The Game Design Forum

Game Difficulty and the Four Phases of FF7

In the first section, on the history of the RPG form and FF7’s place in that history, I talked about how the form of an RPG affects its style and content. The transition from the tabletop space to the digital space necessarily caused the design of RPGs to change. We’ve already seen how the limitations of the digital space—combined with the astonishing completeness of D&D—led to RPGs that focused on one or two aspects of the form. But there are other changes that arise from the transition to videogame form, and one of them is the adoption of orthodox videogame difficulty structures. Between 1978 and 1986, there was a tremendous amount of evolution in the design of conventional videogames. Final Fantasy 1 was released in 1987. It only makes sense that FF titles would be influenced by this evolution.

Two principles form the foundation of videogame design. The first principle is that games should generally become more challenging the longer they are played. The second is that this increase in difficulty is not purely linear, but rather goes up and down.[51] Both of these concepts are important to the design of FF7, but I want to address the latter concept first. Final Fantasy 7 alternates easier and harder quests throughout the game, and many statistical measures reveal this pattern. With a few exceptions, the difficulty of FF7 dungeons features a reliable pattern of tension and release that can be charted clearly. On the other hand, the way in which the designers create a game-long increase in difficulty is much more complicated. The big problem that the FF7 designers face when trying to turn up the difficulty across the course of their game was that many of the traditional RPG mechanisms for increasing difficulty don’t fit the voice or structure of their game. Tactical party composition had already disappeared in FF6 in favor of a party made up of whomever the player wanted to use for plot-and-character reasons. As we’ll see, this doesn’t leave the design team without tools. Even without tanks, dedicated healers, and fragile spellcasters, the designers have ways of keeping the game interesting. Debuffs, elemental resistances, specially-structured attacks and changes in enemy base stats are a few of the traditional ways that the designers were able to modify the difficulty of battles. Novel structures in the design of the level-up system are another way of playing with game difficulty. Overall, the game breaks down into four phases, each of which represents a different approach to making the game challenging.

  • The first phase is an introduction during which the player learns the game’s basic mechanics and can push through most of the content without much critical thinking or knowledge about the game.
  • The second phase adds difficulty by inconveniencing the player in various ways over long periods of time, rather than concentrating danger in any one spot.
  • The third phase of the game is almost the opposite of the second phase, in that it’s all about player characters and their enemies using high-powered attacks.
  • The fourth phase of the game combines aspects of phases two and three, but also uses the level-up system in a novel way to create (and solve) difficulty.

Although the transitions between these phases are so smooth that many players miss them, we’ll see how the designers managed to pack quite a bit of nuance into FF7’s battles and level-up system.

One thing I want to address before diving into the section-by-section analysis is a prejudice which I have encountered many times when doing research on RPGs and JRPGs in particular. There exists a notion that all of the difficulty in a typical JRPG stems from inflation in the base statistics of enemy combatants. That is, JRPG battles become more difficult because monsters have increasingly large HP pools, abd attack and defense stats. This is not really the case with FF7; the truth of its difficulty is much more complicated. In addition to showing that FF7 is not the boring, stat-inflation game that some critics think it is, I also want to address the validity of this prejudice in general. There are games that do consist of a long, uphill battle against progressively higher monster stats. World of Warcraft is one example of this. Although the endgame (raid) content for WoW is actually somewhat nuanced and skill-based, most of the content prior to the endgame is simply a series of dungeons and quests that pit the player character against progressively stronger versions of the same old monsters, over and over again. Considering that this part of the game can take 50-100 hours to climb through (although this climb has gotten easier in more recent versions of the game), it’s fair to characterize the levelling phase of WoW as being exactly the sort of RPG that critics say requires more grinding than skill. Skyrim is another game which suffers from stat inflation as the primary means of challenge. The main difference between the enemies at the beginning of Skyrim and the enemies at the end is that the later enemies have more HP and better equipment. I am not the first person to point this out, but I think that critics of WoW and Skyrim tend to admit that to see those games only in terms of their battle systems is to miss the point. Skyrim and WoW are about more than just finding the next monster to kill. Indeed, most RPGs are about more than finding the next monster to kill. The exploration, atmosphere, character and storytelling aspects of RPGs are Dave Arneson’s enduring contribution to the genre, and ignoring those things is (in most cases) a needlessly reductive exercise.

I have to admit that when I looked at FF7’s statistical measures for the first time, I thought I might encounter the kind of inflation which I discuss above, but it wasn’t so. The easiest way to prove or disprove this notion is to look at how many physical attacks it would take to end a battle, as this is the most basic action a player can take. Physical attack damage can vary, so for my measure I used Red XIII’s physical attack as he falls roughly into the middle of the damage spectrum. Also, he joins the party during quest six (breaking into Shinra HQ), which is the point in the game where the party-size stabilizes at three members for the rest of the game. I measured his attack against every enemy, assuming that his character level is equal to the average enemy level for the quest in consideration (rounded down). For every quest in which there was a new weapon available, I figured that weapon into the calculation.


Over the course of the game, the number of physical attacks required to end a battle does go up slightly. That growth is considerably smaller than I expected, however. Moreover, the way the growth occurs is not in the manner that I expected, either. I did anticipate the possibility of the up-and-down motion of this graph—that some quests would be harder than others based merely on things like HP and defense. But the strange thing is that the average number of physical attacks required to end a battle (per quest) never goes much above nine attacks—not for the entire game! I thought that the battles in the hardest quests would require an average of eight attacks, then nine, then ten, then eleven, but my hypothesis was wrong. Instead of seeing the hardest quests getting harder (in terms of stats), we see the easiest quests getting harder, while the hardest quests stay the same. Or, in other words, the durability of the enemies in the easier quests grows, while the durability of the enemies in the harder quests stays mostly the same (in terms of basic physical attacks).

The fact that the average battle peaks at nine attacks is important, but how I arrived at that average is important as well. The FF7 designers appear to have imposed a consistent limit on the average duration of battles (measured in player turns). I don’t want to conceal the truth of the data from that graph, however. Each data point represents the average length of a battle per quest. Naturally, there are some battles that are higher than the average, and some battles that are lower. There are some battles whose statistical profile is so extreme that it skews the average of the whole quest. These battles generally fall into two groups: battles with an unusually large number of enemies, and battles with one huge enemy. The former category is about as simple as it sounds in terms of composition. These battles are made up of normal enemies in large quantities, and thus they scale normally in terms of EXP, AP and everything else. The only really interesting thing about these battles is that they disappear completely just before the Great Glacier quest. By my count, there are 48 random battles (not including chocobo battles) which contain either four or five enemies, but none after the player leaves the Coral Valley. How can we explain this? I think the answer lies in the player’s ability to use multi-target attacks. Battles with many normal enemies take a comparable amount of time to finish as battles with one very tough enemy—but only if the player relies entirely on basic physical attacks. The moment a player starts using multi-target abilities, he or she can finish the large-population battles in one or two turns. This isn’t true against those beefy enemies, since their HP is concentrated in one spot. As the game goes on, the player’s access to powerful multi-target attacks increases markedly, meaning that high-population/low-HP battles would be way too easy for a player who has the Slash-All Materia, or any of the many other skills which can wipe a battlefield clean of weaker foes. The harder enemies prevent the player from abusing multitarget abilities to gain EXP, AP and Gil at a rate that would break the game.

