Skip to content

the problem with scissors

April 20, 2009

“Ok, let’s say warlocks have problems against rogues. I don’t know why you’re complaining. Warlocks aren’t supposed to beat rogues! PVP is based on rock paper scissors.” – Idiot Rogue

If you’ve never seen the Roguecraft movies, you really need to go do that. They’re pretty funny and are a great example of the way some players want to view games. Now, I’m not sure where the rock paper scissors analogy started, but it fits many games very well. Many card games and real time strategy games use this mechanic. So do some class based shooters, such as Team Fortress and Battlefronts. Many fighting games also use this in the same manner: attacks beat throws, guards beat attacks and throws beat guards. When looking at games where this balance scheme works a pattern starts to develop and common themes can be seen in the games.

  • Team sizes and resources are kept even between players.
  • Players are given multiple choices to select at once.
  • The choices are active during gameplay.

First, each team, or each player is given an equal force to the other. This means that in the overall game this is not a ‘rock paper scissors’ game; the opponents are on equal footing. This means that the ‘rock paper scissors’ part of the game is actually a game within the game. This is reinforced by giving players multiple choices. This is particularly true of strategy games where the player, or team will be fielding many units all of which have different properties. Here ‘rock paper scissors’ is not used to determine the winner, but instead to prevent “zerg” strategies. Interestingly the origin of “zergging” is in a strategy game, but unlike its typical use, it actually refers to an early-game rush strategy. It does not refer merely to attacking with a larger force than the opponent. Finally, the choices in all of these games are active choices. None of these games involve, for instance, picking rock and forcing you to play rock permanently.

All of the games which successfully allow this design scheme allow you to fluidly change tactics, just as the real rock paper scissors game. The same cannot be said about MMOs. Here players choose their characters and then build them up over the course of the game by upgrading their equipment, skills and stats. In this case the rock paper scissors style of gameplay can only be applied properly if the game is a fighting game model. This means that while one may be locked into their character, that character has various options to counter other players characters. This causes dramatic problems when an MMO attempts to model itself around strategy, or shooter-style rock paper scissor models.

The largest problem applying this type of model to an MMO deals with character commitment. Since a player is only controlling their set character at any one point they begin to grow attached to the characters they play. Now, take for example three players, aptly named Rock, Paper and Scissors. Rock decides to go for a walk and runs into Scissors. Rock quickly defeats Scissors. However, Scissors happens to be friends with Paper and tells them where Rock was walking. Paper shows up and defeats Rock easily. How does Rock feel? Rock defeated Scissors, but can’t feel good about this because Rock is supposed to beat Scissors. Afterwards Paper defeats Rock, but Rock feels helpless because Rock is supposed to lose to Paper.

This doesn’t apply to a game that is balanced around the fighting game model. Here we have two players, this time named Ken and Chun-Li. Ken decides to go for a walk in the forest and runs into Chun-Li. They face off and Chun-Li decides to pull out a massive pair of scissors, while at the same moment Ken throws a rock at her. The rock breaks the scissors, but the fight continues. Ken decides to throw a second rock, but Chun-Li has caught on to his strategy and catches the rock with a large piece of paper. This goes on for some time and after a long and skillful battle Ken counters a couple dozen scissors in a row, then smacks Chun-Li in the face with a giant rock. Here the players of Ken and Chun-Li were presented with options of how to counter each other instead of being being forced into the decision upon character creation.

The difference in the models when applied to an MMO is that the strategy/shooter model forces decisions out of the players hands, while the fighter model allows players to develop strategies around their playstyle. However, this is not the only problem in basing MMO balance on strategy and shooting games. Another large problem is that most games do not have three options. In WAR’s case there are twenty-four options. This means characters must then be divided into subtypes and balanced accordingly, or characters must be balanced on a class-by-class basis. In the case of WAR this gives us twenty-four characters with twelve on each side that need to be balanced against the twelve opposing classes. This leads to a very complicated balance model, which looks more like a balance web where each character beats half of the opposing classes, but loses to the other half.

balance-webThe class-by-class method is very complicated. Instead, WAR has broken the classes down in several ways which we can use to subtype the classes. The most important is when looking at the characters roles. In WAR there are five functional roles and a sixth role which is not functional. Those roles are: tanking, melee DPS, ranged DPS, melee support and ranged healing. The sixth role is ranged support, which is incomplete and can’t be fully performed by any class. That leaves the game with five options which need to be balanced against each other. While this is more complex than “paper rock scissors” a model can be created to show balance in a system with five options. In this model each option beats two of the other options, but is beaten by two of the other options as well. There are various ways to balance this, such as in the example below.

