The Game Design Forum

Dungeons, Durability, and Design

Follow tgdfweb on Twitter

The Short VersionThis section uses a statistic called RDur to measure the average length of encounters in dungeons throughout the game. The results of this study reveals that generally, most dungeons are very close in terms of encounter duration, from the beginning to the end. If this is true, it follows that the increasing difficulty necessary for making the game fun must come from another design feature. This feature is identified as being the "long game," which is the practice of making an entire dungeon into the meaningful unit of content, rather than the individual encounter. The final dungeon of FF6 is examined to illustrate how this happens, through the use of various battle-level attrition designs.

HP, defense, evasion, and all the complex equations that make them function boil down to one essential thing, a stat which I call real durability (RDur). RDur is a measure of how many attacks of a certain type character or enemy can withstand before becoming incapacitated. Ultimately, FF6's battles--like most RPG battles--are about who falls down first. The enemy can endure X number of hits, the player characters can endure Y. These variables are the respective RDur stats of each side of the battle. Obviously the RDur for each character, enemy and attack is different, and there are modifiers like healing (which I do take into account, below), but this is a pretty simple way to look at how difficult any given dungeon is, relative to other dungeons. Because RDur scales perfectly with level, it makes the designed difficulty of various challenges understandable as a part of a greater whole.

RDur is made easy, both for us as analysts and for the designers who made the game, by the standardization of stats in FF6. The standardization of stats is a simple practice, common to many RPGs that allows the designers to adjust difficulty on fewer axes. You can visualize it here. It forms a bell curve, albeit an ugly one because there are too few categories for the line to be smooth.

Percentiles of Damage Reduction

The average enemy defense stat is about 112, or a 43% reduction from raw damage. More than a third of the enemies in the game have a defense stat within 5% of this reduction. And, as you can see the same is true for magic defense, although the most common reduction for magic is in the high end of the 50 percent range.

There are numerous possible explanations for why magic defense is higher, on average, than its physical counterpart. Foremost among them is that the formula for determining spell damage is more favorable to the player than its physical counterpart. It's a lot easier to puff up a character's magic attack than his or her physical attack, as we saw on the last page. Magic defenses rise to prevent exclusive reliance on magic, powerful as it can be. The second reason magic defense is higher, I think, is to offer a strange solution to the even stranger manifestation of the Sorceress Problem that appears in FF6. (But more on that below.) Magical RDur and Physical RDur correspond closely, except in two deliberate cases. The Magitek Factory sees a rise in physical RDur that is significant for that moment in the game, but a corresponding drop in magical RDur. It makes sense from a gameplay angle: the player has just received their first Espers and magic, and the game is teaching them how to ration their MP in a longer dungeon. The other example is the Ancient Castle, in which magic is almost completely unusable. This makes some amount of sense from a plot perspective: the place has been sealed since the War of the Magi, and so everything within is well defended from magic. It's actually downplayed on the graph, the actual RDur is much higher (around 50). Short version: don't use magic there.

Real Durability in FF6 Dungeons

(A note on my methods. It's impossible to predict what kind of attacks a player might choose, so when creating an RDur score, I used two methods. For physical damage I used the average of individual RDur scores "fight" with a weapon relevant for that level and the "Dispatch" command. "Dispatch" and "Pummel" are effectively the same--armor piercing attacks of similar yield. You could throw the "Drill" command into the equation, and the result would be a lower aggregate, but the overall trend would have all the same proportions. For magic, I averaged the RDur of enemies vs the first two elemental spells in the WoB, and the second and third elemental spells in the WoR. Because of the ratio of their power to their MP consumption, these spells are very commonly used.)

What's really bizarre is that RDur remains more or less constant throughout. The monsters in almost every dungeon have roughly the same RDur, meaning they all take about the same number of turns--on average--to defeat. So what makes a dungeon difficult? It's not the life of the enemies, if RDur tells us anything. Instead, it must be the way that enemies attack.

Player RDur and the Long Game

In creating the final dungeon of FF6, the development team employed every single design "trick" that had been used in the game so far. As such, Kefka's Tower is a great place to look at the meaning of all those stats, strategies, and such, and how they make the game challenging. There's plenty of data from other dungeons here as well, but a discussion of every single dungeon in the game really wouldn't be worth much for the same reason that bosses aren't worth discussion. Dungeon designers can learn a lot about pacing, treasure distribution, etc from the dungeons, but specific layouts aren't much help. The feature most in need of consideration is FF6's manner of making the entire dungeon a problem to be solved, rather than just individual encounters.

Kefka's tower is the best example of playing the "long game" in an FF6 dungeon. In the age of regenerating health and shields, discrete modules of content, and raid bosses, the long game is often neglected. In FF6, the long game is the ability of the player characters to make it through a dungeon in one attempt. HP and MP are never automatically recovered between battles, and with many battles in any dungeon, it's necessary to find a way to replenish one's player characters. Because of the lack of quantity in significant healing items in the second half of the game, magic is the only sustainable, reliable way to recover HP lost across these many battles. Thus, the player's ability to make it all the way through a dungeon depends largely upon how long the MP pool of their characters can hold out. (Although this is affected by many other factors as well.)

