Thursday, June 21, 2007
Free epics! - An essay on MMORPG difficulty
Have you ever played a platformer, a jump-and-run game? As the name suggests a major part of the gameplay consists of running towards a ledge, and then jumping at the right moment to reach the next ledge. Jump too early or too late, and you fall to your death, or to a point where you have to run back and try again. For the players that appears to be pretty much binary: Either you make the jump or you don't. For the game developers the matter is much more complicated. Because the run-speed is fixed, the game developer has to define the earliest point at which the jump doesn't fall short, and the latest point where you can still jump. And the length of that window of opportunity for making a good jump determines the difficulty of the jump. Everybody can make such a jump if given over a second of time. But if the window is only a couple of milliseconds long, most people will miss the jump and will need many tries before finally succeeding. This isn't binary at all, the developer can basically choose any value from 1 to 2000 milliseconds. But if he chooses a value that is far too short, the game will be too hard, and people will give up in frustration and spread negative word of mouth on it, reducing sales. If he chooses a value that is too long, the game will be too easy, and that isn't fun either. He could program a bridge between the ledges, so that you don't have to jump at all, but then there wouldn't be a game and there would be no interest.
Difficulty of MMORPGs is pretty much the same, only that the parameters you can tune are different. And again some players fail to understand that, considering for example the difficulty of a raid as a given constant: You either succeed with the raid or you don't. These players are mostly the successful raiders themselves, and as they raid mainly for the epics, they torpedo any discussion about raid difficulty with the completely false "you just want free epics" argument. Even the most casual player can see that "free epics" wouldn't be a good solution, as there wouldn't be any raiding game left if there was absolutely no challenge. That doesn't mean that the current difficulty level is the only one possible, or the optimal one.
Another frequent false argument is comparing the raid game with a competitive sport, like football or baseball or golf. Of all the millions of people who play golf, only a very small elite selection is allowed to participate in the PGA Tour, only the very best. That makes sense, because this is a spectator sport, and the money comes from people paying to watch, or advertisers paying to display their ads to the people who watch. Obviously a golf tournament between the world's best players draws more spectators than a golf tournament between a bunch of average Joes. A MMORPG isn't a spectator sport at all, and it isn't the people watching who finance the whole thing, but the people playing. So if you only give access to a small elite, you only get a small number of paying customers.
The genius of World of Warcraft, and the root cause of its success, was that the developers understood the importance of accessibility for the early game. So they made World of Warcraft considerably easier than previous MMORPGs, in spite of all the criticism from hardcore Everquest players. It only takes a quarter of the time to reach the level cap in World of Warcraft than it took to reach the level cap in Everquest. Furthermore in World of Warcraft you can reach the level cap just by playing through solo-content, without ever joining a group, which was impossible for most classes in Everquest. At the same time the developers of World of Warcraft managed to tune the difficulty that it was still interesting and challenging enough for the average player. WoW is "much easier" than Everquest, but far from "trivial". World of Warcraft also introduced a revolutionary quest system, where you are basically doing quests all the time, and the sequence of quests is designed to guide you and point you in the right direction. Thus players who would otherwise have been a bit lost in the open-ended gameplay of a MMORPG now had a guiding thread to follow. By being so much more accessible, World of Warcraft attracted nearly 20 times as many subscribers as Everquest had in its prime.
The criticism that has been directed at World of Warcraft already since the first wave of average players hit the first level cap, and which intensified with the Burning Crusade expansion, is that this principle of moderately difficult content accessible for the average player has not been continued in the raid end-game of World of Warcraft. There is a noticeable and significant break in difficulty level between the last quest you do to reach the level cap, and the first raid dungeon. You can fill that gap partially with 5-man dungeons, but there is still a big remaining gap. How big the gap is exactly, and how many people exactly are able to raid is a matter of much dispute. But I think it is safe to say that more than half of the players of World of Warcraft never got to see Ragnaros, the final boss of the first level 60 raid dungeon Molten Core. And more than half of the players of World of Warcraft will never see Nightbane, the final boss of the first level 70 raid dungeon Karazhan.
A major part of the difficulty gap is that playing in a group is inherently more difficult than playing alone. Imagine that a particular action needed to beat an encounter has a 80% success chance, making the encounter pretty easy. But now you design a raid encounter for 10 persons, where every one of them has to succeed in this 80% success chance action, and the raid wipes if only one of them fails. Then the total success chance drops to (0.8^10) or about 10%, meaning you need on average 10 attempts before getting it right. If 25 players needed all to succeed a 80% success chance action, the total success chance would be (0.8^25) or about 0.4%.
