How to Make Virtual Worlds

Theory and Practice

How virtual worlds ought to be put together and how they are put together are two different things. You can spend 30 hours in a classroom learning how to drive a car, but fifteen minutes at the wheel is going to teach you a whole lot more.

At times, the practice is more useful than the theory.

This section discusses some of the things that look like they should be important, but aren't, along with those that look like they shouldn't be important, but are. It's a bit of a mixed bag, but is fairly representative of the kind of hidden depth (or lack of same) that only becomes apparent when you actually develop a virtual world. There are plenty more, but I'll leave them for you to discover them for yourself; as I said, you'll learn more by doing than by reading about it.

Modes

When you start up your web browser, what page does it point at? Surprisingly, for most people the answer is "whatever my ISP set it to when I installed its software." Their first view of the World Wide Web every morning is what their ISP shows them.

Virtual worlds have many ways to let players customize their experiences. In a textual world, for example, some people want full room descriptions the whole time, and some want short ones the whole time. There are commands that let players choose for themselves which of these to go with. However, one option will be the default, and that's the one that newbies will use. In MUD1, the first time you entered a room you saw its full description, and on later visits you saw the short version. This helped stop newbies from getting lost. Later, they would use exclusively verbose descriptions when making accurate maps and exclusively brief descriptions at all other times, but later is later. When they started, they got the default, and the default said, "Explore!"

Defaults set the tone of virtual worlds, because all newbies play under them. As they become more experienced, they'll inevitably customize some of the settings; most options, however, will stay at the default. Thus, the designer's choice of defaults can have long-term influences on how a virtual world is perceived. Defaults are more important than they look.

To illustrate this, let's take a look at modes, because in a sense they establish what the world is "about."

All virtual worlds assume a hardware device for entering freeform data; it's usually a keyboard. The player types some text, hits return, and—what? It depends on the current mode. In a textual world, the line is usually interpreted as a command.

open door
north
get book

This would be command mode.

In a graphical world, the line is usually interpreted as speech.

Follow me!
What happened? You were supposed to be following!
Hey! That's mine!

This would be conversation mode. Chat rooms usually default to conversation mode, too.

So where's the hidden depth here? It's not like you can't act in conversation mode or speak in command mode.

When a newbie enters a world for which the default is command mode, the message that the world is sending them is that this is a place where you can do things: It emphasizes freedom to act on the world. If the default is conversation mode, the message is this is a place where you can communicate: It emphasizes freedom to interact with other players. You might expect, therefore, that players of textual worlds do more and players of graphical worlds say more.

Actually, for graphical virtual worlds the default is to use the less-than-freeform mouse most of the time, forcing a greater distinction between limited doing and unlimited saying. If you want to talk or attempt anything complicated, you have to stop playing to do so. This is sound advice for crowd control, but somewhat dissatisfying for the individuals in the crowd so controlled.

The designer sets the default mode. The default mode shapes the style of play. The designer can therefore encourage or discourage a style of play by changing the default mode. Thus, a simple, almost throwaway design decision can have a long-term influence on a virtual world's ethos.

Note that although the main choice is between command mode and conversation mode, there are other modes used in virtual worlds. The convention has evolved that the first character in a line is used to switch modes for the remainder of that line. There will often be an option to turn a mode on/off until further notice, but no standard syntax for this has yet emerged.

Table 2.1 shows the most common modes in use, along with the (sometimes conflicting) options for the leading characters used to switch them on.

Table 2.1 Common Modes

Virtual Reality

Read a selection of Science Fiction stories about virtual worlds, and you'll soon discover that the following are inevitabilities:

• True intelligence will emerge from the hideously complex machinations performed by the virtual world engine, with unnerving consequences for human morality.

• Unscrupulous people will transfer their consciousnesses into hardware in order to live forever, muahahaha.

• Virtual worlds will be experienced through virtual reality interfaces so good that the virtual will be indistinguishable from the real.

The first two of these are not of immediate interest to the designers of virtual worlds, being somewhat distant prospects at the moment. Virtual reality (VR), however, while as yet nowhere near the quality envisaged in speculative novels, is far more accessible. Why not create a virtual world with a virtual reality interface? It would attract media attention, if nothing else.

For a virtual world with a closed user base, a VR interface is indeed a reasonable proposition. A small academic research community or a large industrial or military training establishment would be able to experiment with the idea and get fruitful, worthwhile results.

