The evolution of virtual worlds

Masses of data

Looking at all of these games, there are some common weaknesses. The first is that the volume of data needed to generate a 3D world is so high that servers and broadband pipes inevitably choke on it.

Entropia gets around this problem by downloading the entire game environment to disk in a massive 1.3GB package. Second Life takes the opposite tack and downloads objects and art only when they're needed. Kaneva and Multiverse use methods somewhere between these extremes, but none of them are perfect.

Both bandwidth and server speed will have to increase by at least a factor of 10 before game performance becomes acceptable – and then just watch how they blast through many ISPs' bandwidth caps. At 20MB/s, you can easily reach the top of a 40GB bandwidth cap within a week, so caps will also have to rise by at least a factor of 10.

At the server end, these games can make make extreme demands of commercial bandwidth. Kaneva, Entropia and Multiverse are relatively simple and therefore not too demanding, but Second Life's model thrashes its database servers until they weep and beg for mercy. Until commercial bandwidth prices go down significantly, its subscription fees look set to stay high.

Speeding up sims

There's a more basic issue that affects playability. Game areas are typically split across servers, or at least run as separate server processes. In Second Life you can't fly or walk across sim boundaries reliably – turn up the graphics requirements and you can find your avatar coasting along with plenty of momentum but no control. Move fast enough and you'll crash out altogether.

Technically, when a player moves from one sim to the next, their avatar and inventory links have to be copied across and deleted from the old sim. The most popular sims run slowly because they're busier. Meanwhile, the rest of the grid waits in limbo, wasting most of its processing power. The majority of sims are empty most of the time, making this an incredibly inefficient way to run a virtual world.

Any other game that uses a similar split-sim approach will suffer from the same problem, but help is at hand from a project called Darkstar. Darkstar isn't a game, so you can't log in, and it's still in beta. It's designed to handle the back end of online gaming – process assignment, asset management and the nuts and bolts of database handling for multiple players with multiple objects in multiple environments.

When it's finished. Darkstar should solve the sim problem, allowing an entire grid to redistribute its processing power evenly wherever it's needed. Artificial sim boundaries will become a thing of the past and databases and inventories should both scale smoothly. The potential is explosive because Darkstar will make it possible for anyone to add their server to an open gaming grid and contribute both their own world designs and extra processing power, which will be available on demand.

Building the worlds

So what does it add up to? The state of the art isn't necessarily impressive today, but there's huge potential for the future - and it's not necessarily in the obvious places. If you take Project Darkstar's open source approach and combine it with the user generated content and coding of Second Life, along with the social networking and trading features of the other games, you get something that's completely new and addictive.

From the time of Doom WADs onwards, players have been customising their games. What makes multi-user worlds interesting isn't so much that they're 3D and multiuser – so are many other games – but that open customisability is an integral part of the experience. For some people, trading and chatting are less interesting pastimes than being able to build new environments from scratch. For others, the opposite is true.

So far, game and world developers have defined how a particular world works and hoped that users will like it enough to buy it. But the ultimate game is a blank canvas for gamer-created experiences of every kind. All that's needed is some server space and a set of rules for linking worlds to one another. And with Darkstar and Second Life's Open Grid initiative, which allows anyone to run their own server from home, this is already starting to happen.

What made the web interesting was what bean counters like to call 'low cost of entry' – basic web space was never expensive and only relatively simple and cheap tools are needed for development. Most of the bigger, more popular web apps, including Facebook and MySpace, developed from simple starting points. So the magic formula for success is a low cost framework that means anyone can join with the minimum of effort, but which can be expanded almost infinitely.

That's the fundamental principle that turned the web into a global phenomenon in just a few years and it's not hard to see how online worlds could go the same way. Once everyone has high-speed broadband running at 20MB or more, it's going to be easy to park open source server software on a spare PC and build your own virtual homestead, space station, art installation or night club.

It's also going to be possible to create stores that sell real-world goods and services to complement the virtual goods and services that are already available. The money problem isn't solved yet, but you can be sure that someone somewhere is already thinking about it. Once bandwidth starts ramping up again and one or more cash-handling interfaces appear for developers to start adding to their projects, expect to see an explosion of interest and a thriving combined virtual and real economy.

This might sound unrealistic, but most of the current development cost for gaming goes on developing 3D engines and art. If an open source 3D platform arrives, it will leave developers free to work on art and scripting and there's no lack of enthusiastic amateur talent for both. So if you've written off online 3D worlds as chat rooms with eye candy, watch this space – a few years from now they're where we'll all be.