In any game, but especially in a large MMO, the numbers can get crazy if you don't keep them on a tight leash. It's a delicate balancing act between making character abilities exciting and cool (which to most players means they look great, feel unique and do lots of damage or other useful effects) and keeping everything scaled appropriately in relation to everything else.
To accomplish this, everything must be built around a framework of numbers. We establish baselines for damage per second, time to kill (both monsters and other players), time to level, total hours of play (on average of course), value per point of each statistic, damage absorption, damage deflection, xp per level, xp per kill, xp per quest, attack speed, attack range, health for players and mobs, and so on for a dizzying array of interconnected numbers and formulae. Each of these must scale from the lowest available power levels to the highest (and of course beyond, both for high-level challenges and for eventual expansion). This of course doesn't even begin to touch on all of the baselines required for the economy, for quest progression and distribution, etc.
Numbers alone, however, aren't exciting or cool. If your game feels like battling spreadsheets your customer base is going to be very, very small. Players want to call down the wrath of the heavens, turn into wondrous beasts (or turn their opponents into not-so-wondrous beasts), and generally perform fifteen impossible actions before breakfast. How do you turn something expressed in purely creative terms into a set of numbers that can be balanced against all of the other appropriate numbers?
As an example, we'll take something esoteric. Paul Barnett, our resident crazy British guy, mentioned in a few interviews a fun idea where our Squig-herder (a pet class that maintains a stable of nasty hopping toothy critters) could acquire a “Squig Power Armor” ability. This would enable him literally to climb inside a live squig, hide in the throat of the beast, and by yanking on the critter's tonsils and poking nerve clusters, control the squig from within. Bizarre!
Bizarre, but actually pretty easy. All we have to do is use the stats from the squig in question, add a nice bonus for having a smart little goblin “running it from the inside”, and voila! The rest is just art and interface, and the goblin fights like a squig for a while. A very creative and fun ability, but simple in the numbers.
How about a tougher example? Say we want a character to have a buff that scales both with the size of the player's group and the number of enemies faced. Every time an enemy hits anyone in the group, the character gets a small incremental advantage against that enemy.
In game terms, we create a simple buff that the player can place on groupmates. When an enemy hits a character with this buff, that enemy receives a counter. As counters build up on the enemy, we buff the character slightly. All is well and good so far. The trick comes in balancing out the variability in the results; the number of characters in a group can increase the creation of counters, the number of attackers can vary widely, etc. How powerful is this ability anyway? How do we give it a value we can compare against other abilities of a drastically different nature? How does this ability compare to the Squig Armor ability?
Fortunately, as you break down such an ability into its fundamental components, this task becomes easier. Every counter acts as a very small buff to some set of abilities on the character. We know how much, say, a 1% increase to the character's current damage absorption will decrease the effective DPS of the enemy he faces. We can set limits on how many counters can accumulate, and how fast, so in turn we know how big that buff can get, and how fast. So we break it out further: how effective is this ability solo? How effective is it in an average group against a single mob? Against an equal sized group of mobs? Against more? With a full player group? We know the average hit rates and DPS of both players and mobs, so we can get some working numbers on the most effective uses of the ability. From that, you can derive the boundary cases, which just means “the extremes”. What is the maximum possible effectiveness of this ability? What is the minimum? Average? Some complicated math and sample data can derive the “overall” effectiveness of the ability, and we can say with reasonable certainty how powerful the ability is compared to a normal buff (such as a straight damage or stat buff) or an enemy attack (incoming DPS), or another more complicated ability.
Of course, I just hand-waved a LOT of work, but in the end it does come back to numbers. Next, however, those assumptions now need to be tested. So we implement the ability in the game and begin to play with it. Do the numbers match our expectations? Are there any unintended consequences? Mobs may behave in a reliable manner, but can enemy players find a clever way to sidestep the effects? Testing may find, for instance, that players quickly realize that the best way to avoid the ill effects is to take down the character with this ability first. If we're smart, we've made that ability for a tank class, and so the player would naturally want to draw the attention of enemies from his group to himself.
Additionally, we may determine that some factor is making the ability more powerful than we anticipated. An ability of this sort can create drastically different behaviors in enemies. For instance, while a mob will fight as they always do, players will realize quickly that heavy use of area-of-effect attacks near this player's friends may be a bad idea! This results in a further unexpected overall drop in enemy DPS as they become reluctant to use their AoE attacks. This may necessitate a slight revision upwards of the effective combat value of the ability on our scale. Such a result can also alert us to the need to expand our model, to accommodate potential influences on the strategies of enemies.
In the meantime, an army of people are making sure that the art looks great, that the sounds indicate what is going on, that the interface is clean and clear, and that a myriad of other efforts come together. But in the end, we have a cool and interesting dynamic3 that can be balanced in its overall effectiveness against other equally esoteric abilities.
As a library of these efforts is built up, the task gets progressively easier. Rules of thumb can be derived that provide shortcuts in balancing. Counter-based behaviors become a category of effects, and can be given their own baseline. Any ability that influences any pertinent combat value or that provides a combat result becomes understood in relation to the others. At this point, we have a near-universal scale by which we can measure nearly any ability we can conceive. This has become our Battle Point system. Every ability, item, monster, character, trap, and widget can be given a Battle Point value that pinpoints its relative combat value in the game. From there, we can set up ways to balance RvR (e.g., you can only bring a certain number of Battle Points into a Scenario), balance mob encounters against expected player power levels, determine item progressions, and a thousand other issues.
In the end, however, it has to be fun. Everything written above makes it sound like a very dry and mechanical process, and in some ways it must be. But creativity is never “pure” in the sense that it exists wholly without underlying structure. It begins with a good idea, but to build it, you must have a system within which it can take shape. Context is everything. A beautiful home is built according to fixed and reliable principles of construction, to make sure not only that it looks good, but that it is sound and comfortable to live in. A beautiful painting uses perspective, contrast, color and definition in ways that can be described mechanically, but that come together into something that is a delight to look upon. An ability in the game, regardless of the numbers behind it, must be fun, and so requires a good idea to begin. But the system used to build it makes certain that one fun ability doesn't imbalance the game, ruining everyone else's fun.
