I've been thinking a bit lately about some of the high-level principles of PvP design. These principles readily apply to video games and LARPs. I want to be clear from the start: this is not master-class level stuff - for that, you want to talk to Stands-In-Fire. Quite seriously, the majority of what I know about PvP design I either learned from him, or from the hurdles the Fallen Earth Systems team faced in my time there. Also, I am not an expert on PvP-heavy LARPs, but I've had some fascinating conversations with friends who are experts on them.
One of the first decisions that goes into PvP design is the question of how many factions the conflict will have. While I've often suggested that three sides is the Right Number, that really just means that such a system will face the fewest balancing hurdles, because player population issues will be mostly self-correcting. Let's not pretend, though, that such an answer works for all games; you can thank the Zoroastrians for giving us dualism. To that end, I'll go through a variety of models and discuss problems and available solutions.
Zero Factions, or One Faction
Obviously, these come down to the same thing. This is a model in which PvP is not driven by faction membership. It might be a free-for-all, a tournament ladder, or a Highlander-style winner-take-all; this model is functionally identical to "every player is a one-man faction." This isn't common in MMOs, because MMOs gain huge amounts of player buy-in by encouraging players to align with a faction and invest in its story and struggle. (Remember: Anything that is socially destructive in real life is almost always ideal grist for games and stories. Tribalism is just one example.) The good side of this model is that there is always someone to fight! The bad side is that if you can't defeat an enemy on your own (for reasons of skill, level, or whatever), you have no natural allies to call upon.
A faction loyalty, then, is a channel for cooperation and identification with other players. Games without factions, or games where all players belong to the same faction but still kill each other (such as an extremely zoomed-out view of the Camarilla, in campaigns without the Sabbat), have PvP fights that are more like homicide than warfare. Without a defined channel for cooperation, it can be difficult for the players to build anything in the world that they can't accomplish on their own - they don't have a solid, mechanical basis for trust. (This is like real life, in a time of anarchy.) This is fine if your game isn't about the players building anything or changing the world in any way, but if that's true, you might not be playing an actual RPG. For example, you might be playing Banner Saga: Factions, a game I have dabbled in recently. There is no persistent world, just a matchmaking queue, so they made their matchmaking queue absolutely as snappy as possible by using the word "Factions" in their title ironically. (I kid; I'm sure they're talking about each player's team as an independent faction. See my above comment on "every player is a one-man faction.")
Some of the problems of cooperation can be addressed by giving players a way to opt out of PvP, temporarily or permanently; if it's publicly visible that you have opted out of PvP, or you're standing in a safe zone, players have a clear way to trust you. This is a kind of brute-force solution, since you're solving a PvP design problem by toggling off the PvP, but it can work. There might be a more nuanced solution, but in most games it won't be necessary. The upcoming Wizardry Online, which has touted itself as the most hardcore PvP experience possible (permadeath? hokay, dude), might be an exception to this statement; I get the impression that their plan is to let players kill each other at any time, so you only expose yourself to risk with people you have social (not mechanical) reasons to trust. This will make the player community the most important thing in the game, but it will also be a griefers' paradise. Let's call it a really interesting social experiment, but an insanely risky game design decision.
Lord of the Rings Online is a special case of one-faction PvP: your main character always belongs to the "good" faction, but during PvP encounters you can temporarily change over to an evil character (orc, goblin, whatever). I guess you'd call this a one-faction PvE system with a two-faction PvP system. I don't know enough about how the players have used this system to judge its effectiveness. Commentary welcome! As a general case of this concept, you could have a non-factional PvP system that confined its PvP to instances, and automatically assigned players to teams for the duration of the instance.
A G+ post by +Mike Boaz, in which he discussed some of the apparent design problems in Ingress, was the seed for this whole post. He pointed out a number of problems, but the one that interests me here is the imbalance in player population. In that game, as in every two-faction game I've heard about, players do not organically align themselves into two equally-matched groups. On any given server, one side emerges with a distinct numerical advantage. That gives them every other kind of advantage in the game as well, unless there are programmatic solutions in place.