5-way-balanceHere tanks beat DPS careers. Melee DPS defeat ranged careers. Melee healers beat tanks and melee healers. Ranged DPS wins over healers. Finally melee healers beat tanks and melee DPS. This may, or may not be optimal, but it illustrates how five-way balance works. Even if this is balanced properly it leaves players battle outcomes to be determined by their class role. Since this system is itself very complicated it also allows for the game itself to be thrown off balance by class buffs. Below you can see the actual current balance of WAR’s five class roles.

how-it-works-nowThis leaves most of the power within the game in the hands of two classes. It also places the ranged careers well beneath the melee careers. The objection to this thought process is generally that WAR is not meant to be based on one-on-one conflicts. However, this objection falls short when analyzed further. An easy example of this, using the balanced five-way model, would be to increase the group size. The first step is to assign how much of an advantage a class should have over another class. The lower this number is the closer it will get to being a one-on-one balanced model. The best way to do this would be to assign them ten duels and choose a number that the classes counter should beat it. For instance a tank should beat a melee career six out of ten matches. This is a good place to go with as it gives an easy to read advantage of six to four.

This allows us to analyze some two-versus-two groups. This is again a small scale, but is a large step above one-on-one and a good example is testing how this model works. The next step in this is to decide on the groups to test. An easy starting point is the standard group of a tank and a ranged healer, which is a mainstay of MMOs. The other team will be a simple construction of two melee DPS careers. The healer versus the melee is at a four to twelve disadvantage. Meanwhile the melee versus the tank are at an eight to six advantage. This means that under the five-way model that two melee can simply burn down a classically balanced group.

This is complicated further in very large group where balance changes further. The early diagram showing the state of balance in WAR currently was based on small groups. In large groups the balance in WAR is actually much different, which can be seen below.

large-scale-balanceThis balance shift is caused by the advantage of range and position. This gives a heavy advantage to any ranged class, while also giving a heavy disadvantage to non-tank melee careers. It also points to one of the largest errors in this form of balancing: Attempting to balance a game around a strategy-game model while not allowing players to actively change their group leads to poor balance. When characters are balanced around a fighting game model if a situation like the above presents itself, then the group is able to shift into an anti-ranged strategy to gain the upper hand. This is not possible in the strategy-game model unless a shift can be made in army composition.

The final argument for this type of design is that it creates diversity. This argument is very hard to support, often leaning on the idea that because classes are balanced one-on-one that they become homogenized. This is caused by the misunderstanding of balance and the idea that to be equal classes must be the same. The counter argument is that different class roles and even class mirrors as in WAR allow you to provide diversity; different characters can do things in different ways to produce an equivalent result.

An example of this would be a race between two vehicles. Both vehicles need to transport twenty people from point A to point B. The first vehicle can carry ten people, while the second vehicle can only carry five people, but travels almost twice as fast. It takes the first vehicle a minute to travel from point A, to point B, or back and another minute to load and unload passengers. It takes the second vehicle thirty-five seconds going either way and it only takes fifteen seconds to load or unload. The first vehicle will deliver the twenty passengers in six minutes and in two trips. Meanwhile the second vehicle is required to make four trips, but takes the same six minutes.

This may seem like a silly comparison, but its a good way to show how two very different things can be balanced against one another. It also shows how not only can things be balanced one-on-one, but also balanced in such a way that the skills involved are different. The first vehicles driver will have much different concerns than the second vehicle. The first vehicle needs to be quick and efficient with loading and unloading passengers, while the second vehicle needs to handle the driving aspects to a much greater degree. Some MMO comparisons are the differences between ranged and melee. Ranged classes need to be able to defend themselves from close ranged attackers by being elusive, or utilizing crowd control abilities. Meanwhile melee need to be able to defend themselves at range by either closing distance very quickly, or being able to reduce incoming damage and then once at range need to try to keep their target near them.