This is a design challenge. If you wanted 25 players all to have to participate in a raid group action, each one being equally important, and you wanted the whole raid to have a 80% success chance, you would need to make each person's action to have a success chance of over 99%, thus being pretty much trivial and uninteresting. So what developers of raid content do is design it in a way that not everybody needs to succeed for the whole raid to succeed. There is some slack in large raid encounters, especially those with 40 players. A sub-group of key people needs to do their piece exactly right, and the rest just adds to the damage done or to the points healed. It is a lot easier if 10 people together need to do 100,000 points of damage in 1 minute to succeed, than if each of them has to do 10,000 points or fail. Maybe somebody can only manage to do 9,000 points of damage, but if you allow that to be balanced by somebody else doing 11,000 points, the raid encounter still succeeds. The disadvantage of that solution is the possibility of resentment from those players that contribute more to the raid success towards the "slackers" that contribute less.
By making raids smaller, The Burning Crusade eliminated much of the slack. Especially Karazhan, the first raid dungeon of the TBC raid circuit, hasn't go much room for error. If one of the 10 players either makes a stupid mistake, or just gets unluckily disconnected, the raid usually wipes. And with fast respawn rates, recovery from wipes isn't that easy any more, often you need to clear a part of the dungeon again that you already did. This has the advantage of making every individual player in the raid more important. But the disadvantage is a large degree of exclusion, because if every fault can wipe your whole raid group, you can't afford to take anything but the very best players available.
Reasons for not taking somebody on your Karazhan raid are many, and most aren't even based on the real skill of that player in playing his class. One major problem is class balance, and the difference in popularity of various classes. If a raid encounter is designed to be only moderately difficult, and has some slack, the raid group gains a lot of freedom on what classes to take, and what talent specs the raid group members can have. The harder the difficulty, the less freedom. A friend reported from his guild that for Karazhan they even have different raid compositions for different boss encounters, taking more of one class for this encounter, then kicking them out after killing that boss, and taking more of another class for the next encounter. The distribution of classes in the general population isn't equal to the optimal distribution of classes in a raid, which already automatically gives some players a better chance on a raid slot than others, just based on their class. Another important reason to exclude somebody is the way that raidIDs work. If your guild organizes two Karazhan raids on Monday, and on Tuesday only half of each raid group turns up for the second part, you can't just throw the two halves together and continue. People able to attend raids every night won't have that problem, and of course they will learn the specific raid encounters faster through more playing, and thus be preferred when selecting players into a raid group. The "you're my best mate, but I won't take you on a raid with me because (you don't play well | you have the wrong class | you have the wrong spec | you play soccer on Tuesdays)" dilemma is the source of much guild drama. And the root source of that problem is raid difficulty.
Just like the jump-and-run game, the design difficulty of a raid encounter or the whole raid dungeon is completely arbitrary, and can be choosen by the developers in a wide range from trivial to impossible. And it is perfectly possible to make a series of raid dungeons where the early ones are much easier than the current World of Warcraft raid dungeons, and where the difficulty then goes up with every encounter and dungeon. There is nothing wrong with having a series of 10 raid dungeons where only the best 10% of players can reach the final boss of the final dungeon, as long as more than 50% of the players are able to kill the final boss of the first dungeon. If you want raiding to be the "end-game" of your MMORPG, you better design it in a way that most of your paying customers can at participate in the early stages. You don't want to give them "free epics", but you do want them to be able to experience the content and special play style of raid groups, without them having to neglect their family or ditch their mates. The proposal is not to make raiding trivial, but to dial down the difficulty to a level where the average player can have some success, even if his raid group hasn't got the optimum class mix or the very best players in it.
Note that these considerations aren't about abstract concepts of "good" game design. The game design principle of accessibility wasn't invented as a counter-theory to "the Vision". Accessibility by clever choice of difficulty level is simply good for business, the longevity of the MMORPG depends on it. If players hit a wall in the game, where they don't see how they could possibly advance any further without making major changes to their lifestyle, they tend to cancel their accounts. You can't keep players in your game by adding another high-end raid dungeon that they will never see, the pulling power of that is just too small.
Curiously it is World of Warcraft itself that still has the best chance of providing this accessible raiding game. There is no game existing or announced that plans to do anything like that. The end-game of Warhammer Online, for example, will be dominated by PvP, not PvE raiding. Lord of the Rings Online's plans for any sort of end-game are still nebulous. And many newer games are going for a more action-oriented gameplay, which isn't really strategic enough for large-scale raids beyond simple zerg rushes. While WoW's The Burning Crusade expansion raiding end-game turned out to be less accessible than classic WoW one's, at least the intention for more accessible raiding was there and was announced. And recent steps that change the current end-game go in the right direction. So there is hope that in a future expansion Blizzard will get it right, and make the raiding end-game accessible for the majority of players. Wouldn't that be great?