Computer Graphics Volume 34, Number 2, May 2000
Gaming & Graphics

Are Virtual Worlds Worth It?

Are computer games any more fun now than they were ten years ago? Surely they have improved considerably in terms of technology, with flashier visuals and generally more immersive gameplay experiences then ever before. Maybe I'm just a cynic, but the games just don't seem to be any more fun.. One of my own games, Centipede 3D, is a remake or "modernization" of a classic game, the original Centipede as created by Ed Logg in 1980. The new version of the game has a much flashier, more involved, and more graphically lush game-world, but the game-mechanics are largely the same as the original Centipede. Is the game any more fun for having a virtual world? Quite the contrary; I'd say it's less fun. If anything, the 3D world the player gets to navigate in Centipede 3D distracts from the core gameplay.

It would appear that I am not the only one who wonders about the direction games are heading, and whether it is a good direction to take. Game developer and historian James Hague, who Computer Graphics readers may remember as the guest-editor of this magazine's May 1998 issue, graciously agreed to return to the magazine to explore one of his own pet topics: are virtual worlds worth it?

- Richard Rouse III

Are Virtual Worlds Worth It?

by James Hague
Dadgum Games

In the beginning, games were abstract, surreal, and just plain weird.

Looking back on a game from almost twenty years ago, Nintendo's original Donkey Kong for example, there's much that only made sense because of the vaguely representational nature of the graphics. A big ape vandalizes a skyscraper under construction, and somehow manages to make the topmost horizontal girder slope downward to the right, the next girder below slope downward to the left, the one below that downward to the right, and so on. It's not physically possible to animate those kind of deformations on a model of a building. The designers only got away with it by omitting the vertical pieces that the horizontally aligned girders would have been attached to.

And there are other bits of strangeness, too. Barrels roll along the length of a girder, so the player can only see the end caps of each container, yet for some reason they rotate to face the player lengthwise when they roll down a ladder to the girder below. From a 2D side view this sort of makes sense—okay, no it doesn't—but it doesn't jump out at the player as the contrived impossibility that it is.

The Rise of the Tourist Industry

As game graphics improved, there was a tendency to give the player more things to see. This was only natural. No matter how well drawn, Pac-Man or Missile Command only contained one screen. If you expanded the game to include different levels, each with its own set of graphics, then was a big hook for the player. Ads could feature shots of different screens. The player could have the satisfaction of seeing a new screen for the first time on his or her computer. Donkey Kong, again, was an early example of this, with four different levels of custom graphics and animation. Miner 2049'er took this to the next stage with a then unbelievable ten screens when it was first released circa 1983. Ten screens! How would you ever see them all?

The levels in Miner 2049'er obviously took a lot of effort in terms of programming and layout. There was often only one correct way to work through a screen, with misleading "Surely I can jump from here to there?" platforms everywhere. One level included a cannon that the player could jump into, scoot back and forth, and be launched across the screen with. Another featured teleporters that were activated with the 1, 2, and 3 keys. Others included slides. A vat of radioactive waste. Shuffling platforms. The author, Bill Hogue, didn't whip out a level in an editor and plunk it into the game in a few minutes. Jumpman, for the Commodore 64, was a later game in the same general vein, but with even more numerous levels. Quite possibly the author of Jumpman was trying to one up Mr. Hogue.

With memory capabilities growing but programming not getting any easier, there became much more of an emphasis on giving the player new terrain to fly over or walk through, even if the gameplay wasn't directly affected by the current set of graphics. This led to the themed levels that were a mainstay of the ever popular "jump on stuff" platform games popularized by the Super Mario Bros. series. The desire to have widely varying levels that kept the player wanting to see what was coming next spurred a series of cliched scenery sets that no self-respecting game was without: a jungle or rain forest level, a volcano level, a waterfall level, a dark and dripping cave (or a mine with railed carts), and a level covered with ice, snow, and penguins. And I'll be the first to admit that it was fun to get to a new level for the first time. "Hey Bert get up here! I made it to the underwater level! I made it to the underwater level! I—oh no, oh no... died. Never mind!"

