Odds are that you've already read something on Damion Schubert's Zen of Online Game Design presentation. So I'll point out some highlights and then list the questions and reflections I scribbled down during his talk.
Points of wisdom
First, some things he said that are worth repeating, even if they're old wisdom.
If your game is intended for long-term play, like most MMOs are, then the game's long-term potential should be obvious to the player immediately. The player must realize in his or her newb experience that the game is fun and will continue to be fun. If you expect players to work before they play, to hold off for fun later, many of them will quit before they get there. This also means that bugs have a more severe impact in the early game than later.
The more players have "invested" in your game, the more you can ask of them. As their characters are more developed, as they form more social bonds to other players, as they're increasingly engrossed in the story (not that it's ever happened in an MMO, but it could), players will become increasingly tolerant and sacrificial. Now, this can obviously sound bad -- many developers use this knowledge to manipulate players like mere money-machines and avoid the costs of quality. But a developer can also use this knowledge ethically in community management and considerations of the game's pacing.
Hardcore players are often the reason casual gamers enter the game. I'm a great example. My brother and cousin probably would have stopped gaming years ago if I, the avid gamer, didn't try to get them involved from time to time. Even after the sale, sometimes they play a game only because I'm playing with them.
Guilds (large-scale player communities) are powerful elements in the game. Helping players find the right guild is as important as helping them find the right adventure-group.
You "must control your culture." Developers can set the tone of player communities, through communal involvement but also through the game's design. Part of controlling the culture means ensuring that the hardcore players aren't exclusionary. Veteran players need to be inviting to newbs, or else you'll lose a lot of new players.
Questions and reflections
I'm going to try to be brief on these, and not try to answer them here.
One "exit point" (time at which some players quit the game) is the subscription renewal time. If renewal is always at the beginning of the month, could special events or other memorable gameplay experiences be planned for the end of each month?
Players shouldn't be expected to play for longer periods as they progress further into the game. It's true that a player with more invested in the game is more likely to be willing to play for a longer stretch, but no player should find that the game which used to fit his schedule is now hard to find time for. Early game experiences shouldn't disappear at the high end. Rather, the high end should involve additional gameplay opportunities, aside from the original avenues.
Grouping, raiding, RvR, reputation -- none of these gameplay avenues, nor many others, must necessarily involve a particular level of commitment to the game (casual, devoted, hardcore, etc.) or a particular place in the gameplay ladder. They can be provided for newbs and veterans alike, though it may be necessary to provide them differently for different audiences.
How do un-guilded newbs perceive guild leaders? How do people who are new to the MMO genre perceive them?
Is balance really more vital for "games" than "worlds"?
As important (or more) than cooperative and competitive gameplay is player interdependence. They shouldn't have to play together directly or compete towards the same goals to affect each other's gameplay.
Intrinsic rewards vs explicit rewards. Where and when does one make more sense than the other?
Ways should be found to make player communities a more explicit element of gameplay.
How can permadeath be designed to not break player communities? It seems that one answer is a health-power system similar to SWG's, in which newbs and veterans can group together.
How can certainty play into reward systems? Diablo 2 and fantasy football seem to prove that certainty of reward is not as important as many MMO designers think it is. What elements of a reward system are necessary to attract players? Must there be an explicit reward for everything? How predictable do those rewards need to be? How frequent do they need to be?
I almost raised that last question at Damion's talk, but it seemed like a long discussion and a little peripheral to his overall presentation. It was a fun and interesting talk. That's the second talk of his that I've been to and liked.