Bill Murphy recently discussed his thoughts on the end-game experience, and I do think he faithfully captures some of the ideas gamers have about it, but I'm writing this because I don't think it goes far enough in examining this critical gaming issue. Murphy touched on some of the important issues - gear, that feeling of completion and yet the simultaneous feeling of having no pre-defined path (leaving Murphy asking "what do I do now?"); however, he skipped discussing raids in any shape or form, but I think these can be grouped in with the same content issues that Murphy was having problems with anyway.
In any MMO, there is a system of training that goes on that trains not the character but the player. The game has several responsibilities to this training. The first of these is to teach the player to be literate about the game world, meaning how to use skills, how to craft items, how to travel around, and so on. Some of this literacy is more or less generic between MMOs (or between all PC-based video games in general) and thus is unneeded by long-time MMO players entering a game. If you're a longtime gamer yourself, you may recognize things such as role names (tank, cc, healer, dps), keyboard shortcuts (WASD for character movement, I or B for inventory management, J or L for the quest log, numlock for autorun), quest givers with icons above their heads to tell you they have quests for you, similar icons for quest turn-in, quest hubs, resource gathering, crafting trainers, class trainers, leveling, attribute points, and so on. This literacy expresses itself most obviously in games such as RIFT that seem to be trying hard to utilize the preexisting knowledge of the MMO community by making everything look and feel like other MMOs. To this end, many games use some kind of hint boxes or explanatory quests at the beginning of the game to tell players exactly what they should be doing.
The second system to train players is based on Pavlovian theory, where a dog is given a treat every time a bell rings, so that eventually the dog associates the ringing of the bell with the food to the point that the food doesn't have to be present for the dog to start salivating. Games begin rewarding players early on with levels that increase the player's strength in-game and enabling them to use more skills. This goes back to the first point of training - skills are distributed to players slowly as they level partly as reward and partly as training to keep from overwhelming them with skills. (This process was explained in detail by Dr. Brian Cowlishaw.) The result is that we as players are taught that we are good little boys and girls when we do what the game wants us to do, which is to quest and kill monsters in the constant pursuit of XP. It is not fair to say, however, that XP is the only reward. While we level, we're also given quest rewards and random drops of equipment that is, fairly often, better than what we currently have. This equipment continuously gets better and better and is helpfully color-coded to show us exactly how impressive it is. This process further reinforces the Pavlovian response, but it serves a secondary purpose: to prepare us for end-game.
So once we reach end-game, the long and slow questing process that got us to our last level is finally yanked abruptly from us - we have no more level-up to encourage us to seek out the next quest, and we are left with the knowledge that we will receive no more such leveling reward for our work. This is why games become a gear reward alone - the gear provided by end-game instances is the last method of providing reward (many games have now added titles and achievements to the mix, but that is merely another collectible thing, and thus functions in much the same manner as gear). For many of us, that type of reward is not enough. We were trained to appreciate the levelling reward, and the gear reward is not sufficient to keep our interests.
(A quick aside: this system of game-company-driven rewards obviously refers only to extrinsic motivation. Intrinsic motivation - the kind each of us creates within ourselves based on our own desires and beliefs - is not a generated function of the game but is instead internally-created within each player and thus cannot be controlled by the game company. It certainly provides motivation for a great many people. There are intrinsic motivations of completionism, for instance, that will lead players to do quests far below their level, to accomplish every deed, every achievement, and every title, that will lead players to go through raids for the sheer potential of adding that one more notch to their belts. There is also a significant intrinsic motivation toward making friends in-game and helping those friends with their own achievements. More on that part later. The point is that these motivations would exist with or without the game's presence, and regardless of what the game company does with the game. Thus, it is only the extrinsically-motivated players that a game company can control and thus must be the target of the company's future investments. These types of players are the fundamental reason why expansions allow for higher level caps - the possibility of bringing the players back and sending them through more questing & rewarding them with more levelling.)
There are additional problems with the gear rewards. First, the only compulsion to achieve them is the ability to enter into a raid which itself can only provide the reward of more gear. Thus, gear only leads to more gear, and it doesn't lead to any other greater sense of purpose or accomplishment (excepting the possibility of intrinsic motivation). Second, the gear improvements tend to be tiny, as without end-game gear, players still need to be able to participate well enough to get through the quests that would lead them to such gear, but with the gear they cannot be so overpowered as to simply slide through all of that content. Third, because the end-game must last until the next expansion can come out and increase the level cap, such gear must be hard to obtain. As micro-expansions and updates occur and players get closer to the next expansion, that difficulty can be decreased significantly, but it must always return to extreme difficulty for the new gear coming with the new expansion. This difficulty ensures that players who are motivated to complete the raids will be spending a great amount of time collecting gear to reach that point.