Along the way, game characters transformed from blobby eight-by-eight pixel critters into more human like forms. Starting in the mid 1980s, games were more often than not based around controlling recognizably humanoid forms, games like Kung Fu Master, Street Sports Basketball, and Strider. Today, in real-time 3D games the player's on-screen puppet is almost always a human, be him the "Quake Guy," Half Life's Gordon Freeman, a personified Bandicoot who goes by Crash, or some nameless Joe riding a bike in Motocross Madness.

This evolution has culminated in the general formula for a 3D game design: create a 3D world that's broken up into discrete chunks (levels), develop a character, and then let that character run around the 3D world doing something to be specified later. The something isn't quite as important, or as daunting, as the first of these items: constructing a 3D world. This isn't a condescending description at all; it's more a statement of the level of technology and artistic freedom that we have in front of us. We can create worlds. We can create characters. And we can let our characters loose in our worlds. But is it worth it?

Building the Pyramids

An underlying assumption in the development of 3D graphics is that virtual reality is the ultimate goal. It's rarely mentioned outright, but that's what we're shooting for. Oh, we like talk about how gameplay is more important that graphics and wow, it sure would be great to work on a game with the focus on fun instead of frame rate, but that doesn't stop us. Each round of games has more techy stuff behind it than the last batch. Straight texture mapping won't cut it, there need to proper shadow maps. Wouldn't curved surfaces that tessellate on the fly be handy! Characters have to cast anatomically correct shadows that project on nearby geometry. Volumetric lighting sure looks better than twiddling with vertex colors. How about light that properly refracts in water? And wavering heat pockets over lava; we can't forget lava. An up and coming buzzword is "unique texturing," so each polygon in a world can be painted on directly without altering a base texture map. And so it goes.

This is a difficult—maybe impossible—habit to kick. All else being equal, a game with nice shadows is obviously going to look better than one that doesn't have them. Trying to be a retro miser and doing without is only going to make reviewers scrunch up their noses and reach for the "OUTDATED" stamp. And doing this kind of research is compelling. Who wouldn't want to be the first to pull off a crazy new effect, such as a bleeding character walking into a pond and having the blood diffuse into the water?

Technical issues aside, the biggest change that fully 3D game worlds have brought on is a dramatic increase in the amount of time it takes to bring a title from concept to finished product. The rendering engine may be complete, the AI and interface and tools written, but then there are the twenty to fifty levels needed. Even given a game of Doom vintage, thirty to fifty levels can soak up a whole lot of time. Assuming a really good level jockey can pump out a top notch Doom map in three days, then that's eighteen to thirty artist weeks of work—over six months in the latter case. And Doom is simple by modern standards.

Consider a more complex 3D world like that of Quake III. Nobody's going to create a polished and exceptional Quake map in under two weeks, and that's being conservative. We're up to sixty to a hundred weeks of work. Almost two years for one person.

Now go one step further and get away from games whose levels can be put together like so many Lego blocks. Imagine creating a fully textured, fully custom world in a real 3D modeling package. How long would a synthetic museum take to create, complete with landscaped grounds? Or a multi-story castle? Depending on the detail, four months may be realistic. How long for twenty levels? Over six years. Notice, too, the number of levels is even below the original target range. And four months may not be realistic.

What's important to realize in this example is that the emphasis is squarely on creating the virtual worlds themselves. After spending four months modeling and texturing a mansion, there's strong incentive to use that work in the game, because it's difficult to throw out something so expensive in terms of time and money. Developing worlds at such expense can result in awe-inspiring experiences for the player, but as worlds get more realistic and more complex to create, the effort gets further and further away from the original goal of creating games. Is it worth, for example, taking eight months to put together an absolutely brilliant computer rendition of the White House just so someone can run through it, guns a-blazing, on his way to the next level?

