Players generally like to win more than they like to lose. That makes it so that one of the main attractions of pve is that you can usually win. In many games, a player can win over 98% of his battles if not inclined to try something unduly hard.
The problem with this is that the reason why the player usually wins is that it is easy, often to the degree of being boring. Start a battle, leave the room to go get a drink, and come back to find that your deft use of auto-attack has obliterated the enemy mob.
The antidote for this sort of predictability is pvp. You fight someone who really is trying to beat you, and if you’re too predictable or not paying attention, he can adapt to what you’re doing and then you die.
The problem with this is that if one side wins, the other side loses. Players still like to win, and winning half of the time may not be as fun as winning 98% of the time.
Some games will let players get around this by having uneven numbers. If a battle is 4 on 1 and the 4 win, then 80% of the players just won. Getting ganked like that isn’t fun for the one, however--even compared to losing a fairer fight. Even for the four, if you want a free kill with a preordained result, why not just go pve? Mobs won’t even insult you for ganking them.
So in a pvp battle with even numbers on both sides, 50% of the players lose. If there are more than two sides in the battle, then perhaps most of the players lose.
It’s actually worse than that: some players are better at the game than others, and will thus win more often. The common quote is that 20% of the players win 80% of the time and vice versa. That’s usually cited in contexts far enough removed from the original one to be part of the 73% of statistics that are made up on the spot.
Still, the general point is a tautology: players who win a lot tend to win more than players who don’t win very much. That works out fine for the relative handful of players who win a lot. That doesn’t work so well for the other players who mostly lose.
Some games try to get around this by making it into a contest of leveling. Whoever is higher level usually wins, just because he’s higher level. The implicit promise is that, while you’re going to lose a lot at first because you’re low level, someday you’ll be the high level dominating pvp.
For most players, it never works out that way. When you get high level, other players have had more time and are higher level still (or have better gear or whatever). Companies add ever more grinding to let the top players become ever stronger, and most of the players are always behind.
The trick is to make it so that most of the players can win what seem a priori to be fair fights most of the time. What I would propose is to combine pvp and pve. Let mobs take a large fraction of the deaths while players fight both mobs and other players simultaneously.
Start by dividing players into separate areas by level, as WAR does. Next, add a bunch of NPCs allied with one player faction or the other. Give them a lot of different AI routines, so as to be unpredictable. Make their strength mostly toward the low end of the level range, so that they tend to die to players, but with some exceptions for the sake of unpredictability.
The key here is to prevent players from knowing whether they’re fighting mobs or other players. Prevent cross-faction communication. Make the AI allies look indistinguishable from players. It might be enough to simply pick random characters from the character generator, along with random armor skins (but not necessarily stats) from the appropriate level range. If this tends to pick clashing choices that real players would generally avoid, then one could give mobs the appearance of randomly chosen players who happen to be offline at the time.
If an average player wins half the time when fighting other players, but 80% of the time when fighting AI characters, then perhaps he can win 2/3 of the time or so. Thus, most of the players can win more often than not.
Better yet, some other benefits important to RvR combat drop out of this system for free. One is that you can balance numbers, so that both sides have a roughly equal chance of winning. If one side has 50 players in a zone and the other side has 20, it’s not hard to predict that the 20 will mostly get ganked by the 50. Add 20 AI characters to the first side and 60 to the latter, so that the second side outnumbers the former a bit, but with weaker characters on average (because the AI characters tend to be weaker than players) and it’s now a fair fight.
The number and strength of AI characters that spawn could depend both on the number and strength (level and gear) of the players on each side, with more and slightly stronger AI characters spawning when a side has fewer players.
This also fixes time of day dependence. If a zone is designed to hold 50 players on each side, then during peak times, most of those could be real players. During the middle of the night, when only 10 players on each side are there, AI characters could fill in the slack, to let those players have some good battles, too. That might not be as good as fighting actual players, but it sure beats wandering around in a dead game.