The Game Design Forum

Finding Complexity in Wide Levels

A large portion of the content in phase four exists outside of the mandatory dungeons. The purpose of this content is to allow the player to build his or her partyís power through means other than repeating the same few battles in the final dungeon over and over again. Earlier I talked about the concept of ďwide levels.Ē In order to explain what this term means, itís necessary to provide some context to how we think about level-based progression systems in RPGs. Since the days of D&D, a level-up has always been a permanent, periodic increase in character power. The most obvious example of this is when a character goes from being level 23 to level 24. Thatís a permanent condition; it happens at regular intervals as measured by experience, and itís an increase in power. Several other events fit the same criteria, however. An obvious example of this is the levels gained by materia, which go from casting simple spells to more powerful spells, or which allow the player to cast a summon spell multiple times per battle as they grow in level. Equipment is also a form of level-up. The growth from level 23 to level 24 is qualitatively the same as the growth from the Titan Bangle to the Mithril Bangle. Gear may not seem as permanent as character levels because players often sell itóbut they only sell their old gear when they have access to a new and better piece of gear. Players canít ďsellĒ level 23 when they reach 24, but they would if they could. Sold or not, gear is just another level-up system that offers permanent, periodic increases in player power.

If levels can take many different forms, we can assess level-up systems in terms of height and width. The height of a level up system is easy to measure; itís simply the number of increments between the playerís starting level and the maximum. For FF7, the character level system goes from 4 to 99 (no character starts below level 4), for a height of 96 increments. Thatís fairly tall, although many JRPGs have passed it since. There are lots of other heights to measure, though. Cloud has 16 weapons to collect, each one a level-up. Every character has seven limit breaks to learn, each one a level-up. But if weíre looking for width, weíre looking to see how many different systems contribute permanently to character power. My count is five different systems: character levels, limit break levels, materia levels, equipment levels and chocobo levels. Itís not immediately obvious how chocobo levels affect character power, but weíll get to that later. The important thing to realize is that there are many different systems which strengthen the playerís characters.

Thereís more to the design of a wide level-up system than merely having different bars to fill. In order for width to actually work as a design philosophy, the player has to engage in different kinds of activities to fill those bars, or else the player is only grinding. Most of the above systems are based on battles, but not all of them. The best materia and gear are both acquired through various kinds of exploration. Although players certainly can become powerful enough to beat Sephiroth and the optional bosses through mere grinding, it actually takes less time (and fewer repetitions) to pursue non-battle tasks in conjunction with some grinding, rather than grinding until the character level cap.

The point of wide levels is to reintroduce complexity that was stripped from battles when character classes disappeared. The really brilliant thing about FF7ís level-up system is not that itís just wide, but that all of its diverse systems interact with one another in different ways, giving clever players a way of manipulating all of these systems to their partyís advantage. All of the different level-up systems run on different currencies, and these currencies come at different rates. Whatís more, those rates change over time. The primary example of this dynamic is the change in the ratio of EXP to both Gil and AP.



One explanation for this dynamic is that the charactersí EXP needs grow more than their AP needs grow, and the total amount of EXP needed is vastly larger. But this doesnít explain how the player is supposed to grow materia like Ultima, Contain, Bahamut Zero and Knights of the Round. All of these materia arrive late in the game, start at zero AP growth and have huge AP requirementsówithout a commensurate increase in AP income. The designers arenít simply making a mistake here, however; the game is changing to give the player new ways to play the game.

Before players have even thought about the major sidequests, they are met with a puzzle that comes out of the phase four level system. The difference in currency rates has had two important results already: the player has a Gil surplus and an AP deficit. The ratio of gil to EXP has actually gone down, but because the player no longer has to purchase gear, they will start to accumulate it in the endgame phase. An accumulation of money at the end of a JRPG is a fairly common occurrence, and different games have dealt with the surplus differently. Star Ocean: The Second Story features a shop in the middle of its optional dungeon whose contents are so expensive (although they are worth the money) that players can never expect to buy them all. Final Fantasy 8 employed an item transmutation system that allowed players to turn certain common, purchasable items into more powerful items or even directly into magic, thereby putting excess money to use. Final Fantasy 7 takes advantage of the surplus too by allowing the player to invest that money into chocobo breeding. Neither Star Ocean nor FF8 has anything like the AP deficit that exists in FF7, however. (Star Oceanís ability points system is folded into its level-ups, while FF8ís ability points are actually much easier to farm than EXP is, if the player knows where to do it.) The ideal solution to the surplus/deficit problem would be to turn gil into EXPóand by complicated means, this is actually possible.

EXP Alchemy

The signature cycle which drives the end-game content in FF7 is the transformation of excess gil into EXP and AP. The means of this transformation is through the Gold Saucer, primarily via chocobo racing. Because the player no longer needs to buy gear, he or she can invest their excess gil into something other than equipment purchases. One strategy would be to simply invest it all in purchasing Ethers, which would allow for the replenishment of MP while in a dungeon. This strategy will work, but itís not efficient and requires nonstop grinding. Although the upfront costs for chocobo breeding are high and the payoff is not immediately visible, itís still the best investment in the game.

For players who are not duplicating items, the cost of chocobo greens, stalls and nuts can accumulate quickly. The rewards, although not immediate, are extremely good. To the right is a table of items which can be won from races. The items here are a mixed bag. Some of them are useless trash, and some of them are useful but not worth doing races for. Some of them, however, are incredibly valuable. Materia like Enemy Away and Counter Attack have clear uses, but itís the elixir item which holds the most value. Elixirs can be converted into large amounts of EXP (and decent amounts of AP) through the Magic Pot monster. The Magic Pot can be found in the Northern Crater (its exact position and encounter rate are discussed later.) After being doused with an Elixir, the Magic Pot will become vulnerable and give the player 4000 EXP per hitóor more than ten times the normal per-attack rate. The Magic Pot is also one of the best sources of AP in the game, dishing out 1000 AP, about four to five times the normal rate.


Chocobo racing is a great way to earn extra elixirs, since they cannot be bought anywhere. In fact, at the S-Rank level (the highest rank of race) the player is twice as likely to get an Elixir as any other reward. It does seem that the designers made this part of the game with this exact conversion in mind.

One of the great things about the chocobo racing mechanics is that thereís never any wasted effort, because the player never has to accept an item they donít want. All of those undesirable items can actually be converted into the useful currency, GP, which can be spent at most places in the Gold Sauceróincluding at a unique shop in the Wonder Square.

The amount of GP scales depending on the quality of the item, but the purpose of this mechanic is to make every victory useful. GP buys one of the best items in the game, the EXP Plus Materia, which will reduce the amount of grinding the player has to do by directly augmenting earned EXP. There are other ways of acquiring GP to buy this item, like the minigames of the Wonder Square. Like chocobo racing, these games require gil, but they take a lot longer to produce GP than chocobo racing. In the end, all of these paths are ways of converting gil into EXP, either through Magic Pots, augmented EXP income, or both.

For exceptionally skilled and/or knowledgeable players, there are other methods for stockpiling Elixirs than chocobo racing. Players who know where to look can use the Morph command to turn several enemies into Elixirs (although this is grinding, which is what theyíre probably trying to escape). Players who are preternaturally good at the basketball shooting minigame in the Gold Saucer can rack up GP fairly quickly, too, to buy the EXP Plus Materia. Neither of those things can net the player the unique materia which is strewn about the world map in hidden caves; that is only available through chocobos who can cross terrain which is inaccessible by airship. Those materia are some of the best available in the game, including Mime and Knights of the Round. This is the trick of the chocobo breeding sidequest: itís always giving the player more than one kind of reward. Racing gives the player three rewards: items, GP and access to higher-level chocobos. Those rewards can be either directly converted into EXP and AP, or else they augment the rates at which EXP and gil come in, giving the player virtually everything they need to face the gameís highest challenges.

The last thing to understand about chocobo breeding is that the way it converts one currency to another is cyclical, and that the cycle accelerates. The starting point of the cycle is a surplus of gil which can be spent in the initial costs (housing) or ongoing costs (feeding) of chocobos.

Ideally, this cycle will result in the player gaining character levels (and gil) progressively faster and faster. Not only does the EXP Plus Materia directly enhance the rate of acquired experience, but the supply of elixirs and the Gil Plus Materia will further enhance the income of EXP and gil, which keeps the cycle going. Players that understand how to use the cycle can gain character levels from 70-99 faster than they gained levels from 40-70. This is exciting for its own sake; FF7 has established a set rate at which players tend to gain levels (of all kinds), and now the player is taking advantage of the gameís systems to gain power faster than before. Itís also exciting because the optional bosses give shape to the end-game. All that leveling has a pointóthereís a goal for the player to strive towards. Players who use the chocobo cycle to gain levels faster will be excited to see that their goal (of being powerful enough to defeat Ruby and Emerald Weapon) comes more quickly when they cleverly take advantage of the gameís systems.

Other Leveling Tools

Through the chocobo cycle gives the player extra AP through the elixir/Magic Pot conversion, the player is not able to directly augment AP income through that cycleís products. There are other means for accomplishing this, however. Numerous weapons in the game have double AP growth for the materia that are equipped to them, and two weapons have triple growth. The triple-growth weapons are of particular interest here because theyíre both found rather than bought. Cidís Scimitar is in a fairly obvious treasure box in the submarine base, but Cloudís Apocalypse weapon is located in the optional dungeon, the Ancient Forest. Thus, both of these items are already part of a wide level-up system, in that they are products of diligent exploration. But thereís also some extra complexity built into their use. The power of these two weapons isnít nearly high as the gameís best equips, so the rate of finishing battles by normal attacks will drop slightly. Itís not a huge drop-off, but it does require the player to concentrate on using abilities other than basic attacks. The other drawback to using the triple-growth weapons is that both of them have a small number of materia slots; the Apocalypse has three and the Scimitar has two. This smaller number of slots can deter players who arenít thinking critically about their end-game goals. As we saw above, however, critical thinking is what the designers are trying to reward. The point of the triple-growth weapons is to power up those late-arriving materia like Contain, Mime, EXP Plus and Knights of the Round. In that function, the triple-growth slots serve quite well, but the player has to consciously make a decision to level a few specific materia. Although JRPGs donít always have a reputation for meaningful decisions, and mid-90s Square RPGs in particular are seen as being too simple, this is an example of an implicit tactical choice. The choice has moved from the battle to the preparation for the battle, but itís still there and it definitely matters.

The last part of the level-up system in phase four I want to address is the usefulness of the two optional dungeons, the Gelnika and the Ancient Forest. In all of the graphs about the difficulty of dungeons, I deliberately left these two out. The reason for this omission is that the structure of the optional dungeons is so different from the structure of the story dungeons. All of the gameís mandatory dungeons have to be played from start to finish, and first-time players donít know how long each one will take. Thus, players have to be cautious and conserve resources like items and MP. The two optional dungeons donít operate under the same dynamic. The Gelnika is a dead end, and itís only four rooms large.

The player is never more than two rooms away from the save point at which they can cheaply restore the entire partyís HP and MP and save. Thereís no reason to conserve resources here; tents restore MP too cheaply to necessitate conservative behavior. The Ancient Forest is a little different; it does have an endpoint, but itís also quite short.

After getting the few items this short dungeon has to offer, the Ancient Forestís primary use is for grinding. This can be done in the first screen, so as to maintain access to the world map and the use of tents for cheap restoration.

Because of the structure of the dungeons and the configuration of the enemies in them, both the Ancient Forest and the Gelnika enemies offer decent experience with only moderate risk and time expenditure. The best experience in the game is located in the Northern Crater, specifically in the ďswampsĒ section.

Thatís fairly deep into the crater, however, and it makes for an inconvenient trip. Moreover, the enemies in the Northern Crater are the most dangerous in the game. Of the two optional dungeons, the Gelnika is the more dangerous, but no battle in that dungeon is likely to wipe the party out instantly. No enemy in the dungeon can cast Ultima, Death or Level 3 Flare, but there are some fairly powerful attacks. The three variants of the ďUnknownĒ enemy all have attacks with component stats of three which they tend to use as their most common attack. The Unknown 2 routinely inflicts paralysis and confuse. The Unknown 3ís most common attacks (with component stats of 3) inflicts poison and fury. The Serpent enemy casts Aqualung, which hits the whole party with a component stat of 3.25. Because of these attacks, the Gelnika battles are not easy, but the player doesnít have to worry about conserving MP to heal over the long term because of the easily accessible save point. If the player isnít ready for the Gelnika when they reach phase four, the Ancient Forest works as an easier version of the Gelnika. Enemies in the Ancient Forest are weaker than those in the Gelnika (both in terms of HP and attacks), but provide only about 60% of the EXP and AP that enemies in the Gelnika do. There are two enemies in the Ancient Forest which can use attacks with component stats of 1.25 or higher which cause darkness or poison. This lower level of difficulty and lower level of experience points is better-suited to characters below level 50 that need to grind out some EXP quickly. That said, with the right abilities, players can clear battles in the Ancient Forest in one turn because the enemy HP is lower. So while the amount of EXP and gil per battle is lower than in the Gelnika, the amount of EXP and gil per turn or per minute can be higher with the right character set-up.

Why is FF7 so Back-Heavy?

The chocobo breeding/racing quest brings back the complexity lost in battle tactics and party composition by giving the player a complicated feedback loop to manipulate in the last phase of the game. Although most of the game is straightforward and requires only a low level of critical thinking, the optional content gives players a lot more to think about if they want to beat the optional bosses without having to invest 20 to 40 extra hours grinding. Accelerating the level-up system is interesting on its own, but the FF7 team also decided to do it through something other than battles. It would have been easy for them to simply have sent the player hopping between dungeons where different currencies like EXP, AP and gil were plentiful in different proportions, but they didnít. Instead, they substituted the process of breeding and racing chocobos for those tasks. Eventually, races can start to feel as repetitive as battles, but by the time the player is getting bored, he or she should be well on the way to the goal of a black or gold chocobo, and should have racked up plenty of items and GP along the way.

This still leaves the question: why does the chocobo quest have to come at the end of FF7? The deeper question is: why does all the complexity of FF7 have to come at the end? The answer is the same as it has been in every other aspect of the game. The FF7 teamís first priorities are telling a meaningful story, crafting compelling characters and building a persuasive world. Battles, levels, items and the tactical complexity those things can create are secondary to storytelling. Once the story is almost over and the only thing left to do is defeat Sephiroth, the designers can move complex systems to the fore without impacting their primary goal. Obviously this is not the only way to construct games; plenty of RPGs have extremely demanding combat and other challenges right in the middle of their plot. The Dark Souls series has a reputation for robust lore and extremely demanding battles, to give one example. And yet I think itís telling that when people talk about FF7, they talk about their favorite characters or their favorite scenes, whereas when players talk about Dark Souls, they talk about their favorite character builds, bosses and dungeons. Intentional design choices cause this difference.

Understanding Items as Levels

In the previous section I described every permanent increase in power, including the acquisition of equipment and materia, as a form of level-up. These level-ups are not as frequent or as uniform as character levels, but thatís what makes them interesting. This section attempts to compare all of the meaningful levelups in the battle system to character levels, even if that comparison is somewhat imperfect. The general results of this exercise show that comparing weapons to character levels is somewhat meaningful, especially in terms of ultimate weapons. Comparing armor to character levels, however, is not especially meaningful. Comparing materia to character levels, meanwhile, reveals a great deal about the nature of spells in FF7. The materia comparison is the most flawed in the sense that when characters gain power through materia, they often (but not always!) have MP costs which complicate the analysis. Nevertheless, there are several ways in which the comparison between character levels and materia shows us something about the game design which would otherwise be obscure. The stat which I like to use to compare one form of level-up to another is called level equivalence, or LEQ for short. The purpose of the stat is to determine how many character levels an item is worth. The comparison works best on weapons. To calculate the LEQ of a weapon, simply find the point in the game at which a character switches from one weapon to another and determine the raw base damage (component stat of 1, no defense calculation) of each of those weapons. Then, calculate how many levels the character would have to gain in order to equal the base damage of the new weapon while still equipping the old weapon. Below I have visualized the LEQ for all of Red XIIIís weapons.


For most of the game, each weapon is worth about four levels, although the number increases slightly as the game goes on up to just about six levels. Itís not a huge increase, but it does warrant the spending of gil. The only outlier is the ultimate weapon, which is worth more than 50 levels, which would mean that Red XIII would have to be higher than level 99 (which is impossible) to achieve that damage without the weapon. The point is clear, however: ultimate weapons in FF7 are ludicrously powerful. Although I didnít visualize them, the same thing is true for the weapons of Cloud, Cid or Cait Sith. Each weapon increase is only worth a few levels, except for the ultimate weapon, which is more powerful and valuable than any other item in the game.

Itís not practically possible to perform LEQ calculations for armor because there is no equivalent of the raw or average damage which are used to calculate the LEQ of weapons. Enemies vary too much in the kind of damage they put out, and in what increments. Materia, on the other hand, are actually easier to perform LEQ calculations for than weapons. The very simple method for this is to look at the component stats of various materia-based abilities and compare them to a hypothetical spell which does base (1x modifier) magic damage. The only complication is that the point at which a player gains access to new weapons is fixedótreasures and vendors come along during certain quests, and thereís a fairly clear suggested level for each quest (based on the levels of the enemies in the relevant dungeon). There is no clear indicator in the game that says at what point a player will learn Ice 2, Ice 3 or Flare, however. Therefore for my LEQ calculation, Iím going to use a level 30 Red XIII as the casting character. Itís an arbitrary choice, but it is one that is consistently applied.

Spell LEQ

Although it does not take into account efficiency, we can calculate the LEQ of each level of elemental spell in FF7, and thereby understand how many character levels a materia level is worth. In order to do as much raw damage with Fire 1 as he does with Fire 2, Red XIII would need to gain 99 levels (i.e. the LEQ is, coincidentally, 99). This is impossible, since the cap is 99 and heís already level 30, but extrapolating the statistical trends for the character shows that it would take a level 129 Red XIII to make up the difference in power between Fire and Fire 2. The LEQ of Fire 3 vs Fire 2 is 80, meaning that Red XIII would have to be level 110 to make up the difference in raw damage between the two spells. In other words, the loss of MP efficiency is completely justified given the amount of power of each level of spell. So each level gained by a basic elemental materia is worth upwards of 70 character levels in damage (if you donít take MP into account).

The tactical significance of spell choice gets deeper when spells target more than one enemy at a time. For normal magic (green materia) spells augmented by the All Materia, the damage formula is simple: raw spell damage is cut in half. Certain materia do not incur this penalty, however. Most of the unaffected spells are summons, but a few are enemy skills. For example, what is the difference between casting Fire 2/All, Ifrit, and Beta? All of these are multi-target fire spells, but there are differences in both LEQ and MP spent.


When cast by a level 30 Red XIII, Fire 2/Allís raw damage is about 339 to each enemy. At the same level, Ifrit does about 764 raw damage to all targets (because it isnít affected by split damage mechanics), for an LEQ of 57. Beta, the enemy skill used by the Midgar Zolom, is also not affected by the split damage mechanic to which Fire/2 all is subject. It does about 1528 raw magic damage, for a surprisingly large LEQ of 214 vs the Fire 2 spell. The Fire 2 spell is the cheapest in terms of MP, but itís only 13 MP cheaper than Beta, despite its much lower power when cast on multiple targets. Against a multi-target Fire 3, Beta still has an LEQ of 28 (i.e., in order to beat the damage of Beta at level 30, Red XIII would need to be level 58 when casting Fire 3/All), despite costing 20 fewer MP, making it a vastly superior choice for clearing multiple targets.

Why design Beta and Ifrit to be better than the Fire spells of similar level? The goal is discovery (i.e. wide levels) rather than choice. A great deal of RPG design comes down to making the right choice at the right time; this is a trait that RPGs inherited from their wargame ancestors. Unfortunately, the choice between spells like Fire 3 and Beta isnít a very meaningful one. Fire 3 is a little better than Beta against single targets, and Beta is more efficient than Fire 3 against multiple targets. Itís important for players to notice this and choose accordingly, but itís not a particularly deep and meaningful decision. Instead of awarding power to players who make the best decisions, the designers are awarding power to the players who explore the game the most thoroughly. The Fire Materia is acquired through any of several materia shops. The Ifrit summon is obtained through beating a mandatory boss, but it is unique. The Beta ability can only be obtained by surviving the attack when it is cast by one of the few creatures that uses it in random battles. Each one of these abilities is harder to obtain than the last, and so it makes sense that they would have varying levels of power. Thatís the whole point of a wide level-up system.

Independent Materia and LEQ

Independent materia, which include such gems as Magic Plus or HP Plus, are the easiest kind of equip to measure by LEQ. The low effort required to measure them reflects the low effort that went into balancing them. The strange inequality of these materia shows up pretty clearly in any comparative graph. Below is a graph of the LEQ of three different kinds of independent materia affecting three different character stats on a level 30 Red XIII.


In the graph, each of these materia is at master level, granting a 50% bonus to a particular stat. Because the growth rate for each of these stats is so different, a 50% boost can have wildly different results in terms of equivalent levels gained. For example, a level 30 Red XIII will have an average luck stat of about 18 (stats can vary a little bit in different playthroughs). A level 99 Red XIII has a luck stat of about 24. Therefore, a 50% boost to luck at level 30 is worth more than 60 levels. On the other hand, HP grows a lot per level, so a 50% increase is only equivalent to gaining approximately eight levels worth of HP. A growth of 50% means something very different for each stat. The designers did seem to spot the discrepancy between luck and HP to some degree; the Luck Plus Materia requires twice as much AP to master as the HP Plus Materia.

Where the designers failed to notice a difference is in the Magic Plus Materia. Mastering the Magic Plus Materia requires the same amount of AP as the HP Plus Materia, but it allows Red XIII to cast magic as though he were 23 levels higher. If the Magic Plus Materia were hard to find, this would be another example of the enhanced power of wide levels, but this is not the case.

Although slightly out of the way, the Magic Plus Materia is visible from the main path through an obligatory dungeon. Itís certainly within the realm of possibility that the designers didnít care that this materia was more powerful than others in its group. But itís equally possible that they applied a template to it without checking the math. In this case, LEQ shows where there is still a nail sticking up out of the floorboards, so to speak.

Using LEQ in Other Games

The point of the Reverse Design series is to make available to its readers the design tools that were used in the making of classic games. On its face, this is an impossible task, because the ďtoolsĒ that the FF7 team used were probably the subjective judgments of people like Hironobu Sakaguchi, Yoshinori Kitase, Hiromichi Tanaka and Yasushi Matsumura. In lieu of having those people on the staff making your game, I have tried to create tools which faithfully recapture their creative output, even if these tools cannot replicate their creative judgment. The LEQ stat is one such tool, although it has its limitations. The LEQ value works best for one-to-one comparisons, like one weapon or spell against another of its kind. But itís still useful for broader analysis of a game. Using LEQ shows us that exploration was mathematically more important to the designers of FF7 than grinding. The ultimate weapons, none of which require grinding to obtain, are usually worth more than 50 levels. The most useful enemy skills, which are worth 20 to 100+ levels, require no grinding to obtain; they require exploration. The most useful independent materia are easy to find and a little overpowered. There is no ďcorrectĒ LEQ for any given weapon or spell, but LEQ can tell you when items are statistical outliers which might destroy the game balance youíve worked so hard on. This tool doesnít turn us into Sakaguchi and Kitase, but it does help us to measure our work against theirs in a meaningful, precise way.

The Four Phases of FF7 - Back | Next - Enemy Archetypes

Site questions?

All material copyright by The Game Design Forum 2016