Storms on Cloud 9 #37:

Filling the Sandbox

by Scott Holliday
2004-10-29

As I look at various game development projects, I keep running into the issue of content. Game worlds are getting bigger and bigger, but without content, it's all empty space. It seems to be a bragging point now just to see how many square miles online games contain. Majestic mountains, miles of trackless steppe, huge endless deserts, forests even an ent could get lost in. The better offerings have even started to claim virtual ecologies (and bestiaries) to fit each these huge environments and fill them out. All of this is impressive, but it still is not a game.

Content generation is one of the most difficult ongoing processes in online games. I've argued before that using computers to randomly generate content can free up developer time. Of course this is true, however the results will be pretty generic. The problem is that random generation can only place what has already been hand-crafted. For example, let's look at trees. First, a developer must design a tree. If the game uses graphics, that means time spent in art and possibly collision handling. Now, using smart random generation, a computer can assemble your world's forests, right? Yes, it can… if every tree is exactly identical.

So let's back up a step. In a perfect world, instead of designing the tree, the developer would instead make a program to design trees. Assuming all goes well, now the computer can plop out hundreds of thousands of tree designs. You could even tailor this set so that they are classed by function, tropical to arctic, brushy to deepest forest. Ignoring the diskspace requirements for this library, a computer can assemble your world's forests right? Yes, it can… if every forest is just empty space filled with trees.

So start adding stuff! Spice it up with animals and monsters (that you have to design). Throw in an occasional landmark (that you have to design). If you're good, you could even have the rare NPC wandering about (that you have to design). Now, use a computer to randomly assembled all of your pieces. What do you have? Assuming all went perfectly, at best, you have a fairly bland, but maybe believable world. Once players have seen all of the variations of trees, beasts, and landmarks, they're done. Sure, they'll stay a little longer to see a taste of everything, but you still had to spend time designing each of the pieces.

So, what is still missing? Assuming you've done an absolutely incredible job, you've just simulated real forests for a whole world! What could be wrong? How could anyone complain? In short, players don't want a real forest. They want content. They want uniqueness. They want spiffyness. They want a game! You could do the exact same thing with dungeons, and still run into the same wall. In effect, once you've seen one dungeon, you've seen them all.

Alternately, you can spread out your content. Perhaps certain landmarks are incredibly rare and useful in some way? Maybe those wandering NPCs I mentioned are difficult to locate, but can offer special services or story leverage? So, now you've introduced the slot-machine idea. The tireless, the lucky, the desperate - these players now get to see a final level of the content that you designed. Sure, you might keep stringing your players along a little bit further. But you're just as likely to alienate those who aren't quite so tireless, lucky, or desperate. You could even limit it, such that only the oldest or most experienced players can even find the best sites. So now it's a slot machine that requires a minimum input before it starts paying.

So what's the answer? You only have so many hours in a day. Even using the computer to augment your abilities, content is content. Unique places and things require time to create. You could hire more staff, but each person can only produce so much. True, you might be able to reach a critical mass where individual players cannot each keep up with the mass of content that you add. This is one solution.

The other answer is perhaps better. Look instead at the game design. What do your players do? Assuming you have a standard online game, they are each ravaging the countryside, finding and devaluing the content that you've created. And yet, each of them, merely by playing is generating their own content. If players have reason to interact, their interactions are probably better content than any NPC you could have designed.

So this is where you can produce self-propagating content! Encourage your players to interact. Give them new ways to interact. Speaking and grouping are assumed procedures in most modern online games. So, expand on this! Let them create quests for each other. Let them construct their own environments and monsters. Let them assemble their own factions and interests. Then encourage this. Acknowledge their designs by adding to it with your own content and attention. Including a player's content in your story is the highest praise for a player builder. With a few relatively simple tools, your players could be adding content that is just as interesting and valid as your own. In fact, considering the imagination and devotion of some players, many of them are likely to be better at the process than your own staff.

Following this exact line of reasoning, several games in the past few years have described themselves as a sandbox for their players. This is especially true in the text-based online game offerings. No story. No plot. No events. Up to a point, the developers leave all of that to the players to create. I think they have the right idea. Developers are there to add to the mix, perhaps show how things can work, and so get the ball rolling. Unfortunately, these sandboxes often don't come with the needed player tools… and a sandbox without sand is just a box.

[ <— #36: Cliché Practicality | #38: What Went Wrong —> ]