All of this, however, fails to consider one of the other major problems of end-game - grouping. While some MMOs have developed great systems for organizing groups (WoW and DDO are two that spring to mind), most MMOs leave the grouping to the players to figure out on their own. In a game that should be multiplayer by its definition, collections of players form that seem bent on preventing such grouping. Guilds serve to prevent the grouping of players for quests by virtue of being exclusive. Players learn to group with others in their guild, but also learn that grouping with those outside of the guild might prove dangerous (as it might be a collossal waste of time when the unknown "other" player can't make it through the boss fight at the end). Thus, you form typically two types of players - one who only groups within his/her guild, and another who doesn't bother grouping at all. For the former, there can be an extinsic social motivation to participate in raids and thus gather gear, but for the latter, the game is pretty much done.
Part of the problem leading to this point for players is that guilds recruit throughout the leveling process and players are often worried about alienating their guilds by leaving them at higher level. Part of the problem is also the necessity within these games of joining a guild to participate in so much of the game's content or more easily gather certain rare resources. Another part is the feeling so many of us have that if a guild invites us, at least we haven't been picked last for the baseball team - we are wanted for some unknown quantity we bring to the table, and that makes us feel appreciated.
So far, I've listed a lot of issues leading to the negative things leading up to a bad end-game experience, but now I'd like to offer some suggestions based on these things that MMOs can do to overcome them and make for a longer-lasting and more enjoyable end-game experience.
Firstly, we know that guilds not only cause problems but also solve them. Thus, we still need some kind of guilding system that makes it easy to organize people and get them into instances. This will improve the liklihood that players will experience extrinsic social motivation and also the chances of players enjoying the larger group content of "end-game." But we need to undo the damages caused by current guilds based around the "clique" mentality. I propose a social-networking style of system. For such a system, I would suggest doing away with "guilds" as we know them now. Instead, make the social interface of a game work more like a social networking site. Give players the possibility of joining multiple guilds, each of which can be managed by one or more persons, and to each of which those players can chat in a chat room. Make this interface both inside the game and outside the game so that players who are not in the game can still keep apprised of events in-game and even be invited to groups (keeping in mind that they will have to log in to the game to actually join, but if they're in chat for a guild and see that there's a group forming, they can say "I'll join for that" or whatever and then simply log in to the game). Events can be organized among multiple groupings of people this way, and thus everyone can keep track of them easily in- or out-of-game. In such a system, events could also be made public, so that if a person needs a quest done and none of the people in his or her guilds can do it, maybe other people can and maybe the friendship circles will grow as a result.
Secondly, we know that part of the problem of end-game is the sudden conversion people undertake going from a game that was entirely soloable up to this point and into a game that is completely unsoloable. For this reason, I would consider a system that ramps up difficulty considerably and makes it so that at around the halfway mark, players absolutely have to be grouped with one other player to get through quests, and at about 3/4ths of the way up, they really need a full group for all questing. Yes, there is a problem with this in that some people might get left behind through their inability to log in at a given time, but this is the reason for making quest events publicly available in the social networking platform of the game. Finally, make it so that XP actually increases slightly with multiple people in the group, up to a certain point, so that people don't feel like they're losing out by asking one more person to join.
Finally - make all quests progress for all people within a certain radius of the event that causes them to progress, regardless of group affiliation. The idea here is one to combat the "I gotta get this kill first so I get credit" mentality that comes from having too many PCs and not enough MOBs in a given area. The worst in those cases is quest bosses who are killed by solo players and then everyone else has to wait for a respawn. Instead, make it so that everyone who is standing around nearby and who is waiting to kill that same mob gets credit for that mob if they're on that quest. Similarly, make resource nodes individual - the node is always present, but you as a player can only use it every so often (it vanishes from your tracker when you use it until it repopulates for you). This way, another person could come and mine the same node and you're not having to fight for it.
This would be a new kind of training - a training away from the "me first" mentality and away from the solo-questing-not-really-an-MMO mentality and construct a group-based mentality that sees progress of the group as progress for the individual.