In the particular case of Ingress, the Resistance and the Enlightenment are duking it out over portals all across the world, and the Resistance has a 25-percentage-point advantage worldwide. Now, the designers naturally needed for success to carry perquisites; once you gain territory, there are things you can do to make it more difficult to attack. That's good! Players can point to ways in which their actions have changed the setting. On the other hand, victory becomes an entrenched state. Fewer numbers and a higher bar to success? The real world works like that... and now you know why social justice is a big deal. (This is a case of something that sucks in real life and sucks in gaming.) At least in the game, we're talking about more people opting out of the uphill struggle, rather than just being trapped. This is not much consolation to the designer or the game's investors, though.
Star Wars: The Old Republic, World of Warcraft, Warhammer: Age of Reckoning... all of these games have chosen two-faction PvP over other models. All of these games have been successful to one degree or another, and all of them have had to take steps to address the population imbalance issues. SW:TOR and World of Warcraft handled the problem by shifting focus away from open-world PvP and toward instanced PvP Battlegrounds. In an instance, the game controls for population differences by pitting two teams of equal size against one another. This is great, except that instanced content means there is no real room for victory and defeat to have lasting consequences. It's a sport, not a war. This is good for keeping the game fun in the long-term, but bad for most concepts of roleplaying and narrative, as each individual instance completion has zero significance in the world.
At the same time, World of Warcraft has always tried to retain some elements of open-world PvP even on Normal and RP servers. From classic WoW through at least the last expansion I played (Lich King), there have been zones in the game where players can complete PvP objectives as a mostly-seamless part of the things they're otherwise accomplishing there. This softens the population advantage through self-selection - typically there aren't more than 10-15 people in a given world-PvP zone anyway. The benefit of victory is typically a minor buff to all members of that faction within the zone. Only in Lich King did we see a zone in which victory was important enough that Blizzard worried about leveling out the population imbalance, handled with a scaling zone-wide buff to the disadvantaged side. Warhammer: Age of Reckoning does something basically similar, since de-emphasizing world PvP was never an option for them.
Mike suggests a dynamic handicapping system for Ingress, which is another name for a zone-wide buff. His alternate solution is a penalty to the effectiveness of the winning side; this is roughly equivalent to victory carrying an increase to maintenance cost. I don't yet know the ins and outs of Ingress, but I will say that any small adjustment to the damage done by low-end XM Bursters won't be sufficient to make the low-end players feel useful. There's an additional PvP design issue at stake here, beyond population balance: level scaling. Ingress is a glorified version of tug-of-war, in that characters don't really attack each other directly - they instead attack the progress that other players have made against elements in the world. Level advancement grants the right to use more powerful (but rare) consumable items, and that's about it - but those more powerful items are a lot more powerful, and they apply to both attack and defense. A useful discussion of level-scaling design issues in PvP will have to wait for another day, as this post is long enough already. The brief version is that most games handle it by creating level brackets to match players up against vaguely appropriate opponents, and Ingress might benefit from a light-touch move in this direction.
SW:TOR has handled this problem by allowing some or all of their PvP instances to feature two different teams from the same faction. This allows their instance queues to keep flowing even if 100% of a server's population belongs to the same faction.
The one other thing I want to say about two-faction PvP is a lesson from Stands-in-Fire. If PvP is your thing, figure out which faction is dominant on your server of choice, and if at all possible, play the less populous faction. This does three things for you.
It gives you a target-rich environment, which often takes the form of much shorter queues for instanced PvP. This means you will advance faster.
It teaches you to be a better player by presenting you with actual challenges more often. The smaller faction also tends to be more tightly knit, creating a more enjoyable gaming community with more experience (through necessity) at working together.
It may be a drop in the bucket, but it's what you can do to level out the population difference and improve the game for everyone.
