Tuesday, September 29, 2009
Aion and the death of the traditional server model
Players are notoriously unable to agree upon what features a MMORPG should and shouldn't have, and what level of technical difficulties are or aren't still supportable. But the one thing they do agree upon, the absolute minimum requirement, is that you have to actually be able to log on and play. Aion does not fulfil that basic requirement, not for the large part of the customer base which doesn't have hours of time to wait in a queue before playing. Melmoth says:
And to all those who have said that this is the smoothest launch they’ve ever seen, of course it bloody well is, nobody can get onto the servers to stress them. I could solve all of the public transport problems in England if I only let ten people on to each bus and then thanked everyone else for their patience while they stand around for two hours to get on to the next one.after having found that there was a login queue for every single European server, of up to 2 hours. That is clearly unacceptable, and will drive a lot of players away.
Funnily by driving players away NCSoft has created a self-fulfilling prophecy. It appears as if the low number of servers have been created in a Darkfall-like attempt to only let in the most dedicated players, and thereby keeping the player numbers stable over a longer time. Opening up sufficient servers for launch risks the fate of WAR and AoC, losing a lot of players after the first months, and then having to deal with empty worlds, which according to Brad McQuaid is much harder than dealing with overpopulation.
But of course the root of that problem lies in the traditional server model. Why, oh why do we have to select the name of our server first, thus being tied to that server, and unable to play if that server is full or down? Maybe for fantasy games the one-server solution of EVE Online isn't feasible. But have a look at a game like Free Realms, where you create your character first, and select a server later. You can always jump to the server where your friends are on, without having to use the server with the same name every time. And that solution can easily be downscaled, as how many servers are on offer is determined dynamically. Less players result in less servers open, and the number of players per server always close to an optimum.
Whether you decry them as "WoW tourists" or simply realize that the overall market size for MMORPGs has grown due to World of Warcraft, fact is that there are now millions of players out there with a certain familiarity with online role-playing games. And these might well be willing to try out a new game, so any new game must be prepared for a strong initial interest. Which is an opportunity, not a threat. Because if the game is actually very good, most of these players will stay. If they were determined to go back to WoW anyway, they wouldn't spend the money to try out the new game. But every game, even WoW, loses players one day. And a scaleable server model deals much better with that, both with permanent player losses and with temporal dips in player population due to summer holidays. And the scaleable option is also far superior with regards to public relations: Nothing hurts a game as much as announcements of server mergers. So why not just avoid that?
I sure hope that the experience of Aion teaches game companies to start looking for a different technical solution. The traditional named server model just isn't suitable for today's requirements any more. We need more flexibility to avoid both queues at the start and server mergers near the end of a game's life.