Following on from my earlier list of games that MMOs could learn and build on, here are a few more games that really could really help add to the MMO genre or serve as a warning to it:
Syndicate (PC / Amiga): Few games have built a world as thoroughly as Syndicate did. Every time I read that GTA was the first game to really capture that feeling of a big city, I'd think back to watching civillains walk around the maps in Syndicate, seeing cars wait for traffic lights to change and generally acting like it was a living city.
Then I'd have a mini-gun firefight against rival corporate troops, sending civilians screaming for cover before destroying the cop cars that were sent in response to the disturbance. Ahh, vulgar displays of power...
Syndicate (which could be an awesome MMO in and of itself given some tweaking) offers two things to MMOs: 1) having a good approximation of a 'living city' that reacts to your actions and 2) offering multiple solution paths to missions. Some MMOs have taken steps to dress a map up as though it is a living city - City of Heroes / Villains springs to mind - but to my knowledge those cities have been a bit poor in reacting to the player. You can't generally blow up a car or scare the NPC civilians (or alternatively repair the car and hop in it so you can drive around) - the world doesn't really react to what you do outside of pre-scripted results to missions.
Offering multiple solution paths is also rare in MMOs - typically its slash your way to the big bad guy before killing him too. In Syndicate (and I won't pretend every single map allowed for multiple pathways to achieve your goal, but a lot did) you had access to multiple tools that you could use to succeed - it didn't matter if you were subtle / stealthy or staged a full-on assault provided you achieved your goal. Most MMOs seem to be reliant on a player playing through a mission exactly as the developer intended rather than letting players pick their own path.
Vagrant Story (PlayStation): Leaving aside the excellent story and gameplay contained within Vagrant Story, it's the brilliant item and crafting system that could easily be lifted into a next-generation MMO.
In short, within Vagrant Story, it is the weapons and armour that level up rather than the player. If you spend all your time fighing zombies with a sword, that sword gets stronger versus zombies (or whichever of the different power attributes that zombies possess... iirc, it was Undead and Humanoid). That's probably pretty straight-forward. However, you could also combine weapons and armour at the Blacksmith that would then take on the attributes of the items you combined. Not everything could be combined to make what you wanted - two daggers combined became a better kind of dagger - but it was a complex system that theoretically made all types of weapons and armour useful because you could always combine it to make something (hopefully) better.
I'd like to see more MMOs allow customisation of weapons and armour, while also allowing players to convert their items back into base materials that can be combined in new ways to create new things. The metal in a breast plate is also metal that could be used in a sword after all (within a fantasy "we don't care about metallurgy" context, anyway). It would also be great if weapons could 'learn' from the opponents you fought and start to provide bonuses against those kind of enemies. Most MMOs have a pretty disposable item economy - get a sword, use it until a better sword drops, dump the old sword, repeat - so it would be a change to make it worthwhile to hold onto weapons for a longer period of time.
E.T. (Atari 2600): Considered by many the worst game of all time and responsible for Atari's first bankruptcy (which in turn fed into the video game crash of 1983), the lessons of E.T. should be tattooed on the inner eyelids of every game developer considering developing a MMO title (even if it doesn't have an established property involved) and every publisher rubbing their hands together at 'hot title' .
The critical factor behind E.T. failing as a game was time. Howard Scott Warshaw, the programmer, wrote the game design in two days and programmed it in five weeks in order to meet a shipping deadline for Christmas of 1982. No time was available for programming some of the more "sentimental" ideas that would have better tied the game to the film while player testing was skipped entirely. If Warshaw had been given a more reasonable timeframe - between five and seven months - he'd have created a very different game and one that might have met Atari's sales targets.
It should be noted that E.T. actually sold 1.5 million units - that'd still make it a very successful game in today's market on a units sold basis - but what crippled Atari was that they'd manufactured 4 million units for sale. So E.T. the game failed because it wasn't given enough time to provide a quality experience while at the same time Atari management saw it as such a sure thing they overestimated demand. History shows the result of such hubris.
Don't get me wrong - I think MMO development actually needs to be more stringent in meeting deadlines and MMOs shouldn't take years upon years upon years from the time they are first announced to the time they eventually ship. But fixing the release date and demanding that the game ship no matter what on that date is just a recipe for disaster.
You can read a fuller account of E.T. the video game's history on its wiki page.