As I've said, this is the lazy designer's ideal case. As long as there is no one faction that is substantially more powerful than the combined strength of the other two, there's a lot of room for variation in populations that can still provide an exciting challenge, and paradoxically the smallest faction often has the most power, because they ally with one of their larger neighbors and (all other things being equal) decide who will win. To draw a real-world political comparison, it's the undecided moderates in the center who decide presidential elections in the American two-party system, and it's the smaller groups that get pulled into a coalition that allow multi-party systems to form governments.
Dark Age of Camelot is justly famous for implementing this structure, known there as Realm versus Realm. It is ideal for world PvP, but it presents serious logical problems for instanced PvP (a thing that became popular long after DAoC's heyday). We had planned on a similar system in PROJECT NAME REDACTED, but that project was canceled long before even the end of pre-pro. I'm sure it's been done in other games as well, but for some reason I can't think of any of them right off; I suspect that instanced PvP becoming more popular than world PvP has a lot to do with that.
The instanced PvP issue can be addressed, at least in part, by having players queue for PvP in general rather than one specific Battleground. Though not widely understood, this is something games need to be doing anyway. It doesn't give the player what he wants in the short term, but it keeps content in use and it keeps the queue flowing, which is a higher good for all involved - including the guy who didn't get what he wanted. Back to my point - if there are three factions, let there be three Battlegrounds (or some multiple thereof) - one for A vs B, one for B vs C, and one for C vs A. When a player of faction A queues, he will get either A vs B or C vs A, depending on which roster the game's matchmaker can satisfy more quickly. This does mean that there is content that player will never see, but a design team scared off by that idea can't handle two-faction PvE, much less three. (I may be overlooking or oversimplifying issues, though.)
In the aforementioned G+ thread, I agree with Kainenchen and Mike in saying that adding a third faction is probably the best thing that Ingress could do to turn its game into a perpetual conflict machine, unless they plan to implement some wildly unexpected feature that parallels instanced PvP. I don't expect them to pursue this solution, because I think their metaplot is probably strongly attached to the conflict of their two factions.
There is another kind of faction system that I can imagine (because it's how American politics works) that might, hypothetically, be possible to implement. I can't think of when I've seen it implemented as such, though Fallen Earth's original version of six-faction PvP was intended to get at something near this idea. Let there be two factions; for ease of use, the Elephants and the Donkeys. A player can emotionally identify with one of these factions, and a player can mechanically identify with one of these factions, but at any time the player can leave a faction, to either join the opposition or remain independent. Why would a player leave the dominant side to join the Other Guys? Because this is violent game conflict and not real life, the player does it because the rewards for completing missions and winning battles are dynamically determined, and the richer rewards go to the underdog side (sort of like long odds in gambling).
Four or More Factions
As the number of factions increases past three, the game increasingly approaches being a Guild versus Guild game. There's nothing wrong with that, but it is a radical departure from the assumptions made in a game with fewer factions. This style is more appropriate to a sandbox game, such as EVE Online or the upcoming Pathfinder Online, where each faction holds territory in its own name and only concerns itself with fighting or allying with its nearest neighbors. We're talking about a total focus on world PvP and actual player politics; it would be exceptionally difficult and largely undesirable to stage these conflicts within an instance. As a result, the game shouldn't correct for population imbalance in these cases; that population imbalance comes from conscious choices on the part of the players, and the game has to let them stand.
The only one of these that I have personal experience with is Fallen Earth, and I decline to comment on its strengths and weaknesses both because it is too close to home (I work for a different company, but under the same roof), and because I have not kept up with the design team's innovations over the past almost-three years. I am under the general impression that it has changed greatly since my time there.
On the LARP side, the games I'm used to put a lot more emphasis on interpersonal politics (competition intermingled with cooperation) than on open violence. Since simulation is at a premium in LARPs, there's really no mechanical room to correct for population imbalances, but these problems mostly take care of themselves through social dynamics. Probably the ideal case, from the game-runner's perspective, is a constant simmering conflict that never quite erupts into a bloodbath - a lot like hour dramas on TV.