This post comes from ongoing debates on the MUD-Dev mailing list regarding the advantages and disadvantages of working at different scales in online world design.
I’ll say it flatly–right now, the innovation, for better or worse, is coming from the graphical side of things. Yes, there’s a TON of catch-up work to be done, and basic fundamentals that they (we?) are missing. But what we’re seeing is an explosion of basic typologies–and muds divided into a few typologies so long ago that a new one hasn’t been invented in literally a decade.
Of these two, chat spaces and games are by far the largest. Some of the chat spaces are roleplay-themed and some are not; some of the games are and some are not, whatever, but the presence or absence of roleplay is more a major typology differentiator, to my mind.
These typology differences correlate very strongly with particular code bases; we readily associate MUSH/MUCK with chat, MOO with building, and MUD with the two types of games (competitive and cooperative).
The first wave of graphical muds in the 90s was actually centered around the chat model, not the game model. The Palace, WorldsChat, and similar products were chat spaces, not games. I don’t know current stats for The Palace, but Cybertown has 60,000 users right now, I hear.
You mention group departures; social spaces are particularly vulnerable to this because, after all, roleplayers can do their thing with just IRC. Chatters can do it with darn neara any communications medium. The chat spaces failed to offer significant value added and were not, by and large, able to monetize the presence of the users.
The building model was also attempted, most notably with AlphaWorld. Providing the building model in a graphical setting is difficult, and AlphaWorld relied too much on users creating their own playspace without providing an overriding purpose. It didn’t go much of anywhere as a result, as it was a directed enough experience.
The competitive game flourished in smaller scale. Competitive games tend to reset in muds anyway, and have ever since the days of the “scavenge everything and drop it in the well” muds. Quake et al played perfectly into that. However, they also did not monetize retention.
The cooperative game turned out to be what unlocked the wallet of consumers. It’s strongly retentive, offers significant value added for remaining in one game for the long term, and provides barriers to exit in that there are in-game assets that you cannot migrate elsewhere and are loath to leave behind.
The building model was successfully co-opted to a limited extent by Ultima Online, and given the success of that limited co-option, almost every game since has followed suit to one extent or another–in fact, I’d say that the integration of a crafting paradigm into the basic advancement game mechanic is much MORE widespread in MMOs than it is in muds. Similarly, the competitive game has also been integrated in limited fashion in just about all of them; just as in the mud world, it’s proven to be a small subset of the game playing audience within the context of an advancement-driven game, probably because good competitive game mechanics depend on parity between contestans and an emphasis on player skills as opposed to statistical advancement on the part of characters.
I know many MMORPG designers who were mud developers first. UO had developers from LegendMUD, NarniaMUSH, TooMush, StarWarsMUD (the LP). Meridian 59 had developers from Worlds of Carnage and other muds. The litany goes on and on. It’s been argued convincingly that the problem with many of the first generation MMORPGs was that they were too mud-centric in some critical ways; the mud folks brought much-needed expertise on multiplayer game dynamics, etc, but there weren’t enough people with experience in commercial services and previous generations of commercial online games.