The opportunity cost of sinking a dozen person years of time into world creation is steep, both in terms of lost game development time and bitterness. Yay! After four years the 3D masterwork is complete! And Barbie Girl Gotta Groove is outselling it fifty to one. Is there no justice? Sometimes there isn't, and hitchhiking across rural Canada is the only option.

Lost development time is more nagging. Somebody's going to set the industry on its ear. Somebody always does, every four or more years, somebody that nobody's heard of. And when that new big thing comes along, being locked two years into a three and a half year development cycle is going to be murder, especially if that new big thing doesn't follow the "game = world + characters" equation that's the current norm. Really, there are other options. The view that games have to run along a preset path to virtual reality is hopelessly limited. Ah, the wonder of PaRappa the Rapper for the PlayStation, of The Sims for the PC, of Crazy Taxi for the Dreamcast, or of a yet to be written game about a lunatic who mail orders everything and never steps outside.

The Limits of the Level Designer

At one time the term "designer" implied "programmer." Really, how full of hubris could one be, to think he or she could jot down a game concept on a cocktail napkin, then hand it off to a coder to do the grunt work. Or perhaps, more correctly, how could this person think that someone else could take a loose idea, craft it into a finished game, and expect that game to embody the spirit of the original inspiration. Yes, eventually the career of designer became separate than that of programmer, but there's still the fundamental truth that a designer is somewhat of a guidance counselor who can help make decisions and steer people in the right direction, but ultimately isn't able to make all the small creative choices that result in the bulk of the game. A designer may say "this level takes place in a shopping mall," and give some general input about the features of the mall, but it's the person putting together the levels who decides that there should be a fountain in the middle and how that fountain should look. And somewhere down the line, when a player is being chased by mall security guards in the shipping game, the player is going to have to decide whether it's quicker to run around the fountain or to try to hide in the Macy's dressing rooms. Something out of the designer's control had an affect on gameplay.

Issues like these have become so prevalent in game development that a new term has come into being: level designer. A person who works from the guidelines set by the game designer and programmers and puts together the actual worlds that the player wanders around in. This isn't some grunt task, but the job that bridges an engine and a game. "Level designer" is an impressive term on a business card, and rightly so.

A side effect of this division is that there may be one designer, six programmers, and fifteen levels designers on a good sized project, and it's hard to shake the feeling that the direction of a game is largely free floating until the level designers—the people with the most time consuming task—have done their work. Does the combination of game engine and editor reduce to a construction kit from which dozens of games can be created? If so, it is the job of the game designer to design such a construction kit rather than a finished game. By definition, this eliminates a large class of games that don't fit the construction kit model—consider such classic non-electronic games as poker or baseball. If baseball were designed by a game development house, one inning would take place underground, the next on a rock floating in a pool of lava, and the next on Easter Island. And the bases would always be in different places, and maybe there would be ten bases on one level spaced hundreds of feet apart and dinosaurs roaming the outfield of the next. Face it, the game of baseball is hopelessly outside the realm of modern computer game design. Maybe if we did away with the whole "pitcher throws the ball" thing and just had a guy named Hank who ran through levels in a striped uniform swatting critters with a wooden bat?

Okay, that's over the top. But there are certain design considerations and rules that come into play once you start down the "player in a world" path, and those considerations tend to conflict with games that don't fit that model. In many ways, the desire to architect explorable worlds inside of the computer is different than what can be categorized as game design, the selection of rules and interactions that result in something that's fun to pick up and play. Creating 3D virtual worlds for players to run around in, that's an amazing endeavor, but a different one, and it should be categorized as such. One that has different time sinks and payoffs. And depending on the particular situation, one that might not be worth it.


James Hague has been designing and programming computer games since the mid 1980s. He founded developer Dadgum Games in 1996, and is currently with Volition, Inc., as lead programmer for the PlayStation 2 role-playing game Summoner. An avid electronic game historian, James is the author of Halcyon Days: Interviews with Classic Computer and Video Game Programmers and maintains an associated web site at He can be reached at