I'd say the 2-8 player thing. They all start neutral, and wander like they're playing a 1p version of the game. However, they can stick close to each other, and join each other's battles, each taking command of their troops in the fight. Players joining an existing fight could choose a side to ally with, rather than simply being placed on an arbitrary side, thus, players could help each other, or enter battles against each other. Out of the 8 or so players, say 4 join each faction, now you have warring armies that maneuver intelligently and support each other!
A good use example would be player A likes to be a "solo guy" so he never invests in much leadership, but has a small contingent of archers that follow him, he enters combat, finds a nice vantagepoint to order his troops to stand and shoot from, then rushes headlong into the fray. Player B loves cavalry and the thundering doom of hooves. So, he invests in leadership and riding skills, and builds a force of knights, leading them in rushes on the enemy in a battle and intercepting possible trouble headed for Player A's archers. Player C likes infantry, so builds a skirmisher unit that follows the cavalry into the initial fray, remaining behind to help keep the riders from becoming trapped, protect those who are unhorsed, and play the "center of the maelstrom", remaining in the heavy fighting straight through the entire fight.
This, of course, would take a few other changes to how combat worked, and some extra considerations:
- It would mean increasing the total number of units that could be on the field, but decreasing the number any one party could bring in at once to a percent of their side's total.
-In addition to the increased total unit size, one thing that should be considered is that more players mean that that side will have a more dynamic strategy. So, to help ballance this, imballance of players on one side will be offset by decreasing the maximum number of units on that side by 2/extra player, and increasing the maximum number of units on the opposing side by 2. This gives the AI a net bonus of 4 units on the field for every extra player they have to face, while players fighting on opposite sides would face equal odds.
-More players should also increase the size of parties roaming around, not as drastically as the combat ballance above, but it should be considered, since players are likely to ally against AI parties, meaning the AI party will need to be much larger to be able to pose a real threat (especially since cooperation can be devastating).
-A cycling que system would be needed for reinforcements. So, as casualties mount, the reinforcements come from rotating partys involved on a particular side. In the above example, say you have players A, B, and C on Side X, fighting a huge army of dark hunters/knights on side Y. Side X loses 5 units (an arbitrary threshold for the game to decide reinforcements are called for), so player A's unit is refferenced for reinforcements. He only has 5 archers, all have already been deployed, so the que looks at player B's forces, he has 10 Swadian Knights in reserve, so it adds five. The black knights prove a bit more than the players expected, and 5 more fall. The que system checks player C's party, he has 3 infantry he wasn't able to field at the beginning, so it adds those three, then moves to player A, no troops, moves to B, fields 2 more of his cavalry. 5 black knights die, there's only one party on side Y, so the que always just fields more of that party's army.
- Late-comers would have to wait until their side suffered some losses to be brought in, but would be placed before other reinforcements in the "que", with the player being the first to be added.
- Because of the issue with large-scale combat (noticed from the Mod thread on increasing battle size), you could have corpses dissappear in a multi-player game, unless the server was running as a stand-alone, in which case it could likely handle it better.
- It would make this system even cooler if the AI would join fights (perhaps only player-involved fights, conserving units for fighting players, and increasing the challenge, since enemies could get reinforced, losing a bit of realism if they didn't join AI only fights but, meh, can't be perfect). It could even be an invisible rule that AI will only join in on losing sides, or sides opposing more players. That would make the challenge cooler, but might, again, detract from realism. "Man, when we fought on Swadi land, they got like five armies of backup showing up on the field! Those damn Vaegirs just stood there and watched those Dark Hunters lay into us!"
Anyway, it'd be a big project, but I do like the concept of a 2-8 player full version to M&B as opposed to an arena system (I mean, there's a half dozen or more of any given incarnation of this play-style out there, and capture the flag and such, or even counterstrike. . .can get old doing it over and over again, there's other forms of multi-player) I think it would just be more fun to play with and against players in the actual game, where they could explore the full aspect of the game. I think the arena thing would be boring and repetitive, while riding around waiting for the main player to get in a fight so I can do something also doesn't sound like much fun to me.
As for MMO. . .

Honestly, MMO is a thing that has to be part of the core design, not really to be added later. Moreover, M&B is an awesome game to sit down with a handful of buddies I know and play until we decide to do something else. I'd hate to play in the world of M&B and have to put up with "Yu fuggin' hacked! I know it! I'm reporting you!" "Ha! I so ninja'd that fight from U!" "OH! Totally pwnd ju!". Don't get me wrong, I play, and love playing MMOs, but I think it would destroy the M&B experience, even if it were designed in to be Massively Multiplayer. :shrug: