The Monster Rancher games, like them or not, had a unique approach to having multiple characters (or in their case, pet monsters) that could dramatically increase the longevity and depth of MMORPGs if applied. This article will briefly go over how it worked and what potential applications could be used.

Note: I know that there was a Monster Rancher MMO that came up a few years back and died before it ever made it to the states, so I can't comment on how well it may or may not have applied these principles.

Key points of Monster Rancher:
In Monster Rancher, the top achievement of the game was to become a master breeder, which essentially meant that you would raise a monster and get him through a series of really tough arena fights. If it was designed with the current MMO mentality, the player would make a monster, play him up to this point, then if he desired, take other monsters up to this point as well (i.e. level up a few characters and take them all through the hardest content).

In Monster Rancher, however, the game isn't set up for each monster to go through the same process. Because of the limited lifespan of each one, they could only accomplish so many tasks. As a result, players trying to raise an "everything" monster would never be able to achieve their ultimate goal, no matter how many attempts they made.
Instead, players would succeed by raising monsters with different intents. One would be to get lots of money, another would be raised for breeding, a different monster would be sent on expeditions to get rare and useful items, etc. Ultimately, each of these monsters would be used to prep a highly powerful monster so that he could actually get through all the hardest content.

How would you implement this in MMOs?
While it would take a VERY different MMO for characters growing old and dying to be a good thing, there are aspects of this that could be very applicable, sometimes with only moderate change to existing games.

The first thing this would require would be to have a strong level of interactions between characters owned by the same player. Items that bind to an account would be the norm, rather than the exception, and an interface for easily interacting with each other would make the experience much smoother.

Also, this design would work best if raising some of the alternate characters didn't take as long as the leveling process in an MMO usually does.
Once these changes were implemented, the designers could be focused on content where characters support each other. For example:

1. The ability to build a character who sacrifices fighting power for increased ability in other areas such as crafting: The hardest part about this option is, of course, having a game with non-fighting progression that is sufficiently interesting. Your specialized blacksmith wouldn't be able to go through top level fighting content, but he could make amazing stuff you couldn't get any other way.

2. Level-capped content with wide-reaching reward: What if the only way to get a component for a powerful item was to do a level 20 instance with a group made of level 15's or less? While many players currently consider leveling to be the boring part of an MMO, adding specialized challenges like this would make this experience a lot more enjoyable.

3. Content designed with non-standard challenges: Picture a 3-man instance full of glass-cannon type enemies whose damage output was so high that it was nearly impossible to clear without a group focued on burst damage and crowd control. Not every character would even be capable of being in a group that would be able to make it through. As long as there was enough variety in this type of content that no character type felt underutilized, this would allow a greater degree of new experiences with a new character, while at the same time, acquiring rewards for your account that your main high level character is not capable of.

There is far more that could be done with this, but I wanted to keep my suggestions in line with what would fit into the standard MMO (i.e. WoW, Aion, Warhammer Online, Guild Wars, etc.). The most important thing to keep in mind would be to focus on ways characters in an account could support each other, rather than treating each character as an individual experience.

Don't overscript it:
The one danger here would be making the mistake of having all of this work in highly scripted ways. Examples:

Bad: You have to be a mage to enter this dungeon.
Good: This dungeon was designed to be most doable by mages, but we're letting players have the room to be creative and find alternate means.

Bad: Your character can either ally with Faction A or Faction B, each of which offers a unique set of tasks and rewards.
Good: Faction A and Faction B both offer a unique set of tasks and rewards but the style of challenges from each faction are vastly different, making it difficult for a character to progress far in both.

Besides the fact that the scripting would make gameplay feel confining, it would also encourage lazy design. In both of the above examples, it is far more difficult to design the good content than the bad, but it would be far more gratifying to players who would feel like they solved and conquered a challenge, rather than just being handed something because they made a particular choice.

Conclusion:
Though imperfect in its implementation, the design of Monster Rancher, encouraging monsters/characters who fill different roles that support each other, would go a long way toward taking MMOs to the next level.