I do a certain amount of critiquing here of the game designs that others have worked on. What about the plus side? What's on my short list of "things that would keep me playing a MMORPG for years?"
I think such a list might look a little something like this:
I'd like to see a game developed in which all the systems were explicitly designed to fit together smoothly. More specifically, I'd like to see a game in which there is a conscious, deliberate determination on the part of the developers to create and maintain balances all through the game.
1a. Playstyle balance.
Combat should not be the only focus of development -- catering to today's hardcore gamers will not bring in the many, many more potential gamers out there. Combat is an important feature to provide, because today's gamers do matter; it's just not a good long-term strategy to focus on current gamers exclusively.
So I'd like to see a game in which combat gameplay is effectively balanced with the other major gameplay styles: commerce/crafting, exploration, and social. Treat these other in-game abilities as equally worthy of respect when providing new game content. If combat gets some major new content, so should the other playstyles -- maybe not in the same release, but overall.
1b. Game vs. world balance.
MMORPGs are games, so gameplay matters. But MMORPGs are also worlds, so immersiveness matters, too. Both of these aspects of MMORPGs ought to be balanced with respect to each other -- the gameplay should deepen the lore and appearance and emotional depth of the game world, and most world enhancements should try to sharpen and make more exciting the gameplay.
In particular, this means not imposing out-of-context gameplay. If it's felt that some common gameplay feature "has" to be provided, find a way to provide it that fits within and enhances all the other parts of the game and world.
And don't be afraid to add some world features just for fun. Some of the most memorable parts of games are the offhanded bits that didn't have to exist, that had no gameplay functionality or value, but that fleshed out an NPC or a location in a believable or fun way.
1c. Skill balance.
Skills -- especially at higher power levels -- need to come with a balance of strengths and weaknesses. For every benefit, there should be some appropriate cost.
That cost could be a cost to use the ability, or a weakening of some other ability, or a susceptibility to negative effects, or some similar consequence of choosing to use the high-end ability, but it needs to exist. A powerful and useful ability with only positive consequences winds up being something every player wants, and the result is templating and cookie-cutter characters. Balancing ability strengths with weaknesses helps insure that players will choose different skills, helping each character define his or her unique identity.
2. Crafting as a process that's fun in itself, not just as a means to a predetermined result.
For various reasons, too many designers have treated crafting as a mere support function for supplying combatants with gear. The focus is placed on the outputs because those outputs are what non-crafters care about.
But if crafting were primarily about being fun for crafters, and secondarily about supplying gear to other players, the focus would shift from the results of crafting to the process of crafting itself.
To make the process of crafting fun for crafters means allowing for surprise. If I follow the same recipe and get the same output every time, there's no surprise -- in fact, there's no crafting; what I'm doing is mere production, a purely result-oriented system. A fun crafting system would focus on making the process of crafting fun.
The process of crafting should be consistent enough to be experimentable, but deep and nonlinear enough to allow the results of experimentation to be surprising. By "experimentable" I mean that if the inputs could be held exactly the same between two crafting sessions, the outputs would be the same. This allows for science -- the forming of a hypothesis about how a process works, testing to evaluate the hypothesis, and reevaluation of the hypothesis. And by "deep and nonlinear" I mean that there are generally a lot of inputs to producing any item, and it's really, really hard to perfectly control all the inputs; there's some variation in inputs between runs. This helps lead to results that aren't totally surprising (you'll always get a sword if you use sword-making inputs), but that aren't totally predictable, either (each sword will have slightly different qualities).
To accomplish this, a crafting system needs two specific features. One is that the attributes of the inputs should be reflected in the qualities of the outputs. If I have a recipe for making candy that calls for some liquid as an input, using pure water should generate a different kind of candy than using grain alcohol. When I make a sword blade, that blade should have different qualities depending on whether I used bronze or steel or meteoric iron, on the amount and type of carbon I added from my fire, on the temperature of my fire, and on the liquid (water? olive oil? blood?) in which I quenched the blade to temper it.
This should also mean that whatever strengths and weaknesses an input has, both are transmitted to the final product in proportion to the amount of those inputs. A particular type of iron could be both very strong but very brittle, leading to a sword made with that iron having those properties as well. The corollary to this is that every input should have both strengths and weaknesses. This will insure that it's really, really hard to come up with a "perfect" object... or, worse, a perfect formula for making perfect objects.
The other desirable feature is a "construction kit" approach to crafting. Making a thing -- especially a new kind of thing -- should be a product of combining numerous components in many possible ways, and of altering components and groups of components through many different possible processes. The fun of crafting comes from playing (in the best sense of that word) with different inputs and processes to see what comes out at the end. Ultimately it should be possible to build large-scale and highly complex objects with reasonable but somewhat surprising behaviors, as long as the crafter is willing to take the pains necessary to make each individual piece work and put all the pieces together properly.
Real crafters like trial and error! Other types of people (*cough*Achievers*cough* ;-) can't stand this approach because they're so strongly result-oriented that all they see is the error. A focus on minimizing costs in a production process leads to a style of play in which unpredictability isn't fun.
But Explorers love this kind of crafting because they perceive every "failure" as an addition to their sum of knowledge about materials and processes. A certain amount of surprise in a production process is enjoyable to those whose focus is on maximizing results (rather than minimizing costs) because it offers opportunities for creativity and improvement.
I'd happily play a game whose crafting system was designed by someone who actually understands and appreciates crafting as a playful and creative process, rather than the kind so common in today's MMORPG, most of which seem to be designed by a developer who grudgingly cranks out a crafting system as a game-necessary manufacturing function.
There's more, but to sum up: Give me crafting as a creative process, and take pains that it and exploring and socializing are always balanced with combat when it comes to active content, and I'll be playing that game (and talking it up to other gamers) until they pull the plug.