For an open user base, though, VR isn't yet an option. It's simply not value-added enough for developers at the moment, that is, the costs of putting it in outweigh the benefits. Until the technology improves⁴⁷ and the installed base reaches some critical mass (whether because of computer games, 3D movies, 3D video cameras, or something else), VR would be available to but a few, lucky or wealthy players.

There is an argument, though, that VR access to a major virtual world could itself be enough of a draw that people would be willing to acquire the necessary hardware to play. Of course, for this to happen VR would have to bring something quite special to the virtual world experience.

So what would that be? It depends on the set-up, of course, but in the first instance the chances are that a basic VR kit would mean full-vision headsets with surround sound and orientation detection, plus gloves incorporating movement sensors and some degree of positive tactile feedback. Using such an interface, the virtual world could

• Let you see it in 3D⁴⁸

• Let you hear it in 3D

• Tell in what direction you're looking

• Tell what your hand is doing

• Deliver sensation to your fingers

• Accept speech input

What advantages would accessing a virtual world through such a VR interface confer over the traditional mouse and/or keyboard approach?

There's some convenience in being able to control what you see and do by means of head and hand movements, countered only by the slight inconvenience in having to sport the equipment needed to support it. Nevertheless, it's unlikely that a player using a keyboard and mouse would be at a serious disadvantage compared with someone kitted out in the modest VR set-up described here, at least in terms of their ability to control their character in a virtual world.

Speech as an input form seems, at first glance, to be an attractive proposition. It wouldn't be all that great in command mode, because speech recognition software still has a long way to go before it could be of genuine utility. In conversation mode, though, it would be much more effective—but only if it could be completely disguised. Again, although voices can easily be distorted using today's technology, it's likely to be some time before they can be altered so well that they don't sound altered. But why is some form of disguise necessary anyway?

One of the main attractions of virtual worlds is the ability to be whoever you want to be; anything in the virtual world that anchors you to who you are in the real world is a disincentive. When you hear an elf in the middle of a sylvan wood speaking a New York accent, or a mighty-thewed, bare-chested barbarian who sounds like a schoolmarm, sadly reality is intruding a little too much. If voices aren't disguised, players aren't disguised, and then the virtual world is just another aspect of the real one.

3D vision and sound, and to some extent tactile feedback, are the real gold for VR. They make the virtual world more persuasive, which helps players immerse themselves in their (virtual) surroundings. The change won't be to everyone's taste—some people are always going to prefer text, for example, on account of how it speaks to the imagination rather than to the senses—but it'll help anyone who gets on well with graphics⁴⁹. That said, flat images are already quite capable of immersing players into a virtual world, and they don't have to cut off great swathes of the real world to do it. Players like being immersed in virtual worlds, and will happily ignore negative cues if they get sufficient positive ones that they can will themselves to suspend their disbelief. You don't have to trick them into it; they'll go for it anyway. Giving them more 'realistic' visual and auditory stimulation will provide additional signals, but are they mere luxury? Designers can and do encourage immersion by many other means⁵⁰.

So, would VR merely amount to a marketing ploy to attract newbies?

Cynics might suggest that this is all that adding graphics to virtual worlds ever did, but even they would have to concede that it worked. The allegation is unfair anyway, in that a graphical interface to a virtual world does actually make a tangible difference. If, for example, in a textual virtual world you happened on a gathering of 50 characters, you'd have to read 50 names to see if there was anyone there with whom you wanted a chat. Spotting a friend from among 50 faces in a graphical virtual world is far, far quicker⁵¹. Would a VR interface add some genuinely useful feature that flat graphics and two-point stereo don't?

Frankly, probably not. Unless the virtual world has very counter-intuitive physics (for example, the further away an object is, the bigger it appears), you're not going to learn much more about it from experiencing it in 3D than you can already figure out from the perspective and motion conventions of 2D.

Unless...

Unless designers can exploit the extra capabilities that a VR interface imparts. Instead of clicking where you want your arrow to land, you raise and point your bow, then release the string—with true depth to your vision, you can now judge trajectories. Instead of running to attack the lightly armed man, you flee—you can now see he's a giant. Instead of walking in the gaps between well-spaced trees, you push through dense forest—the tree trunks no longer merge to look like fences when they're close together. VR really does offer new prospects.

Unfortunately, virtual world design is not yet sufficiently developed to make the best of this. Game designers still have to make more use of sound as a gameplay element, so it seems unlikely that virtual world designers will successfully embrace VR the moment it becomes available. There may be some centerpieces that show off the technology, but that's likely to be all. When you catch a movie on TV and notice that people seem to be pointing or throwing things toward the camera a lot, pretty soon you realize you're watching an old 3D movie in 2D. Directors never really came to grips with 3D in movies. Will designers fare better when VR comes to virtual worlds? Or will it be a case of, "Aww, man, not the guys with pikes again!"?

