You Keep A'Knocking
by Dave Rickey
Watching new games come out, and new intiatives in old games, can be a bittersweet process for me. Especially, seeing the extension of the craft of online game design unfold the way I expected is simultaneously a source of frustration and validation. Frustration because I can't help thinking "I wanted to do that 3 years ago," validation because, after all, it *worked*. I wasn't crazy, I wasn't drawing impossible plans for castles in the air, but in fact had an idea that was practical, useful, and within reach, but just a little too hip for the room.
I'm not going to say *what* series of feature announcements for which games triggered this; the people who actually made them happen first deserve unfiltered credit for their achievements. Rather, I'm going to go on record with a list of future advances I expect to see become standard over the next three to five years. Some of this may be obvious, some of it may not be obvious for a while, some may turn out to be flat wrong.
1) Distributed computing based on blade-architecture systems using OSS
components (OS, DBMS, and others) will become the standard server
This architecture will center on blade computers because the economics of the blade design are obvious. It's standardized equipment, using PC components with all the economies of scale that implies, and with all of the user-friendly bells and whistles of a PC stripped out to lower the price. Cycle for cycle, you get two to five times as much processing bang for your buck, without getting locked into a particular vendor or having to teach your programmers a totally foreign environment.
The basic plumbing will be Open Source Software for similar reasons. Nobody seriously considers writing a custom Operating System anymore, and few would really want to use a home-brewed database. And where using standard commercial software will add to your overhead and lock you into license terms that may restrict your game design later, OSS is free as in beer and free as in speech--you can do anything you want to with it on the server side without ever having to let anyone not working for you see your source code. This will be a win-win for both the game developers and the OSS systems they adopt.
This standard architecture will be centered on "entities" or some equivalent, small processes representing mobs, players, spawn zones, and other specialized logic that run on blades and communicate with the rest of the server in standardized ways, handling things that apply to a very narrow focus and relying on messaging protocols to receive inputs and dispatch outputs. These message protocols will do the heavy lifting on keeping track of what is happening where, but will contain no logic on what to *do* about any of it, the logic of the entities will all be at the edges, decentralized as much as feasible (truly emergent systems will be a lot longer in coming, if ever, but we'll start faking it better).
I'll go into more detail about how this will work in some future column, for now we'll leave it a black box.
2) AI will become increasingly important, and increasingly
However, while we've been creating armies of thousands of marching morons, academics and defense-industry programmers have been working on related problems with different incentives. In the academic world, Artificial Life is nearly a solved problem, with many different proven methodologies for reproducing lifelike behaviour in silico. And Defense Department combat simulators have come a long way since the days when packs of kangaroos busted out SAM's and started shooting down simulated helicopter gunships.
Seven years ago, Ultima Online's attempt to create an artificial ecology failed. Although there were some fundamental flaws in their model (for example, the lack of aging or natural death combined with unlimited individual creature potential leading to super-trained unkillable predators), in fact the real problem was that the number of top predators (player characters) the system was designed for was exceeded by a factor of ten in the actual live game, who then went out and killed everything in sight, reducing the world to a wasteland. But the failure was so spectacular that it has been nearly impossible to even bring the idea up in a design meeting without it (and often the presenter) being laughed out of the room.
Instead, squads of world builders have populated worlds one database entry at a time, and hand-tuned the parameters of very simple Finite State Machine AI that was otherwise standard to all creatures in the world. However, the size of world that is expected by the players, and the level of detail demanded by the players, is quickly creating a situation where creating those worlds by hand will be prohibitively expensive. The generation of games now on the verge of launch will probably be the last to have worlds built by painstaking hand craftsmanship. Future efforts are going to have to turn to AI and A-Life based approaches. There's actually a large amount of prior art just in single-player games for this: Will Wright and Peter Molyneux have been building AI based software toys for years, and I've mentioned in a previous column that we could take a lot from strategy games to improve ours. Again, I'll have to wait for a future column to explain the nuts and bolts of this.
3) Character art detail will plateau, soon.
So we're going to see a sudden halt in the increasing realism of character graphics, with a standard emerging somewhere close to that of CoH or SWG, and the emphasis shifting towards greater customization and personalization of almost cartoonish characters in increasingly realistic worlds.
4) Someone is going to pre-empt eBay.
It doesn't have to be that way. The game operators already have the player's credit card numbers, and can secure the transaction just like any other trade (taking a cut, of course). Nobody is certain exactly how much real money is trading hands over virtual items, but it is pretty certain that if the more obviously trustworthy game operator was acting as guarantor of the transaction, that number would skyrocket.
For a long time the game operators have been edging up to the line, wondering how close they can get without triggering outrage in the general playerbase. Linden Labs (Second Life) is out in front on this, having worked out an official arrangement with the "Gaming Open Market". Sooner or later, someone is going to grab this danger/opportunity with both hands, and if they get away with it, the others won't be long in following.
5) Cable/DSL Providers will get more involved in online games.
Going out on a limb and making firm predictions like this is risky, but except for #5 (which depends on an industry I have little exposure to and therefore don't understand the politics of), all of these seem as inevitable courses for the industry to take over the next few years. Check back with me in 2009.