The “beefy enemy” category is a little more interesting from the perspective of individual battle design. The one-tough-enemy setup is typified by the single most durable random encounter in the game: Master Tonberry. The Tonberry appears in many FF titles, usually as exactly what he is in FF7—a monster with high HP and some deadly attacks. There’s no easy way around the Tonberry; he takes about 35 (average) physical attacks to defeat, and he has no glaring magical weakness to exploit. The Master Tonberry appears as a random encounter in the final dungeon, and so his toughness is completely understandable. There are other enemies like him from many different quests, however. The Blue Dragon of Gaea’s Cliffs takes an average of 15 attacks to subdue, while the Dragon of Mt. Nibel requires 13. The X-Cannon from the Return to Midgar requires 18 attacks to defeat, while the King Behemoth in the final quest requires 15. These enemies hand out an average to slightly above-average amount of EXP. The one exception is the Ying & Yang enemy from the Shinra Mansion, which takes a whopping 30 attacks to defeat, but which delivers about 57% less EXP than a typical battle in that zone. The reason is that Ying & Yang are a gimmick battle; one side is nearly immune to physical attacks and the other, magic. This is one of a very small number of gimmick battles in the whole game, but it does skew the data for the Nibelheim/Mt. Nibel area a little bit higher. Thankfully, there are not really any other significant instances of gimmick-based defenses in random encounters, and so the numbers on the physical attacks graph are a fair representation of the game as the player experiences it. (We’ll go over this in greater detail in the later section on enemy archetypes.)

Stat Inflation: Sometimes Meaningful, Sometimes Not

Looking at how many basic attacks go into a battle is the most consistent way to compare enemy durability across quests, as it accounts for level, HP, defense and even player-character statistics. That said, this method does not account for many of the other stats that enemies (and player characters) have. I expect that the first big question readers have regarding statistical inflation is about the enemy’s ability to do damage. Even if the average random encounter lasts no more than nine attacks across the course of the game, what the enemy can do during those turns is pretty important. I want to start out by simply looking at the average stats of enemies on a per-quest basis. Here is a graph of average enemy attack stats, per quest.


You can see that there’s a very strong relationship between quest progression and enemy attack stats. This would seem like a clear indicator of increasing difficulty in the game except that it is mitigated by rising player defenses.


Enemy defenses and player attack power rise similarly, but that data is already baked into the durability graph above, and the relationship is neatly linear. This is a form of stat inflation; it doesn’t present the player with much to think about. It does force the player to buy new weapons and armor, and as we’ll eventually see, that’s meaningful; the acquisition of gil becomes more important during phase four. For most of the game, however, the player simply maintains his or her party’s stats in line with the inflation of enemy stats by buying stronger equipment. As far as enemy base damage and armor go, FF7 is guilty of boring stat inflation—at least until the final phase when the use of gil becomes a little more interesting. Although physical attacks and defenses inflate somewhat meaninglessly, the same is not true for magic and magic defenses. Enemy magic stats increase in a straightforward, linear fashion.


Player magic resistances also rise commensurately, recreating the meaningless inflation of the physical defenses. I have on the graph the physical resistances throughout the game to show how, despite starting lower, the magic resistance on armor catches up to its physical counterpart pretty quickly. What saves the magic/magic defense dynamic from being too boring is that enemy magic defenses are quite a bit different.


This is the first glimpse we get of the multi-phase structure of the game’s difficulty. Indeed, this drop-off in magic defense after the Temple of the Ancients was one of the first indications I had that FF7’s difficulty structure could be broken up into distinct phases. That surprising valley in the shape of the magic defense graph is a major part of phase three of the game, which represents a stark and sudden change in the way that battles challenge the player.

Although the mid-game valley in magic defenses is obviously significant, the early zenith of magic defenses is important too. The dialogue analysis shows us a clear “first peak” at the Temple of the Ancients, and the magic defense graph seems to coincide with that. The early peak is typical for a Squaresoft game, and appears in most FF titles from the 1990s, as well as in Chrono Trigger. This peak sits at the end of a long line of quests with high magic defenses, however. Why would earlier enemies be more magic-resistant than later enemies? This question led me to examine the portion of the game which goes from the Shinra Tower until the Temple of the Ancients, and to try to figure out why enemy defense stats would be distributed so counterintuitively. The answer to that question is what revealed the nature of phase two of the game. I’ll explain all that in the section on phase two, but the point I want to make here is that most of FF7’s statistical dimensions did not show boring, predictable and linear growth across the course of the game.

Phase One: Introduction

The first phase consists of the first five quests, from the opening bombing mission up through the destruction of the Sector 7 pillar. It is not easy (or helpful) to statistically compare the beginning of FF7 to the rest of the game for several reasons. The foremost reason is that the party size is not stable, moving between one and three player-characters across the first five quests. Although party composition is not tactically nuanced in this game overall, the player is frequently forced to carry party members who are under-leveled and underequipped during the first phase. Thus, the difficulty of each battle has to be calibrated to take into account a party of frequently varying size and power. There are many other reasons why this phase can’t really be put on the same graphs as other phases. There are relatively few shops during the first phase. The shops that do appear don’t offer weapons, armor or materia as often as they do in later phases. The rate at which the player acquires EXP and AP is different from the rest of the game. In essence, all of the ways in which the player characters would encounter and then master difficulty are different in this phase than they are later on.

Although it’s not terribly helpful to compare early statistical trends to the rest of the game in any great detail, the first phase does show the player all of the essential mechanics. Elemental resistances are a good example of this. The player only has basic fire, ice and lightning spells, but a huge number of enemies are weak against lightning, and so the player has plenty of chances to learn about enemy types and their elemental affinities. Meanwhile, only two enemies in random encounters are immune to a spell the player has. Both of those enemies are located in the Train Graveyard, which is part of the last quest in the phase. The designers are teaching the player through ease and power rather than difficulty and resistance. Similarly, the enemy’s use of debuffs is minimal and incomplete. (Debuff is the catch-all term for the family of spells which Final Fantasy calls “status effects.”) In the first phase of the game, there is only one enemy that can inflict slow, one that can inflict poison and one that can inflict silence. Two enemies can inflict sleep and two can inflict darkness. None of the really dangerous debuffs are present. Moreover, all of these debuff-using enemies are spread out so that no one zone in the first phase is especially rich in debuffs. This is in contrast to the later parts of the game which tend to concentrate debuffs heavily. The designers are introducing game mechanics, but not in a way that is particularly dangerous.

The underlying statistical elements in the beginning section don’t compare well with the rest of the game, but there are still some numerical elements which are indicative of the low level of challenge in the beginning of the game. The FF7 designers did themselves a huge favor when they constructed the damage formula for enemy combatants. Most of the enemies in the game have an attack which is equal to some multiple of their base attack damage. (I’m going to cover this in much greater detail in phase three, where it’s even more relevant.) For example, the formula for the Guard Hound’s special attack “Tentacle” is simply base damage*1.5. I call this modifier the component stat, in that it is a kind of contributing statistic that is a component of the ability itself rather than of the character casting it. The component stat is one of the clearest ways in which we can measure increasing difficulty across the course of the game.

In the first phase of the game, most of the component stats for most attacks made by non-boss enemies are below base damage*2. Only two enemies have attacks that equal or exceed that amount. Moreover, there are only two attacks in this phase that can hit multiple targets. Interestingly, the Hell House enemy in Sector 6 has the highest damage attack in addition to one of this phase’s only multi-target attacks.

Based on its location in the game, I think that the Hell House’s unusual damage output is in place to help Cloud learn his second limit break, but it also serves to teach players that some normal enemies are much more dangerous than others. Most enemies in the first phase aren’t like the Hell House. Highly damaging and multi-target attacks will become more prevalent across the course of the game, and players need to know about that. But in the first phase, most of the damage dealt by enemies is at or near the levels dictated by enemy base stats. Bosses in the first phase are different from normal enemies, and really, they’re also different from bosses throughout the rest of the game. Bosses in FF7 tend to have several shared characteristics: large pools of HP and MP, low physical defense, high magic defense and one or two dangerous attacks which are used infrequently.

Only about a third of bosses in the entire game have an elemental weakness, but all but one of the bosses in the first phase have an elemental weakness. (Of the bosses who do have elemental weaknesses, 60% of them are weak to lightning.) And while the high magic defenses of bosses make elemental weaknesses less meaningful, it’s still clear that the bosses in phase one are supposed to help teach the player to exploit any advantage. Bosses in phase one also do other unusual things. Many enemies in the game have counterattacks, and sometimes it’s not even clear when a counterattack is happening. The Guard Scorpion’s counterattack, on the other hand, is actually explained during the battle.

Many enemies and bosses have conditions under which they are more vulnerable, but the Air Buster (the game’s second boss) takes six times more damage when it is hit in the back. Many bosses use debuffs that require the player to target the victim with a cleansing ability, but Reno’s pyramid technique requires the player to specifically target the debuff itself, which is a great way to force players to think critically about debuffs. Overall, the bosses in the first phase are set up so that the player can see explicit versions of game design ideas that will be subtle in later phases of the game.

Phase Two: Abundant Inconvenience

Phase two of FF7 is best characterized by an abundance of small inconveniences for the player characters which don’t really present a terrible danger, but which nevertheless make the game feel progressively more challenging. There are three design features which cause these increasing challenges. The first is an abundance of elemental resistances, which peaks surprisingly early in the game. The second is an abundance of low-level debuffs, which peaks around the Temple of the Ancients. The final feature is increases in magic defense, which also peak at the Temple of the Ancients. Although none of these are purely linear (because that would defy the essential voice of videogame design), they do see an overall increase toward a clear climax. In trying to understand the structure of phase two, it’s helpful to visualize the enemy’s use of debuffs.

Charting the prevalence of debuffs confirms a clear, quantitative “first climax” of which we have already seen evidence of in the textual analysis and magic defense graphs. Below is a graph of debuffs starting from quest six (breaking into Shinra HQ), the point at which party size and party composition stabilize. The graph continues through the final incursion into the Northern Crater.


To explain how I came to my conclusions about phases of the game, I want to explain how I represent the data here.

  1. 1. The quantity of debuffs is calculated on a per-battle basis, rather than on a per-enemy basis, because this is more accurate to the player’s experience of the game. Therefore, each data point represents the average number of debuffs that are available to be cast per battle in each of the quests.
  2. 2. The graph uses weighted data; some debuffs are worth more than others. Each debuff is weighted based upon the number of turns it costs to remedy. Debuffs like poison, blind, and silence can be removed in one turn by the character who is afflicted by them, and so get a score of one. Debuffs like confuse, paralyze, petrify and stop consume two turns; the character who is afflicted by the debuff loses a turn while afflicted, and another character spends a turn remedying the debuff. Those debuffs are thusly scored as twos. The only effect scored as a three is instant death, because it consumes the turn of the victim, a turn to revive, and a turn to heal the character that was just revived so they don’t die again immediately. Many enemy AI scripts are configured to target the character with the lowest HP, and so leaving a character in critical condition will result in another death. The spell life 2 (or the Phoenix summon) can reduce the number of consumed turns to two, but the Revive Materia takes a long time to level up, has higher-thannormal stat penalties, and costs a lot of MP to use. Thus, the number of turns spent on reviving the victim of an instant-death attack is only reduced at significant cost, and only if the player has properly prepared for it.
  3. 3. I have adjusted for the few instances where an enemy can only cast a debuff once or can only cast it when it is the last surviving enemy in the battle.
  4. 4. I have not included the Gelnika or Ancient Forest in this analysis as they have their own section later which explains the different rules those dungeons follow.

As you can see, the graph does not really go up steeply across the course of the game. If you plot a linear trend line for it, the r-squared value is less than 0.2. Still, there are a few things we can deduce from individual sections of the overall trend. The peak in debuffs around the Temple of the Ancients mirrors the peak in magic defense that occurs in that dungeon. Despite having seen that peak already, I expected that the use of debuffs across the whole game would increase, going from a few enemies that cast poison and slow spells to a large number of enemies who cast stop, paralyze and death. That was not the case. Instead, the difficulty in phase two of the game is mostly about inconveniencing the player with a large number of low-level debuffs rather than endangering him or her with a few higher-powered spells. The enemies in the Cave of the Gi are a good example of this. On the debuffs graph, Cosmo Canyon represents the highest point in the game up to that point, but the primary debuffs in the dungeon are poison and death sentence (which rarely takes effect in short battles). More rarely, the enemies will use paralyze, but it has a low hit rate. Poison isn’t particularly deadly, but in this dungeon there are usually several enemies in a battle that can apply it, and it will eat into the party’s HP. The same thing is true at the Temple of the Ancients. Enemies in this quest can apply poison, slow, sadness, darkness, frog, and berserk. (Two enemies can cast either paralyze or confuse, but do so much more rarely.) For the most part, these frequent, low-level debuffs are simply a way of lengthening battles and forcing extra effort out of the player without actually increasing the real risk of a game over.

While we’re on the topic of debuffs, I want to address a common criticism of FF7 and the series to which it belongs. There is a long tradition of criticizing certain FF titles (mostly FF4 through FF7) for filling the game with debuffs that benefit the party very little, but which benefit the enemies greatly. That is to say, when a party member attempts to inflict the silence effect, for example, the chances the spell will actually affect an enemy are quite low. This is a criticism that I think is true, but I think there is a structural reason for it. The average battle in FF7 should be over in about seven character turns if the characters do nothing but physically attack and never get a critical strike. Using any multi-target or enhanced-damage abilities will shorten battles even further. Even if the player were to successfully apply a debuff, that effect would barely have time to make a difference in the battle, so it’s not worth wasting a turn to try. Sometimes a debuff will be useful as part of another attack through the use of the Added Effect Materia, but such a technique rarely shortens battles by much. On the other hand, enemies can apply debuffs which last longer than the battle, which extend the amount of time spent in the battle, and which significantly increase the difficulty of those battles—especially if the player doesn’t have the remedy for them. Debuffs could still, in theory, be useful for the player in boss fights, since the greater length of the battle would give them time to have an effect. But most of FF7’s bosses are categorically immune to debuffs. Why does this need to be the case? Why couldn’t debuffs work on bosses for a few rounds and then wear off? Why couldn’t debuffs work against bosses but with reduced effectiveness? I can make no defense of this design decision; the designers simply missed an opportunity.

The incidence of elemental resistances follows a trend similar to debuffs, featuring a strong, up-and-down motion and a nadir where one might expect a zenith. For elemental resistances, I counted the raw number of resistances appearing in a battle. Even if every enemy in a battle had the same resistance, I counted each resistance as a separate instance. I felt that this was the best way to replicate a player’s experience of the game, since in the early going, many players would be reliant on one or two elemental attacks like Fire 2/All or the Shiva summon. The sight of four enemies resisting damage from such a spell is thusly replicated by giving such a battle a score of four (or more, if there are other instances).


Again, there is an up-and-down motion, although this time there’s not as clear a peak around the Temple of the Ancients. In fact, the trend for the game is a slow, consistent drop in the prevalence of elemental resistances. The distribution of elemental resistances still reinforces the idea of phases in the game. Phase two has the most elemental resistances of any section, including the highest peak. Combine that with the relatively high magic defenses of enemies in this phase, and it’s clear the designers are creating another kind of abundant (but minor) inconvenience in that phase. Aside from the extremely early peak, I was most surprised by the nadir which comes at the end of the game. I honestly expected a reasonably strong upward trend in the overall prevalence of elemental resistances, and I am thrilled to be wrong because truth is usually more interesting than prejudice.

The up-and-down structure in the incidence of elemental resistances is almost certainly another case of the designers adhering to videogame orthodoxy and varying the difficulty of sequential quests. The line bounces around a lot from the very beginning and all the way through; there are no plateaus or valleys lasting longer than two quests. The most curious feature is the extremely early peak at Mt. Corel. I expect that almost nobody would name Mt. Corel as the one of the tougher dungeons, and yet it has the highest average incidence of elemental resistances per battle. So what are the designers trying to accomplish there? The answer is: nothing special. Mt Corel’s unusually high incidence of elemental resistances is inflated by an extremely abundant resistance to the earth element. The player could have bought the Earth Materia at Kalm, but because it costs 6000 AP to level up (whereas the Fire, Ice and Lightning Materia require only 2000 AP each), it’s not a likely candidate for use in clearing battles quickly by linking it to an All Materia. Moreover, the player has fire, ice and wind-based summons. So why do the designers insist on an earth resistance? It’s because the majority of enemies in the Mt. Corel quest are birds or monsters who have traditionally hovered above the ground in previous FF titles. By tradition, such monsters are immune to Earth, even though Earth is a virtual non-factor in this dungeon. This is what the graph looks like with the Earth resistances removed from that quest.


This version of the graph looks a lot more consistent across the course of the game. Obviously, that’s not the true shape of the game, but it’s representative of what the player experiences. Only players focusing specially on the Earth Materia are going to actually perceive the impact of the bizarre, early peak in elemental resistance.

I want to dig a little deeper in phase two to illuminate some things which the graph doesn’t capture perfectly. After the early (and arbitrary) peak in elemental resistances, their prevalence decreases for a few quests before climbing again.

Similar to Mt Corel, the graph point representing the Cave of the Gi is propped up by resistance to one element. Almost every battle in the cave features several enemies with poison resistance. While the player certainly could be using a Bio 2/All combination, it’s pretty unlikely, because no enemies have been weak to the poison element by that point in the game. Still, elemental resistances are more common in that dungeon than normal. After a few more quests without much in the way of elemental resistances, Wutai and the Temple of the Ancients represent another high point. In these quests, the elemental resistances are not only plentiful but also varied. Enemies in these last two quests resist gravity, ice, fire, earth, water and lightning in mostly equal measure. This is more in line with what we would expect from the difficulty structure of phase two: lots of small inconveniences, rising in number and diversity near the phase’s final quest.

Bosses in Phase Two

Bosses are an important part of the distinction between phases as well. Phase one bosses exist to highlight basic mechanics in the game, like back attacks, counter attacks, elemental weaknesses and debuffs. The bosses in phase two take the player on a kind of tour of the more advanced design ideas before shifting to a more damagecentric design at Nibelheim. Both parts of the phase are important, but the first part of the phase is clearly more interesting because of the different design ideas. Sample HO152 isn’t terribly exciting; it casts poison and has minions and is otherwise unremarkable. The Hundred Gunner/Heli Gunner and Rufus/Dark Nation fights are much more tactically varied. The Gunners fight the party at a long range, meaning that the only way for anyone other than Barret to attack them is to cast magic or use limit breaks repeatedly. The designers are simply forcing the player to do something other than use basic attacks, and to learn about rationing their MP pool as a part of that practice. Dark Nation will cast Barrier and MBarrier on Rufus, meaning that all of Cloud’s damage against him will be halved. I think it’s actually quite clever of the designers that they used barriers in a fight where Cloud can’t do anything about them. By forcing the player to play through a battle extended by barriers, the player gets the proper appreciation for what barriers can really do for his or her own party, and how important it is to get rid of them when the enemy has one (once this becomes possible).

Bottomswell and Jenova BIRTH are the two bosses that really stand out among the phase two lineup, but for totally different reasons. Bottomswell is unique among bosses in that it combines aspects of several earlier bosses into one. Although stacking up various challenges into one is one of the most orthodox design traditions in videogames, FF7 doesn’t do a lot of it. Bottomswell is the exception.

Bottomswell uses a tidal wave attack like Aps, it hovers at long range like the Gunners, and it has a prison attack like Reno. It is this last attack which is the most interesting from the perspective of traditional videogame design. Whereas Reno’s prison ability simply stopped a character from acting and could be undone by physical attacks, Bottomswell’s prison attack actually deals damage and can only be undone by magical attacks. Those two improvements that Bottomswell makes upon Reno’s Pyramid attack are the kinds of evolutions that would not be out of place in a Mario game. Strangely enough, that kind of evolved attack doesn’t appear again until phase four, and only really sees extensive use in the final battle with Sephiroth, as well as the optional fights against the Weapons. Jenova BIRTH stands out from the other bosses because she is out of phase. What I mean by this is that Jenova BIRTH has more attributes of phase three and phase four bosses than she does of other phase two bosses. In that regard, she makes for a good point of comparison for the rest of phase two. Earlier I mentioned component stats, which are the fixed stats of a spell rather than those of a spellcaster. Because the physical and magical attack stats of enemies experience inflation across the course of the game, it’s not that useful to try and compare one boss’s attacks stats to another’s. Component stats, however, compare quite nicely from enemy to enemy, and Jenova is unusually powerful for the section. Up until the Jenova BIRTH fight, only one boss has had an attack with a component stat of three or higher (the Air Buster). Jenova BIRTH has three attacks that powerful, one of which hits all party members, and the others of which she can use several times in a row. This would fit in perfectly well with the kind of enemies and bosses which exist in phase three, but in phase two, she’s an outlier. What’s more, because all of her attacks are so powerful, there’s no lull in which the player can forego healing and deal extra damage. There are other bosses in the phase that can deal a lot of damage like Jenova BIRTH can, but they don’t do it so consistently. The Materia Keeper casts Trine, which has a 2.125 component stat and hits all party members, but two of its other attacks only do base damage, so the player has a chance to regroup in between blasts.

I think the surprising strength of Jenova BIRTH is actually a great example of consonance between the story and the gameplay. Jenova is aligned with (if not a direct manipulator of) the main antagonist of FF7. She has unique boss music. She should be more dangerous than the average boss, and she is! Even the bosses at the Temple of the Ancients can’t compare to her relative damage output. The Red Dragon and the Demons Gate have no attacks between them with a component stat of 3 or higher, and only one attack with a component stat higher than two. Jenova outclasses them both. But what about her second form, Jenova LIFE? She has one very powerful attack in Aqualung, which has a 3.25 component stat. Her other attacks are either equal to base damage or lower than it. I think that the reason for her relative weakness is actually story-related as well. If Jenova LIFE could wipe out the average party too easily, the player would have to watch Aeris die over and over again. That would really spoil the drama of the event, and as we have seen so many times, the FF7 development team always puts the drama first.

Phase Three: Damage Racing

Phase three of FF7 drops most of the design ideas of phase two and replaces them with a race to see who can do damage faster: the player’s party or their enemies. The metaphor of a race works really well in the sense that the player’s best strategy in this phase is to finish battles quickly by using high-powered, multi-target attacks. The enemy will be doing the same thing. The race metaphor fails in the sense that, although each individual battle is a race, the dungeons are about long-term attrition and the rational use of the MP pool. We’ve already seen how magic defense and debuffs drop off after the Temple of the Ancients; that drop-off marks the beginning of phase three. It would be a mistake, however, to imagine that this signifies a drop-off in difficulty. Actually, enemies in phase three are actually a threat to the player for the first time. Most of this increased threat comes in the form of powered-up special attacks. Special attacks are unique enemy techniques like “Claw,” “Tentacle,” and “Extreme Bomber,” which the player never learns. These attacks are governed by component stats which rise markedly in this phase. Equally important, however, are the abilities which player characters gain. Increasing amounts of AP make it so that magic and support materia are more powerful, and the player can cast highlevel magic against multiple targets several times per battle. Beyond that, players also have lots of powerful summons, several of which deal non-elemental damage. The designers put these abilities in the game specifically for the player to use them in normal battles, and phase three gives the player a good incentive to do so.

Scaling Simple Damage

Even from the earliest days of the RPG, there have been unnecessarily complicated damage and hit formulas. Anyone who played D&D in the days of THAC0 (to hit armor class zero) can attest to that. The primary formula for damage in FF7 is fairly complex as well. Several times already, I’ve spoken about the how component stat was useful for tracking enemy damage output across the game. It’s important, however, to understand why that’s the case. Below is the two-step formula for determining physical damage.

(1) Base Damage = Att + [(Att + Lvl) / 32] * [(Att * Lvl) / 32]

(2) Actual Damage = [(Power * (512 - Def) * Base Damage) / (16 * 512)]

We’ve already seen that enemy attack stats and player defense stats (and vice versa) rise in step with each other across the course of the game. That is, the designer has a set of guidelines for how high a monster’s attack stat can be in a given quest. The same is true for enemy levels. Enemies can only be one or two levels higher in the Great Glacier than they were in the City of the Ancients or else the player will have to spend several hours grinding between quests. (There’s a place for grinding in FF titles, but it’s not in the mid-game.) The only part of the formula which isn’t limited to a fixed range by quest is the “power” variable, which is the component stat I discussed earlier. The component stat for enemy attacks is the place where the designers effectively determine damage-based difficulty for each quest.

Looking at the value of component stats across the course of the game tells us a lot about enemy design and the overall difficulty structure of FF7. Like everything else in the game, the value of component stats goes up and down. The graph below visualizes the value of component stats across the course of the game, on a per-battle basis.


Many enemies have more than one ability that has a component stat, and so for this graph I have only visualized the highest-damage ability for each enemy in a battle. I have also adjusted the value to reflect not just the value of the component stat, but also the number of targets that the attack can hit. On top of that, I added together the highest damage attack for each enemy in a battle as a representation of the total top-end enemy power available in that battle. Finally, I charted the average of all the battles in each quest on that basis. This isn’t a perfect representation of how each battle will proceed, since an enemy might not use their most powerful attack every time, but it does give a good idea of how much damage a given battle will dish out relative to other battles and other quests.

One point on the graph is a red herring that needs further explanation. The quest with the greatest damage output from random enemies in the entire game is, apparently, the crossing from Junon to Costa del Sol. I doubt very much that anyone who has actually played the game thinks this is accurate, and in a certain sense, they’re right. The data point is unusually high because the Marine enemy, which appears in every battle, has an attack which hits all party members for moderately elevated damage. Because these enemies appear in every battle in that dungeon, they pull the average component stat up significantly. The reason why nobody thinks of this as the hardest quest is that there are only two rooms with random encounters inside the belly of the cargo ship. Even players who explore every inch of the ship’s interior will only run into half a dozen encounters. So while the component stats for this section of the game are unusually high, the dungeon is too brief to be much of a threat. Or in other words, the data are skewed by a very small sample size. If we remove that unusual quest, we are left with a graph that tells a clearer story.

The last thing I want to do is to alter the data so that they reinforce my thesis, but there are a few good reasons for visualizing the data this way. Firstly, there is only one major exception to the low/moderate component stats which are found in phase two. If there were three quests which broke the pattern, there would be no pattern, but one exception does not negate the larger rule. Secondly, one of the things we have seen again and again in this book (and will continue to see) is that the designers of FF7 put the storytelling aspects of their game ahead of everything else. The cargo ship’s small size (relative to other dungeons) is a product of that. The designers could send the player through all the ducts of the cargo ship, fighting a variety of enemies along the way, but that doesn’t happen. That wouldn’t fit the fiction of the quest (i.e., the size of the ship), and that fiction is the most important part of the game. Thus, we get a smaller sample size of enemy attacks which affects the shape of the unaltered graph.

There are four other points on the graph which explain the philosophy behind enemy ability design in FF7. The first two I want to examine are the Great Glacier and Gaea’s Cliffs/Whirlwind Maze quests, which form the heart of phase three. Both quests are filled with enemies with strong attacks like the Headbomber’s ability Extreme Bomber, which has a 4.375 component stat, and the Lessaloploth’s Avalanche, which has a 3.75 component stat and targets all party members. These enemies appear often and in normal quantities, meaning that players have to make an important choice: do they spend MP to kill these enemies quickly, or do they spend MP to heal after taking several rounds of damage? One way or another, the player is going to have to spend MP. Unlike the Cargo Ship dungeon, the Great Glacier and Gaea’s Cliffs/Whirlwind Maze dungeons are quite long. The Great Glacier consists of 12 large maps and lots of generic connecting maps which make it difficult to navigate. It’s the one dungeon where the player might really get lost.

It’s easy for players to get lost in the many similar-looking sections of the glacier and for their party to become depleted of MP and items, no matter what strategy they use. Eventually, the party will faint and wake up at Holzoff’s house, but the quest can really drag on and endanger unprepared players. Based on the data collected about the design, it seems like the designers intended for the player to use big attacks rather than spend MP healing in this phase. Below I have visualized the changeover from using the strongest available multi-target summon (Titan) in the final four quests in phase two, and then the same thing (but for Bahamut) in the first four quests for phase three. In each quest, the damage is calculated using the average magic defense enemies have in that quest.


The difference is stark. Not only is Bahamut way more powerful than Titan, but the lower magic defenses of the enemies in the post-Bahamut quests makes the difference in damage even more pronounced. This trend continues throughout phase three. In the graph below, the blue line is the damage from the highest available summon, while the red line is average enemy HP for the phase three quests. In phase three, the player gains the ability to wipe out a battle with a single summon, whereas in most of phase two that only works in a few battles.


The player’s ability to blow through battles isn’t limited to just non-elemental summons like Bahamut, or summons at all; non-summon elemental spells are still totally in play. We saw earlier how elemental resistances drop across the course of the game, but I want to examine the elemental resistances in phase three, because they highlight the change in player spellcasting. To that end, I’m going to bring back the elemental resistances graph with another highlight.

There’s a slight uptick in the prevalence of resistances in the Great Glacier quest, but afterwards elemental resistances drop significantly. (The uptick in Great Glacier is mostly just in one element: nearly every enemy is strong against ice, but all the other elements are viable.) At first glance, this drop-off seems strange, but it’s probably deliberate. Designers might round up on the amount of EXP an enemy provides, and they might be careless with the occasional magic stat, but game-wide elemental resistances are not such a casual affair. I think that the reason that the designers chose to use a progressively smaller number of elemental resistances is that it allows the player to use a greater variety of attacks. What’s the point of putting all that time and energy into leveling up the Lightning and Leviathan Materia if they are progressively less useful as elemental resistances become more prevalent? This would make it impossible to blow through battles in one or two attacks, which is the essence of phase three.

The reduction in the prevalence of elemental resistances is actually a good example of how the materia system in FF7 separates it structurally from other entries in the series. In FF5 or FF6, it’s easy to teach a character several different kinds of elemental and non-elemental attacks. There’s also no limit to the size of a character’s spell book in those games; given enough time, every character can learn every spell. Except for having to invest time, there’s no drawback to learning a wide range of spells. In FF7, where an elemental spell fills one of the limited number of materia slots and each materia penalizes the player’s stats, it’s not really possible for every character to have access to every spell. Thus, it’s all too easy to go into a battle in FF7 without having the appropriate elemental spell available on more than one character. In that light, the late-game reduction in elemental resistances makes sense; players would be irritated to find that the 45,000 AP they have gained in their favorite materia is a wasted effort because so many enemies are immune.

Bosses in Phase Three

Bosses in phase three are less interesting from a mechanical perspective than they are from a statistical perspective. The only boss in phase three which does something really interesting mechanically is the Carry Armor, which shows us another evolution of the idea of imprisoned party members. All the other bosses in the phase are built around some version of statistical brute force. Still, sometimes the designers still manage to do something clever in that regard. A couple of bosses (including the Carry Armor, again) have multiple parts which put out significant damage. Multi-part bosses are not new in the third phase. Sample HO512 and Gi Nattak both had minions; they just didn’t do that much damage. By contrast, the multiple parts of Schizo and Carry Armor deal damage far in excess of what their base stats would suggest. There are also some quasi-bosses on the train bound for North Corel which subvert the player’s expectations of what to do in a phase-three battle just by being unusually durable. All of the bosses, however, tend to eschew debuffs in favor of throwing out large amounts of damage in a fairly straightforward fashion.

One of the odd things about bosses in phase three is that the component stats of their attacks aren’t higher than those of bosses in phase two. This doesn’t mean that these bosses can’t output more damage than their predecessors; they just have to do it in a different way. Below I have visualized the average of each boss’s component stat across the course of phases two and three.


For random enemies in phases two and three, I visualized the component stat of each monster’s most powerful attack, whereas in this graph, I visualized the average of a boss’s attacks. My reasoning for this is that boss battles last long enough for the boss to use all their abilities more or less equally (and their behavior scripts virtually guarantee that they will do so). Thus, the player’s experience of a boss’s damage output is, I think, closer to the average of its attacks than to its highest attack. In any case, I’m comparing bosses to other bosses here, so the difference between them and common enemies is mostly irrelevant. The graph can be deceptive because bosses—especially those in phase three—aren’t perfectly represented by their stats. Jenova DEATH, in particular, has bafflingly low component stats to her attacks. She’s a great example of what’s going on in boss design in phase three, however. Jenova DEATH can use her “Red Light” attack three times in a row—and frequently does. Non-boss enemies can use two attacks in a row, but they usually intersperse these double attacks with a single action. Jenova DEATH always attacks twice in a row when she uses Red Light, and frequently attacks three times in a row. Similarly, both of Schizo’s heads can attack in sequence, and all three parts of the Carry Armor can cause damage in one turn. So while these bosses aren’t equipped with especially high component stats, they can deal damage as though they were.

Phase three also features the closest thing this game has to mid- or mini-bosses, in the quest to retrieve the huge materia from North Corel. Each of the enemies on the materia train is unique to the quest and has an unusually high amount of HP, especially the Wolfmeister and Eagle Gun. Both of these enemies have HP on par with those in the monsters of the Northern Crater. On the other hand, neither monster does that much damage. The role of these enemies is to subvert the player’s normal phase-three strategy. Phase three calls for the player to use high-powered attacks to get through battles before the enemy can deal too much damage to the party. Many of these powerful attacks—especially summons—have long animations which will eat up the timer that is ticking down during this quest. Thus, players who have been relying on summons or other spells with long animations will actually hurt their own chances of completing the quest successfully and getting the full rewards. This inversion of expectation is one of the oldest design tropes in videogames, going back all the way to the early Mario games. Although this is one of the only instances of such an inversion in the game, it does show some evidence of the influence of action games on the RPG in its digital form.

Phase Four

As we saw in section one of this book, the kind of RPG that the Final Fantasy team wanted to make could not include the tactical elements which had been essential to the RPG formula from the very beginning. Final Fantasy 6 took the first step towards a class-agnostic system by using armor, stat and magic systems which gave no particular class an advantage over any other. Final Fantasy 7 consummated this trend by tactically denuding its characters, except for Aeris. The second-greatest mistake that critics make about FF7 is in thinking that nothing filled the void left by tactics. The complexity is delivered via the level-up system, which becomes more intricate in phase four. Essentially, the designers move the complexity out of the individual battle and into the long-term preparation for and recovery from that battle. To explain how this happened, I’m going to talk first about how the fourth phase of FF7 presents a new kind of difficulty, and then how the endgame content solves that difficulty.

In phase four, changes to the structure of the level-up system force the player to become more resourceful. The most obvious change is in the jumps in monster levels between dungeons. For most of the game, the player enters a quest either at the appropriate level or slightly above or below it. In phase four, the player’s party usually enters a quest at a level beneath that of the enemies. The chart below describes the median level (based on appearances in actual encounters rather than a bestiary list) of enemies in quests throughout the game.


To some degree, this is just supposed to make the dungeons harder, and it does. First, there is a small jump in the level of enemies in the Return to Midgar quest, and then there is a larger jump in enemy levels in the Northern Crater. Those jumps are the marker which makes phase four distinct.

It would be easy to see the bigger gaps in median enemy level as the worst kind of RPG difficulty. One of the things that free-to-play/freemium games revealed in RPG systems is the problem of boring stat inflation. Games which are trying to incentivize in-app purchases will use a variety of tactics to make their games almost unplayable so that the player will spend money. One strategy for this is to make each character level require a quadratically, or exponentially, higher amount of EXP points to reach. Final Fantasy 7 does not do this; in fact, the number of battles required to gain a level actually goes down markedly toward the end of the game. Of course, requiring the player to level up three times in a quest instead of once is similar to making the gap between levels wider. So FF7 could be said to achieve the same effect through different means, although I’ll explain why I don’t think that’s the case here. The short version is that although the gap in levels grows steeply, this is actually a kind of puzzle for which the game provides several pieces. Another strategy that F2P games frequently use to incentivize purchases is sudden leaps in enemy stats, regardless of level. We’ve already seen that this isn’t true for FF7. Enemies in the fourth phase have medium stats, medium debuffs, medium component stats, and low elemental resistances. It is through the sum of these statistical increases, plus the larger gaps in levels between quests/dungeons, that FF7’s last phase gets its difficulty.

As in the discussion of the first three phases, I want to highlight some of the finer details that don’t appear in large-scale statistical analysis. There are some unusual methods in enemy design in phase four, especially in the Northern Crater. (Although they are a part of phase four, I leave the discussion of the Gelnika and Ancient Forest for their own section.) The first new idea to enter enemy design is an increase in the prevalence of instantly lethal abilities. In the earlier analyses of the prevalence of debuffs, I said that I expected the worst debuffs like death, stop, paralyze and petrify to appear more frequently at the end of the game, and they do. Death spells, in particular, are much more common on enemies in the Northern Crater than in any other place. (Other debuffs make a moderate comeback, too.) But it doesn’t take a direct death spell for enemies in the Northern Crater to kill a party member. Several enemies in the final dungeon have other abilities which deal so much damage that they will kill either one or several party members. The best example of this is the Allemagne’s Level 3 Flare which has a damage modifier set to 120% of the target’s HP, meaning instant death if not mitigated by another effect. Similarly, the Dark Dragon can cast Ultima, the strongest spell in the game available to a non-boss monster, and the King Behemoth will cast Comet 2. In each case, these highly-damaging spells are only used as a counterattack against magical abilities the player uses. The Dark Dragon and the Allemagne will also only use their best attack once per battle, as do several other slightly less-threatening enemies like the Scissors.

With one exception, the attacks listed above aren’t really meant to kill an entire party at once, because FF7 just isn’t that kind of game. Only one ability in the entire dungeon is meant to cause a game over, and that belongs to the Dark Dragon. The Dark Dragon appears in the first section of the final dungeon, and its Ultima spell will probably wipe out parties below level 50, or those that aren’t using high magic-defense gear like the Zeidreich or Aegis Armlet or protective spells like Mbarrier or Big Guard. In that sense, the Dark Dragon serves as a great spot-check to tell the player whether his or her party is ready for the Northern Crater at all. And really, the best answer that a first-time player can receive is “no,” because there is a lot to do and see outside of the final dungeon. The other non-boss enemies tend to kill party members one at a time, or weaken the whole party by a moderate amount. This change reflects an evolution of the design ideas from phase three. It’s still in the player’s best interest to end battles quickly, because the enemies in phase four are deadly in many different ways. It’s just that in phase four, the player can’t plow through battles with summons and high-powered magic as he could in phase three. Not only are magic defenses back up to moderate levels, but many of the enemies will save their best spells for counterattacks against magic. Essentially, many enemies in the final dungeon have become like the first boss, but without the obvious warning of the raised tail.

The other way in which phase four evolves the primary dynamic of phase three is by deploying enemies and battles which cannot be rushed through by any means. Although the player continues to acquire powerful summons and other magic throughout the game, some enemies are virtually guaranteed to get off several attacks before dying. Sometimes this is through obvious means; the Dark Dragon, Iron Man, Dragon Zombie and especially Master Tonberry cannot be easily defeated in one or two turns because they have some combination of high HP, high magic defenses and elemental resistances. Even with attacks like 4x-Cut or some of the later limit breaks, these enemies will probably still get off between two and five attacks, many of which are highly damaging. Other enemies, like the Gargoyle, are guaranteed an attack by a combination of temporary invulnerability and counterattacking behaviors. The result of this is that players—especially those with under-levelled characters—have to both spend MP to get through battles quickly and spend MP to recover afterwards. With only one save point available in the entire final dungeon, there’s no reliably cheap way to restore those MP. Thus, the player is going to have to be careful about resource management in this dungeon in a way they haven’t ever been before.

Rationing and Attrition

More than half of what phase four has to offer is located outside the final dungeon, and because of that, I have given that content its own section. Before getting to that, however, I want to talk about how the final dungeon is supposed to help the player understand how the rest of phase four works. In FF7, as is the case in most of the FF titles up to this point, MP are not cheap. The purchasable item which restores MP in FF7 is the Ether, which grants 100 MP for the price of 1500 gil. This isn’t terribly expensive as far as items in FF7 go, but it’s not trivial. Battles in the Northern Crater drop an average of 2,600 gil. Or, to look at it another way, the party earns about 400 gil per basic physical attack, on average, in the Northern Crater. (Actually, both of these averages discount the Mover enemy, which dispenses tons of gil but appears so rarely that it doesn’t affect the overall rate much.) At that rate of return, the cost of Ethers will seriously cut into the player’s gil earnings, and that has real drawbacks which we’ll discuss below. The player has to figure out some other strategy for getting through the dungeon with a fresh party and a good stockpile of items for the three sequential last bosses.

There are several ways around the price of MP, but those methods (mostly) force the player to engage in activities other than grinding in the final dungeon. The most obvious way that the player can mitigate the cost of MP is by stealing MP from enemies. I wrote extensively about MP management and dungeon attrition in Reverse Design: Final Fantasy 6, and many readers made the totally justified criticism that everything I wrote was moot, because the Osmose spell was incredibly effective at stealing MP from virtually any target. I can offer no comprehensive defense against that criticism. I don’t know whether the part of the FF6 design team in charge of dungeons failed to communicate with the part of the team in charge of spell design, or whether it was a deliberate decision. Whatever they intended for FF6, the ability to steal MP in FF7 is significantly reduced. The MP Absorb Materia makes a nice pair with a number of other attacks, but it’s a unique item, it only absorbs 1% of damage dealt as MP, and it takes 100,000 AP to master it. If the player wants to spread the ability to steal MP around to more than one character, they’ll have to be clever about power-leveling the materia. As we’ll see in the next section, specifically targeting certain materia for powerleveling is one of the things that the designers wanted the player to do. But that doesn’t mean MP are cheap and abundant—especially not in long dungeons.

The player can also cheat. In every Reverse Design, I try to make a point of writing about the game that actually exists, rather than the game that the designers wanted to make. For FF6, this meant I had to acknowledge the legendary Vanish/Doom trick, and the fact that the evade stat doesn’t do anything. There’s a big difference between the Vanish/Doom exploit in FF6 and the W-Item duplication trick in FF7. When the player uses Vanish and Doom to instantly kill a boss in FF6, those spells are working exactly as intended. Obviously the designers did not foresee the power of this combination, but there weren’t any programming errors involved. Moreover, most of the toughest bosses are totally immune to this combination, so the game still presents a challenge. In FF7, the player can take advantage of a programming error to duplicate items (like megalixirs) with the W-Item Materia, thereby totally bypassing any concerns about gil or MP rationing. It would be wrong of me not to address the reality of this, but at the same time, I think the W-Item trick is philosophically different from the Vanish/Doom trick. The former is a bug, the latter is an emergent strategy. Beyond that, I know from firsthand experience that many players did not even obtain the W-Item Materia on their first few trips through FF7. Even though the W-Item trick obviates much of the analysis to come, I’m going to continue with it because it reflects the experience of those who played without exploiting that bug (deliberately or in ignorance), and those who continue to do so even to this day on platforms for which the bug was repaired.

The most important way to mitigate the struggle to ration MP in the final dungeon is to collect the numerous powerful weapons, items and materia which exist outside of it. In phase three, the player gains numerous new attacks, but virtually all of them are in the form of MP-expensive materia. In phase four, the player has the time and resources to obtain powerful command materia like 2x-Cut, Mime and Counter. There are also powerful weapons, many of which don’t require any dungeon crawling at all, like the Ultima Weapon, Venus Gospel, Death Penalty, Limited Moon and Final Heart. There is a summon materia so powerful that its high MP cost doesn’t even matter, because it ends virtually any battle in one or two casts. There are also materia and equips which enhance the rate of EXP and AP gain. Essentially, there’s an entire peripheral level-up system which exists in phase four which does not require the player’s party to be at a high character level. (Technically, some of these things can either be obtained at the end of phase three, or the process to obtain them can be started then, but for most of them it’s much easier to simply start in phase four.) It is by exploring these other options that the player will encounter the rich complexity which FF7 has to offer. But the peripheral level-up system—or what I like to call “wide levels”—is so deep that it requires its own section, which follows the discussion of phase four bosses.

Bosses in Phase Four

Battles with bosses in phase four become longer, but not always more difficult. Below is a graph of boss HP across the course of the game vs the growth of several other stats.


There is an explosion in boss HP in phase four. The important things to notice about this explosion is how sudden it is, and how it’s so much larger than the previous increase (which happened right before the beginning of phase three). The interesting thing here is that HP is the only stat which goes up so sharply. Magic defense, for example, doesn’t go up consistently across the course of the game, as you can see below.

Boss attack stats (not pictured) do go up, but they only do so in the same way that all attack stats go up—the growth is not explosive. Also, boss component stats don’t really inflate at all. None of Diamond Weapon, the Turks, the Proud Clod or Jenova SYNTHESIS regularly uses an attack with a component stat above 3. Diamond Weapon has a fraction attack which depletes party health to 1/8, but this is preceded by a long countdown and succeeded by a brief cooldown. Similarly, Jenova SYNTHESIS will cast Ultima after a long and obvious cooldown, giving the player lots of time to prepare for the blast. For the most part, the boss fights in phase four are just longer, not harder. The two exceptions to this are Hojo and Safer Sephiroth, but I’ll address their more powerful (and frequent) attacks individually after talking about the HP explosion.

Two factors account for the increase in boss HP: the caster’s advantage and the introduction of wide levels. The caster’s advantage is a design dynamic that arises from the scaling of spell effects when MP costs are fixed. That is, the Fire 2 spell gets stronger as the character casting it gains levels, stats and new equipment. The MP cost for Fire 2 remains at 22 forever. Thus, the damage output of a character’s MP pool grows like this for a single spell:


Because bosses are always preceded by save points, and because MP can be restored to full at a save point, players never have to ration their characters’ MP during a boss fight. The caster’s advantage makes that strategy even more effective as the game goes on. There is a drawback, however, in the need for healing spells. The player’s pool of HP grows along with the boss’s ability to do damage. That is, although the player characters are always growing into a higher maximum HP, the boss’s ability to reduce that HP grows commensurately. Thus, the caster’s advantage is not as advantageous when it comes to healing.


Although the amount of healing per MP spent grows, the need for healing grows so much that the caster does not enjoy as comfortable a margin as in the case of attack magic. Much of the MP spent during a boss fight will be on recovery, cutting even further into the marginal gains from the caster’s advantage on the damage side of the equation. But there is another factor driving up boss HP which supplements the gains from the caster’s advantage.

The other component driving up boss HP is the implementation of wide levels. There’s an entire section on this coming up, but I want to give an illustration of the idea. Sephiroth is the only boss in the game whose HP and stats scale up in a really noticeable way, as long as the player characters meet certain criteria. For every party member at level 99, Safer Sephiroth gains an additional 80,000 HP,[76] which is equal to his base HP. All of his character stats also climb markedly. In spite of this increase in stats, players who get every one of their party members to level 99 invariably obliterate Safer Sephiroth in a few turns. Along the way to level 99, the player has collected so many items and leveled up their materia so many times that it doesn’t matter if Sephiroth climbs to level 99 with them. When equipped with the best weapons, the best materia and using the best items, the player’s party doesn’t just climb to level 99—it’s more like they’re using characters who are effectively at level 150 or 200. That power, and the path to acquiring it, are the place where FF7’s design gets really interesting.

Hojo and Sephiroth both benefit from this explosion in HP, but they also have other factors which make them dangerous. Hojo uses debuffs often: he will routinely inflict poison, confuse, sleep and silence. His real strength, though, is that his basic physical attack Pile Banger has a 3.125 component stat, which is high for a basic attack. Hojo’s third form, meanwhile, actually combines the numerous debuffs with a strong physical attack. The third form’s Combo attack has a total component stat of 2.187, but it also inflicts poison, darkness and sleep on its target along with all that damage. For most bosses, the extra HP doesn’t do anything except make the fight longer. In the battle with Hojo, the extra length gives all those debuffs a chance to eat up lots of player turns. By contrast, the fight against the Proud Clod is long just for the sake of being dramatic, much like the Black Tyrano fight in Chrono Trigger. The Hojo fight is both long and dangerous.

Sephiroth, especially in his second form, does everything that Hojo does, but to a more extreme extent. Sephiroth’s first “Bizarro” form relies mostly on magic, including third-level elemental spells which have component stats of 4 to 4.375. His debuff attack, Stigma, has a 1.375 component stat (backed by a high physical attack stat) and applies poison and slow. He also uses multi-target fraction attacks, including a unique one that reduces all party members to 1 HP. The real cornerstone of the Bizarro form, however, is Sephiroth’s ability to heal himself. Although a properly-prepared party will have no problem dealing more damage than Sephiroth can heal, players whose parties are too low in level can easily become demoralized by his regeneration. This is especially true of players that have to consume too many turns healing their own party. Overall, though, only the fraction attacks are a real threat to knock out the player’s party.

Safer Sephiroth is considerably more deadly, as he possesses all the most dangerous types of attacks the game employs. Even his physical attack is dangerous; although the component stat for this attack is only 1.5, it inflicts paralyze and darkness. His weakest single-target magical attack has a 3.18 component stat and inflicts sadness, frog and small. His strongest magic attack (the legendary Supernova) hits all party members for 93% of their health, and inflicts confusion, silence and slow. He has two other magical attacks with component stats above 6. What separates Safer Sephiroth from all the other story bosses is that he has no attack that constitutes a lull. Jenova SYNTHESIS has a variety of powerful attacks, but there are points in the battle during which she’s counting down to Ultima, and in that moment the player has a little extra time to heal, apply buffs or deal extra damage. In the Safer Sephiroth fight, there’s no such time; Sephiroth seriously hurts at least one party member per turn. Thus, at least one character usually needs to be healing and cleansing debuffs all the time. Earlier I weighted the distribution of debuffs based on the number of character turns those debuffs would consume. Sephiroth’s constant use of debuffs in this battle certainly consumes some turns, but so does his constant damage output. A player who is spending turns removing debuffs and healing cannot attack. Thus, the length of the final battle is extended, not by a gratuitously large HP pool, as in the case of many phase four bosses, but by an endless series of deadly attacks.

Optional Bosses

The optional bosses Ultimate, Emerald and Ruby Weapon all follow the Sephiroth template to some degree. Ultimate Weapon is the easiest of the three, but it still employs one part of the formula that makes Sephiroth dangerous. All of Ultimate Weapon’s attacks are damaging enough that the player has to dedicate a party member to healing at least every other turn. Its weakest attack is a spell with a component stat of 1.5 that hits every party member. It doesn’t use many debuffs, but all of its attacks are above base damage, and most of them can hit more than one party member at a time.

Emerald Weapon is the best example of pure brute force in FF7. Obviously, it has incredibly high base stats—although we’ve seen many times how those aren’t the most meaningful indicator of difficulty. That said, Emerald Weapon’s component stats absolutely back up its statistical underpinnings. The monster’s most-used attack has a component stat of 5.6, while two of its regenerating minions have a primary attack with a component stat of 6.9.[77] Although Emerald Weapon does not use debuffs, nearly all of its abilities remove party member buffs like regen, mbarrier and haste.[78] Because of the huge amounts of damage that it and its minions can put out, players can easily lose turns to Emerald Weapon while trying to constantly heal and re-apply buffs—even more so than in the Sephiroth fight.

The closest thing to fighting a powered-up version of Sephiroth is fighting Ruby Weapon. Like Emerald Weapon, it deals tons of damage, but it also uses debuffs and has a nasty gimmick attack. Ruby Weapon’s most prominent attack has a component stat of 6.25[79], hits the entire party, and inflicts paralysis. It also has a spell with a component stat of 3 that inflicts confusion, and four attacks which cut a party member’s HP by a set fraction and inflict frog, small, poison and slow. The difficult thing about the Sephiroth fight is the lack of a lull; the player can easily get behind in the damage department while spending turns healing. If the loss of character turns is dangerous, however, then Ruby Weapon has the deadliest weapon in the game: he can remove up to two party members from the battle entirely. This is the final evolution of the design idea that began in the first fight with Reno. Reno imprisoned party members in a pyramid that could be cleared by physical attacks. In the Bottomswell fight, that prison attack became a bubble that dealt damage and was only vulnerable to magic. The Carry Armor is able to imprison two enemies until its arms are destroyed. But Ruby Weapon can knock two party members out of the battle entirely—an attack that the player cannot do anything to cure. Although the AI of Ruby Weapon can be exploited to avoid this ability, most players on their first trip through the game aren’t going to figure that out. This leaves the player with one party member who has to both heal and deal damage. That—in this author’s view—is the absolute peak of the challenges in FF7.

Quantifying the Game Script - Back | Next - Examining a Wide Level-Up System

Site questions?

All material copyright by The Game Design Forum 2016