The theory is good, but the practice may take some time to measure up to it.

Extensibility

Virtual worlds are designed such that they can be extended over time.

Why? To add content, to correct imbalances, to allow for more simultaneous players—there are lots of reasons.

Who extends the world? The live team.

Who created the world? The dev team.

Is the dev team a subset of the live team? At some point, the answer has to be "no."

For large-scale, graphical virtual worlds, the answer is usually "no" immediately, because the live team is a separate entity from the dev team. Even for small-scale worlds, or those for which the live team is built from the dev team, the original designers aren't going to be around indefinitely. Students who create virtual worlds eventually leave college; professional designers who create successful worlds are offered new opportunities⁵²; people move, their circumstances change: Ultimately, no one lives forever.

The live team may therefore differ in attitude to the dev team⁵³ when it comes to assessing how the virtual world "should" work. There are many opportunities for divergence:

• The live team may fail to understand aspects of the design, seeing flaws where there are none.

• The live team may misunderstand the dev team's intentions, believing they're doing the right thing when they're not.

• The live team may have a different overall philosophy, and "correct" the design where it runs counter to this.

• The dev team's design might fail when exposed to real players.

• The dev team may have higher quality staff than the live team (or vice versa), with a consequently better handle on things.

Because the live team is in control, there is great scope for a virtual world to shift away from the designers' original vision over time. This isn't necessarily a bad thing—adapting to circumstances is how systems evolve, after all. Neither, though, is it necessarily a good thing—survival of the fittest is great when you're one of the fittest, but not so great otherwise.

At this level, it's a classic conflict between theory (what the dev team wants) and practice (what the live team gets). It's a little more complicated than that, though. The live team has to deal with players, every single one of whom believes they know just as much about virtual world design (if not more) than anyone in the live team—and are prepared to argue the point.

The bad news is that players know nothing about virtual world design. Nothing whatsoever.

Well, that's not strictly true. A very small fraction of them do⁵⁴, but these are generally indistinguishable from normal players except in the benighted eyes of people who actually do know about virtual world design. Message boards are full of erudite arguments by players able to put their opinions cogently, politely, and convincingly. That doesn't mean they're right, though. It's like listening to a religious discussion between people of a religion different than your own: They obviously know exactly what they're talking about, in great and profound detail, but from your point of view they're at least misinformed and at most completely misguided. Player discussions are frequently like that: Designers can recognize some truths in what is being said, but these are so mixed with dogma, rhetoric, and downright falsehoods that the conclusions they reach are often bizarre and irrelevant (whenever they reach conclusions at all, that is).

Yes, I realize I've just insulted about three million people, here.

It's not that most players don't know about virtual world design, but that their knowledge is too personal. As mentioned in Chapter 1, players tend to view all virtual worlds in the context of the one they "grew up" playing. If a new idea is suggested, many players will immediately consider how it would fit into their preferred virtual world, whether or not the virtual world for which it is intended is remotely similar. If the debate actually concerns "their" virtual world, they'll figure out the short-term repercussions of their own playing style and use that as a basis to decide whether they're for or against. They'll only refer to long-term effects or other playing styles when they're trying to win allies or to convince the live team that they are responsible people whose opinions should carry weight.

This is because actually playing a virtual world adds a subjective element to all discussion. Designers have to be objective. If you can play a virtual world for fun, it's very hard to be a designer; every decision you make is related to your own experiences as a player. Designers can't play virtual worlds for fun. When I enter a virtual world, all I see is the machinery, the forces at work, the interactions—it's intellectually interesting and can be artistically exciting, but it isn't fun. Other designers are the same: The price you pay for being able to deconstruct a virtual world is that of being unable not to deconstruct it. Magic isn't magic when you know how the trick is done.

That's why most players aren't good at design. They still sense the magic.

Unfortunately for the live team, that's not quite how the players themselves see it. Players want improvements made to their virtual world, and most of the time they are appreciative—even fanatical—of the live team's efforts. When they aren't, though, oh boy, do they ever let the live team know! The pressure can be phenomenal. It can reach the stage where it's more gainful to implement the change that everyone is screaming for than it is to answer all the emails they'd send if you continued to hold out.

At this point, the live team often surrenders. Top-down design gives way to bottom-up experiment.

Whether this fills the original dev team with pride or despair depends on the extent to which they'd planned for its occurrence.