When it comes to the MP pool, FF6 players are governed by two spell-specific design ideas that are curiously at odds. These two design features are the Caster's Advantage and the Sorceress Problem. The Caster's Advantage is a balancing feature for the long game, common in many single player RPGs. The advantage in question comes from a situation in which the cost of a spell is fixed at an absolute amount, while the yield of the spell grows proportionally with its task. This happens in FF6, but only until the beginning of the World of Ruin. You can see on the graph of the effectiveness of the "Cure" spell.

The Healing Plateau

The early growth of healing from a single MP pool makes sense if the dungeons are getting more dense with difficult encounters. So why does it plateau--and even shrink a little bit--when the hardest dungeons come along? I can think of two answers: one is the ease with which players can grow their magic power stat for meaningful results. The other answer is that in the WoR, there are a large variety of accessories that can either drive up the power of healing (and damage) or drive down the cost of each spell. In that regard, eliminating the Caster's Advantage makes a lot of sense; it forces tactical decisions about equipment.

This decision is a little strange, however, when one considers the fact that the FF6 team clearly chose to step into the Sorceress Problem. For a more detailed discussion of the Sorceress Problem, you can read more at the going deeper page. The short version is that area-of-effect and multi-target spells are usually disproportionately powerful. For some reason--a reason I don't entirely understand--when a spell or ability in FF6 has more than one target, its damage (or healing) is cut in half, no matter how many targets it has. So, an attack spell targeting three enemies will actually do more total damage (not accounting for defense) than a spell concentrated on a single target. As far as I can tell, this is a free lunch.

With magic, this is forgivable, because even though it would be easy to kill ten or twenty or even fifty enemies with a single spell, the player is never given the opportunity. Battles never exceed six enemies on screen at the same time, and they rarely exceed four. Accordingly, the amount of MP spent on multi-target spells is reasonably commensurate to the yield of those spells. When you take into account that attack magic is drawing MP away from healing, there's at least some modicum of tactical tradeoff.

With non-spell damage, and especially through most of the early game, multi-target damage is a giant problem because it heavily favors a small number of characters. Sabin, Edgar and Cyan all have MP-free attacks that hit all enemies on screen for decent damage. This happens before magic is available to all, and even after magic is introduced, these attacks are still MP-free. To me, this seems harmful to the design, because the core principle of FF6 is that characters should be roughly equal. While the introduction of powerful, abundant magic can solve this problem, neophyte players are not going to realize this quickly enough, and may suffer through a kind of "doldrums" period in the final dungeons as they rally their severely neglected characters. (Although I will credit FF6 at making even this circumstance, at worst, a fairly brief annoyance.)

It seems that there is an easy way out for FF6 that the designers didn't use. First, use different fractions for different scenarios: halve the damage of only certain multi-target attacks, and cut other abilities by more or less. Why not? Players will catch on pretty quickly to which spells are good as multi-targets and which aren't. The other solution is to simply have every spell with a multi-target option be cut by n, whereas n = the number of monsters on screen. Since there are rarely more than 4 monsters on screen at the same time, how would this be too much of a challenge? I think that this would make for interesting tactical choices: should the player use multi-target elemental attacks, or single-target armor piercing attacks? Every situation would require different actions.

Enemy Design and Player RDur

We still need to figure out some frame of reference for the RDur of the player, especially as it shapes the long game of a dungeon. As explained earlier in the assesment of stats, both magic and physical damage grow in a linear way relative to the level of the caster. This is important because it means that the RDur of enemy-against-player is not going to change as the game goes on--especially when you take into account the player's ever growing armor total. Moreover, none of the enemies in Kefka's tower fall into the top decile of stats for attack power or magic power. (And few even fall into the top 20%.) Something else is making these enemies difficult.

What the FF6 designers did to achieve the difficulty was actually terrifically clever. Not only were they able to make dungeons as difficult as they wanted, but they were able to make it very easy for themselves. The RDur of any given battle is more or less a given--it's going to hover around the average. So that's one decision already gone. Crafting the monster's roster of attacks is easy because of the standardization of stats. A designer simply picks a standardized stat tier from the bell curve, figures out the damage it's going to do, and factors in the armor the player characters ought to have. The designer figures out what the player character's RDur should be, and chooses a multiplier to determine a player character's estimated RDur. The result is the inflated "special" attack most monsters come with. There are dozens of names for these special attacks, "Swing," "Mirror Orb," "Tackle," etc. The important thing is that these abilities calculate normal fight damage, and then multiply it by a variable. If you track that variable across the course of the game, it shows a very clear trend: it goes up. A typical special muliplier for the early game is about 1.5 to 2. By the time you've reached Kefka's tower, most special attacks are hitting for 4-6 times normal fight damage.

Of course, that's not the only trick the designers have in their repertoire of difficulty modifiers. There's an entire array of attacks which, unlike the multiplied "specials" ignore armor completely. These attacks are:

-Status debuffs: armor doesn't do a thing against poison, seizure, zombie, sleep, slow, stop, freeze, etc. Some of these debuffs do damage, reducing RDur directly, but many of them simply incapacitate the player characters. For each incapacitated character, it's harder to defeat the enemies, who continue to pummel the characters. (Certain accessories can block some of these debuffs, at the tradeoff of losing other protections or enhanced battle power.) As you might expect, Kefka's tower is filled to the brim with enemies who can inflict these negative effects--particularly the Evil Oscar who can inflict almost all of them on the entire party at once. You'll be seeing a lot of him.

-Fraction attacks: more than any other game I've ever played, FF6 uses many attacks which reduce character HP to a certain fraction of total. Attacks like "Launcher," "Quartr," "W Wind" are frequent and often very likely to hit. There is no way to change the percent damage these attacks do. In Kefka's tower there aren't a huge number of these, but where they are used, they're particularly severe. Several sections have the Fortis enemy that frequently retaliates with "Launcher," for massive damage. A few sections have the fearsome Doom Dragon enemy that can reduce the entire party's HP to 1. Talk about cheating the RDur!

-Instant KO attacks: what game doesn't have a few of these? Instant KO attacks are pretty self-explanatory, and the only important datum is: how often does it hit? In Kefka's tower, there are quite a few IKO attacks, many of which have high surprisingly high accuracy, although because of behavior programming players tend to only see one of them per battle.

Most of the enemies in Kefka's tower are programmed to use their armor-ignoring attacks just before they die, or when are the last enemy standing. This is actually a game-wide practice, making sure that battles don't always become easier with the defeat of each enemy. (One enterprising designer saw this behavior, and then organized single-enemy encounters, which result in that enemy using its 'last stand' attack immediately, as a form of added attrition.) The big difference in Kefka's tower is that these attacks are so frequently fatal or otherwise severely damaging.

The Two-Attempt Dungeon

The last thing that contributes to the complex "long game" of Kefka's tower is the level scale. The first enemies in the dungeon are on average, level 50. The enemies at the end of the tower are about level 55. There is not enough time for the player characters to gain those five levels during the course of the dungeon (at most, they'll gain 2 or 3), and so characters who don't start out that high are going to get pummeled a bit harder than usual. Most players aren't going to enter the dungeon prepared to replensish themselves enough to level up to 55--at least not the first time.

A second trip through the dungeon (courtesy of the Warp spell) almost certainly will get everyone up to a sufficiently high level; and there are more reasons why a second trip will be necessary than attrition. For one thing there's a surprise boss fight against Inferno for party three. I'd wager that quite a few newbies put their worst characters in that last part, and suffered a humiliating defeat. Reorganization requires exiting the dungeon, but hey at least you got all that EXP, right? They can't take that away from you. Also, there's a heck of a lot of gear in that final dungeon, and it's better for the player to unequip everyone and take a look at the total equipment pool at once, when making decisions. That, too, requires the party to exit the dungeon. And of course the player, now aware of just how much endurance the final dungeon is going to take, will proceed with more save-point supplies, etc.

The Post-RDur World

RDur loses some of its extended meaning in the era of modular content. I understand regenerating health; it can be very annoying to be stranded somewhere you can never exit, because you're one battle away from death, but five battles away from home. It has happened, and it's a dealbreaker. I think, however, that a lot of designers take refuge in the notion that modularizing content is always the best practice. By doing so, they never have to balance anything beyond the set piece. There's no need to think about how that set piece flows into the next or previous one, just throw the player into a differently shaped room with more or fewer chest-high walls.

Still, RDur serves well for assessing the difficulty of individual encounters, particularly in games rich in simulated skills. For all the inflated numbers, formulae, etc, every battle is basically about who falls down first. RDur isn't the only stat that matters, but it's great for dispensing with all the window dressing that can sometimes obscure what a designer needs to know.

Trivia!The Genji helm would provide you with the best RDur of any helm... if you were just counting defense. The Red Cap provides a 25% increase in HP in addition to its defense, which blows everything else out of the water. Why didn't I include evade into my assessment of RDur? Evade is meaningless because of a bug; magic evade controls both stats. Of the enemies with enough magic evade to actually matter, only one of them is loose in a dungeon (Pug), and many of the best spells are inevitable. About one in ten enemies has an evade stat that might have meant something, if the system hadn't been bugged.

Examining the Design of FF6's Level-Up System

Next: A Final Look at FF6

Site questions?

All material copyright by The Game Design Forum 2011