It’s possible to broadly classify features within an MMO:

• Micro features concern a single player or character, and may not extend beyond the player’s client. Examples: character stat displays, the user interface, inventory, most quests.
• (Normal) Features concern multiple players or characters, but do not significantly change the world state. Examples: combat, looting, respawning mobs, vendors, alternate movement (flight, swimming, horses).
• Macro features concern the entire game world, with significant and lasting effects on the world state. Examples: player housing, territory capture, unique artifacts, the Sleeper (Kerafyrm) encounter in EverQuest, Horizons’ and EQ2′s unlocked races.

Difficulty of development of these features is proportional to how macro or micro the feature is: a character display in the client or packet to update world state is relatively straightforward, while an encounter like Kerafyrm is a complex balancing of script, world impact, rewards, risks, etc.

Game development has traditionally pushed macro features to the end phases of MMO development, in part because they rely on many subsystems functioning properly. How macro features are implemented is one of the biggest differences between single-player and MMO games: in a single-player game, the world altering event can be as simple as a flag (and in cinematic or “rails” games, old content won’t be revisited so it won’t have to reflect the change). In an MMO, other players need to be made aware of the change, which may involve the developers scripting reactions in otherwise unrelated parts of the world long before the event takes place. Many games, like DAOC, don’t script macro events into NPCs or game world at all and instead rely on custom reports or user interface elements to reflect the change in world state (think, /realm).

Redefining what “MMO Generation” means
Are micro, normal, and macro features useful for identifying first, second, and third generation MMOs? UO, a first generation MMO, had significant player housing, and at times, ways players could significantly impact the game environment (such as placing a house on a spawn point). EQ had no significant macro features at release–the biggest impact players could have would be to kill Lord Nagafen or Lady Vox and have to wait a week (or until the server reset) for the respawn. Even with a number of expansions, EQ is a relatively static world. EQ2, on the other hand, has macro features as part of the patch cycle, such as the recent disease storyline. Horizons, like EQ2, had many macro features, especially player towns and server-wide quests (that coincided with the storyline put out with monthly patches). CoH, like EQ, has no significant macro features.

It would be fair to say that games with more macro features are more complex, better simulations of worlds and more deserving of the title “second generation” MMO. With those criteria, UO, Horizons, DAOC, and EQ2 are second generation, while EQ, CoH, WoW, Guild Wars, etc., are first generation because the long-term player impact is minor (leader boards, at best). Third generation contenders, like Vanguard, may actually be first generation games with a lot more shiny and better or more complex AI–characters are still living in groundhog day worlds.

So, if rather than defining generations of MMOs by date that they were released, MMO generations were defined by how many macro features they implement, we end up with this list of characteristics:

• First Generation: Minimal player impact on the game world, usually reset over time or when the server is reset.
• Second Generation: Limited player impact on the game world, restricted to building, crafting, or questing. All macro feature activities take place within well defined limits set by the developers (you can’t dig a tunnel to the satyrs just anywhere, but only in three specific locations).
• Third Generation: Strong player impact, with engaging content created by players. Should include limited scripting. Is the Saga of Ryzom with the Ryzom Ring system moving here?
• Fourth Generation: Dependent on player work to create the game, with tools for graphical and scripted manipulation. Environments like Second Life, while not games as such, are fourth generation for the amount of player customization possible. The MMO game disappears in favor of player created games.
• Fifth Generation: A dynamic space, with portable models, code, and trust relationships. In this generation, the user becomes the service provider, minimizing publisher and developer on-going costs.

By definition, in first and second generation games, a player exercising a macro feature will prevent another player character from taking a related action, even if it’s as simple as purchasing a plot of land on which to build a house or being able to safely enter a recently captured fortress. This may act as a significant barrier to inclusion of macro features in favor of giving every player a chance to try all of the game content.

Why don’t we see more second and third generation games?
There aren’t more second and third generation games out there because of risk, cost, and game content. It’s risky to attempt to implement macro features: they can break an otherwise successful game design, and can introduce difficult (or impossible) to correct bugs. The time required to design, implement, test, and debug these features increase the costs of production. Finally, as the game industry puts more money into game content, with large palettes of detailed artwork and 3D models that require the latest video cards to display well or quickly, professional voice acting, and other high cost, high production-value assets, average, creative players can’t compete with the professional artists for quality of output. Allowing players to create new game content can lead to interesting legal situations (such as, who owns the content? The creator/player, or the game company that is now hosting it?). Finally, don’t discount the belief that all players should get a shot at game content: permanently killing the ice giant king means that only one group gets to experience that success–other players will be bitter.

What about MUDs?
Many MUDs are second, third, or even higher generation games: players may be able to impact the MUD world significantly, especially those with built in building tools (especially MOOs). Of course, prose is much, much easier to develop and host than 3D graphical worlds. MUD developers are proud of showing how they’ve had macro features in their games for a decade that major MMOs are only beginning to approximate.