huge unforeseen oil reserve -- but what are the odds, out of all the millions of neighborhoods in the world? I wouldn't bother doing more than about three different levels. Beyond that and it becomes too hard to know how local events will be meaningful at the global level.