This type of balance leads to interesting outcomes in large scale battles. Because melee are now able to defend themselves from ranged attacks they are not in as dire a situation as they would be in a rock paper scissors scheme of balance. Meanwhile in open combat ranged are no longer at huge disadvantages against melee. This means that battle outcomes are no longer determined primarily by class choice and instead lean heavily on player skill level. Good ranged players can defeat mediocre melee players in single combat. At the same time good melee players will not be utterly destroyed in large conflicts. In this way small scale balance serves to even out large scale battles when properly constructed, while ‘rock paper scissors’ balance designs serve to cripple MMO balance.

The old rock paper scissors game functions very well when applied properly to gaming models, but only when done at the proper level. Rock paper scissors should never be implemented in such a way that players do not have the option to change their strategy. Instead it should be implemented in such a way that it allows flexible and quickly changing strategies. This is seen in real time strategy games where a player who is amassing many aerial units may be countered with land units that are particularly strong against airborne attack. It’s seen prominently in fighting games as well, which often emulate an ever-flowing literal version of the rock paper scissors game in which players attack one another with rapid sequences using low, high and mid-level attacks to counter one another. However, this type of design has no place in any game where rock, paper, or scissors is a permanent decision. Instead games should strive for balance which allows players to all be equal, while allowing different strategies and playstyles to thrive. And that is the problem with scissors – they divide things and create a thin line which can be almost impossible to repair.


5 Comments leave one →
  1. Squirrt permalink
    April 21, 2009 7:50 am

    Good right up. I think that you might want to consider breaking your ranged option into 2 though. I play a squig herder on magnus and I very seldom defeat a Iron Breaker, Swordmaster, Knights or WL’s. But the Sorcerer I run with defeats the very quickly and easily.

    In reflection, I don’t think WAR will be able to accomplish what you are suggesting. The reason is that the original concept was designed around grand large battles. But now they have appeared to have lost this focus and are trying to balance the sides. IMO this is what has lead them to where they are currently, no real direction. The original designers are trying to go one way but are being pulled by a player base another way. Creating what appears, at times uncontrolled gut reaction changes to the game.

    For instance in 1.2 if they had not buffed Bright Wizards damage, nerfed the talisman resists, and hard capped the resistances at 40% all at once, but phased in the changes and getting some feed back you would not have such a huge knee jerk reaction of 1.3 to the AOE stacking damage.

  2. April 21, 2009 8:15 am

    @ Pancakes: You just applied the principle of Sabremetrics to WAR – well done!

    @ Squirrt – I agree. When you change too many variable during the course of problem determination you tend to cause cause unforseen problems. Of course they have to balance slowing the phase-in of their fixes against customer complaints. It’s not a position I’d like to be in if I were Mythic.

  3. April 21, 2009 11:04 am

    Very astute analysis… though from my experience tanks can’t beat ranged heals in solo play .. both as a tank and as a ranged healer.

    This is a problem with a PvP-focused game, this sort of micro-balancing is really necessary to optimize PvP play but it doesn’t have the same appeal to potential subscribers, casuals, and people outside of the game that other game play or world enhancements would have.

    Once you commit to a PvP focus, I think, you agree to sacrifice the rest of the game a bit.

  4. pancakez permalink*
    April 21, 2009 1:04 pm

    @squirrt: The charts are a representation of how it works on average, not really how it works for every possible setup. For that you’d have to work out how that balance web works, because WAR is balanced in a very random manner. Squig Herder, Magus, Engineer and Shadow Warrior all lose to tanks. Bright Wizard and Sorceress typically lose to tanks that have better damage, or a lot of defenses with a two-hander (ie: Blackguard, Swordmaster, Ironbreaker).

    As far as the huge kneejerk that 1.2 caused, I would object that this was caused by the testing cycle of 1.2. Patch 1.2 was only tested in depth for roughly four days. Other companies with this large of patches test for months. Even when 1.2.1 was released Mythic was still throwing in bug fixes for things that were broken by 1.2.

    @Czarnal: Tanks in this analysis were considered to be geared for open PVP, which means very high strength and specced for CC. I’ve never seen a healer able to stand up to a tank that properly controls the fight. Detaunts get countered by taunts. If the healer tries to fight back the detaunt is removed and AP is wasted, then the healer is interrupted with taunts, knocked down when they try to recover and just doesn’t have a way to adequately fight back.


  1. Long time coming « Book of Grudges

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: