Friday, August 31, 2012

Devil Survivor

Devil Survivor is another Shin Megami Tensei spinoff by Atlus. The game was never released in Europe and is a Nintendo DS game, so it took quite a while for me to actually muster up enough interest to play it. I have an irrational distrust towards any Nintendo system games, deal with it. Now that I've finished the game I'm pretty damn glad I did, and I'm going to order the sequel right away. Devil Survivor is a tactical RPG with all the power of Shin Megami Tensei packed into it. It doesn't even contain anything particularly unique or new, but the overall design is just brilliant. The recipe for success here is in fact really simple: it's a very faithful adaptation of SMT into a tactical RPG.

1. SMT lite?

Battles in the game run in a very typical isometric battlefield where units move and fight each other. The terrain is oversimplified: there are no modifiers from terrain, there's just obstacles and open ground. Fighting takes place in the streets of Tokyo so the lack of actual terrain does make sense. It's clearly not a game of map positions. Instead it's a game of unit composition. Each unit in the game is made up of one to three characters, most of which are demons. The player controls up to four humans who each can have two demon allies to form a unit with, whereas most opponents will be teams of demons only. It's worth bearing in mind that these are SMT demons which makes them inherently more multidimensional than your typical standard tactical RPG units. I'll go into the demons in the next section.

During a unit's activation, every member of that unit is allowed one action. In addition, the unit can move and initiate a battle. The way the game relies heavily on correct unit compositions, the leader of each unit can use his or her action to summon a new demon, provided another one uses their ability to withdraw first, or has died in combat earlier. Other possible actions mostly involve healing and certain tactical level abilities possessed by demons. Moving and attacking both affect a unit's turn order. If they do only one of those things (or neither) their next turn will come up sooner so it is not always desirable to fully utilize a turn. The reason the game is so big on unit composition is the combat system, which is basically SMT lite. I think a similar approach was taken in Bahamut Lagoon.

Each battle is a small game in itself with its own screen as the combatant's fight a full round of SMT style combat. There is nothing astonishing about SMT combat: it's just a turn-based system. The player sets a command for each of their combatants, and actions are then carried out in a slightly varied order based on agility. Some predicting ability is needed to figure out a good approximation of the order actions are carried out in. One UI slight makes this a little more annoying than it needed to be: there's no way to view agility scores of combatants while in the combat screen even though it is possible to do so on the main battle screen. Otherwise it works like a charm, mostly because it relies on a tried-and-true mechanical core of turn-based combat.

The SMT twist comes from extra turns that can be scored but also stolen from other combatants. The conditions are quite familiar for people who have played Lucifer's Call, Persona 3/4 or Digital Devil Saga: hitting a weakness or scoring a critical hit guarantees a bonus turn and often steals one from the opponent. Likewise hitting someone's immunity can give the defender a bonus turn. These bonus turns are taken in another round of combat that is fought immediately after the normal round. Only characters who scored a bonus turn participate in this round. This system is simple yet clever because it puts a lot more emphasis on weaknesses and immunities. It's not just more damage, it's also another turn in which to do even more damage.

This is especially aggravating with AoE spells: if there is even a single target that is weak to the spell, then the caster gets another go which can be used to cast the spell again. In this case the attack does more damage to everyone just because one of them was weak against it. See why team composition is important?

2. Demons all around

The reason team composition forms a big part of the game is the way that the game handles demons. Best demons are always obtained through fusion because they inherit abilities from their parents. Devil Survivor has streamlined the inheritance process quite a bit. Instead of 8 general slots where abilities are inherited randomly, demons have 3 active slots and 3 passive slots, and the player can freely choose which abilities to inherit into each slot. The only limitation is that the demon's natural skills cannot be replaced. Inherited skills can greatly enhance a demon's power by giving it a larger range of attacks and even covering its weaknesses with rarer abilities that provide immunity. The entire process of demon raising is about passing along the most important skills and obtaining new skills from demons on the way.

However just mindlessly passing good skills along won't do, because the other half of good demon raising is to pay mind to their natural attributes, their weaknesses and resistances and immunities. If a physically strong demon inherits powerful spells, it often doesn't utilize them very well. Devil Survivor includes another, even more important aspect into this equation: racial abilities. It's a long standing tradition in SMT that demons are categorized into different races which are used to form the rules of fusion. However, in Devil Survivor each race has an important tactical ability that cannot be changed. The fusion process therefore has another goal: not only is it important to match skills to to demon attributes, you often want to have demons from a specific race because of their ability.

The cleverness of the demon fusion system is well-proven by its ubiquity in the series and its spinoffs. As far as character development schemes go, it is special because of the huge amount of variables that go into the system, and the complex dynamics that dictate fusion results. Note that Devil Survivor has actually simplified the process quite a bit, particularly inheritance. It also includes a nifty search system that makes it easier to figure out how to create a demon of a given race. These are necessary amends because the number of demons that are needed in battles is eight at minimum and often more to adapt party compositions to different situations. They have done a marvelous job with this simplification: the system feels as intriguing as more complex systems from other SMT titles.

The reason this all works so well is that while they have made inheritance simpler (free choice instead of random selection with complex rules)  they have at the same time limited its power by putting more emphasis on the demons' natural abilities. This is especially achieved by introducing racial abilities which are tactically more important than any combat abilities. There's also an interesting balancing mechanism: not all racial abilities are equal usefulness, but often demons who have the better racial abilities are either harder to make, especially of desirable components, or generally weaker.

3. Structuring for pacing

If there ever has been one glaring flaw in SMT titles, its pacing. Modern titles have it better: Persona 4 and Devil Summoner 2 have really good pacing. The problem is that the games often have really intriguing plots but they become hard to follow and go into because between every event that moves the plot forward there's hours of running around in dungeons. As a tactical RPG, Devil Survivor is already naturally structured better. Even if the game only contained battle events with some dialogue before and after, it would still have a lot more going on because each battle takes maybe an hour at most. It's not just that though. The game is structured by time limits. The player can often choose from multiple places to visit, and each visit is either a dialogue that deepens one plot line in the game, or a battle.

The nice thing is that this system incorporates choices quite naturally. Each event is only available for a limited time - sometimes even just one single hour in the course of the game - and each event takes the game clock forward 30 minutes. There is never enough time to go to all events so the player has to make choices. These choices mostly affect how other characters end up in the game. A lot of them can die as a consequence of the player's choices. Overall the structure makes the game go smoothly and have meaningful decisions all the way through. Selecting events is a simple matter of choosing where to spend time from a menu so it's also time efficient on the UI side.


I was swept away by Devil Survivor. It should have been predictable, I mean, I love SMT, and I love tactical RPGs. Still I was surprised how smoothly these concepts came together in the game. It's full of elegant design choices. I might actually go as far as to say that it's the best tactical RPG I've encountered so far. This is largely due to the complexity of the system when it comes to units. Raising and selecting demons is a delicate business. Moreover, they've had the guts to make the game hard enough to actually encourage the player to explore their possibilities and pour some serious thought into character development. The icing on the cake is an intriguing plot with a lot of characters that are meaningfully incorporated into the story. That, and the fact that the plot really feels like it's moving forward constantly.

Friday, August 17, 2012

Notable RPG Bosses vol 2

Since I'm still playing Shin Megami Tensei: Devil Survivor I don't want to write about it just yet. Instead, we're going to continue our boss series. This time around you're in for a real treat. This particular boss is pretty notorious and indeed does test the limits of fairness. It is however also a really brilliant design. Here goes:

Demi-Fiend (Digital Devil Saga)

The Demi-Fiend is better known as the protagonist of Shin Megami Tensei: Lucifer's Call (Nocturne if you're on the other side of the Atlantic) but makes a cameo as an optional boss in the second playthrough of Digital Devil Saga. Half-human, half-demon, he is able to summon and control demons and also possesses a nasty array of skills. The Demi-Fiend has six different demons which he will summon in order whenever he has less than two demons out. This is the list of demons he uses: Cu Chulain, Girimehkala, Pixie, Arahabaki, Titania and Parvati. He starts the battle with the first two already in the field. All in all, this is an "endless allies" type boss battle which are often difficult due to the high number of turns the opposing side is able to take. This is particularly true in the SMT's press turn system where you get a bonus turn for hitting an opponent's weakness or getting a critical hit. He does the latter a lot.

If you've played this game or taken a look at its skill list, you'll notice that there are skills that allow you to become immune to pretty much everything, so how can this fight be any challenge? Well, that's because this guy doesn't really approve of unfair fighting and rewards your "cunning" with an almighty spell, Gaea Rage,  (almighty, as the name suggests, being the only damage type that cannot be cancelled) that does more damage than it is possible to have hit points in this game. He does generously allow you to have resist skills though, which will be needed to cover any weaknesses of the characters you bring because otherwise his allies will be creating way too many bonus turns exploiting your weaknesses. He also allows you to have two null skills because they don't actually work like the rest of them. I find it just a little baffling that they didn't give them more discerning names.

Although he doesn't approve of your unfair fighting, he has absolutely no trouble fighting unfairly himself. He's immune to most damage types with three exceptions, two of which are actually useful against him. He also gets a second turn to start off with. His attack patterns are actually pretty limited, he's just throwing out various physical attacks that deal a lot of hurt to all characters. The most annoying of his attacks is one that causes mute which, if it sticks, makes you unable to cast spells. Another attack causes stun. Curiously this is a blessing more than anything else because if a character is stunned they cannot be muted. Since by this time you'll have access to a healing spell that cures full hit points and all status effects, mute is usually only a nuisance but occasionally can kill your entire turn.

Of course his allies add their own ingredients to this mix of nasty. Stuff like instant kill attacks, petrification and tons of debuffs can be expected. As long as you're prepared, you should usually survive all this. I'm saying usually because sometimes an unlucky streak of attacks can really throw you off. Such is life in the Junkyard. Anyway, so far so simple, so what's the big deal? Remember Gaea Rage, the attack that kills your entire party off if you dare to have any immunities on you? Well, he actually throws it out at specific times during the battle even if you are fighting according to his rules of fairness. This will happen exactly after he has summoned Pixie and after he has summoned Parvati. There is only one way to avoid getting wiped out by Gaea Rage, and that is a skill called Null Sleep.

This may sound a bit weird but it's mostly because of poor naming. What actually happens is that after Pixie and Parvati are summoned, they will cast Dormina, a spell that causes sleep. How's this important if you are equipped with a skill called Null Sleep? Well, that's because the skill doesn't make you immune to sleep. Instead, when afflicted by sleep a character with Null Sleep is guaranteed to dodge every single attack thrown at them. So that's good news, Gaea Rage won't hit you. Bad news is, there's no guarantee that sleep will stick. This is where the fight is borderline unfair, because there's not much you can do besides bringing Cielo who is weak against all ailments, it's all based on luck. The only thing you can do is to minimize the number of times you have to roll this particularly deadly die. Easy right, don't kill his demons, he'll never get to even Pixie.

Right. That's where the other nasty mechanic of this battle comes to play. Whenever a demon ally has taken 30 turns using anything except Dekaja or Dekunda (these skills clear enemy buffs and ally debuffs, respectively) they will cast Recarmdra which kills the demon but heals the other demon and Demi-Fiend back to full HP. Which is typically worse than having to start all over again, so at this point retrying is a good idea. This forces you to kill those demons. You just don't want to do it too quickly, otherwise you're pushing your luck with Gaea Rage. The odds of sleep sticking are by the way uncomfortably low, so you really don't want to do that. On top of that, even if it does stick and you do survive Gaea Rage, the aftermath is messy. There's no telling how long your character will sleep and on whose turn they wake up.


So, this boss sounds mighty unfair, no? I mean having something depend on luck is bad design right? Although that's what it initially looks like, the design can be defended. There is one way to have single-use failsafes against Gaea Rage which will somewhat mitigate the effect of luck. Most importantly though, having it depend on luck is what gives this battle its urgent nature without introducing hard time limits. They could have easily switched to some lose condition that works similarly, like after a total of ten demons are killed, Gaea Rage is imminent. However, selecting that threshold would have been a tough call. Too lenient would make the battle too easy; too strict would have made it even harder.

Actually, having a hard limit might have actually made the battle even more luck-dependent in a paradoxical way because then it would matter a lot more how many turns are lost to unlucky Javelin Rains (the attack that causes mute) and attack combinations. Although these need to occur more often to screw up your battle plan, they also can occur on about every single turn in the fight. But, because there is no hard limit, it is always possible that the next Gaea Rage won't wipe you out and you get more time. It is still infuriating to lose the battle to an unlucky Dormina. There is however also the thrill when everything is hanging on just (figuratively) one roll of a die.

Unfair or no, the two mechanics that make this battle unique, that is Gaea Rage and Recarmdra, are brilliant because they implement a two-sided time pressure. Defeat the demons too soon and you will be getting hit with more Gaea Rages. Defeat them too slowly and you risk redoing the battle. Everything else in the fight mostly serves one purpose: it forces you to think about your skill choices very carefully. There are only eight slots, and there's 5 skills that you'd want to set for every character to maximize survival. After this you're still going to need skills that heal your party and deal damage to the demons and Demi-Fiend. The high damage output also means that maximizing your characters' stats is a requirement for this battle. This consequently means that there is no way to over-level to make this battle easy.

Although the appendix is actually longer than the post itself, I suggest skimming through it just to get an idea how tightly planned a strategy needed to be to clear this fight. It also shows that the randomness of Gaea Rage is in fact a lot less random once you really prepare for it. Controlling the chances in random scenarios by the way is a really enticing game mechanic. That's actually explained by the flow theory's notion of the paradox of control. I quote: "Only when a doubtful outcome is at stake, and one is able to influence that outcome, can a person really know whether she is in control."

Appendix: How I defeated it

First of all, I'm afraid the details might be a bit hazy. It's been quite a few years since I did this battle and although my recall is pretty good, this probably won't be as detailed as the Gilgamesh entry. I actually checked my saved games to see if I still had my skill configuration in one of them but no dice. Overall it took me about half a year to clear this one. Not all of it was spent playing of course, and I did have a four month break from the game after deeming this battle just too damn hard. At some point though I had to get back to it. This time around I gave up my pride and looked up some strategies (well, two) online but, and this is the second time I recall this happening, they didn't really work for me. Both required too much luck to my taste so ultimately I ended up developing my own. I'd be the first to admit those two strategies did affect my own quite a bit of course.

For starters I had to do some more grinding to get all the relevant stats to 99. That's magic and vitality in this battle for all characters and strength for one character. Maxing these stats is important because they reduce damage. Magic also increases magic damage which is by far superior to physical damage in this battle. Cielo had to be in the party because he's more likely to go to sleep than anyone else. Serph, being the leader that you're free to develop, is an automatic choice and finally I chose Argilla because a) I like her and b) she has the highest magic in the game and the least points in useless stats for this fight which means less grinding. Because both Serph and Argilla have an elemental weakness they are going to need to cover that with a resist skill. Everyone needed to have Death Resist, Phys Resist, Null Sleep and Null Critical (which does only decrease the chances for critical hits, huh?)

After trying this battle a lot of times I figured that one really important aspect is to be in good control when exactly the Demi-Fiend's allies die. More specifically, I wanted them to die at the same time. The best way to do this is to have an attack that does a lot of damage to all targets because the higher the damage of a single attack, the less you need to worry about the difference in remaining HP the two demons have. The most damaging attack in the entire game is Black Sun. It does very high and very constant damage (also important) but it's a combo spell that uses up everyone's turn. It also means forgoing the most powerful area of effect almighty magical attack (Celestial Ray) in the game for a slightly weaker one because the weaker one (Megidolaon) is required for the combo to work. The other two required skills, Ragnarok and Salvation, are fortunately necessary for this fight anyway.

With AoE taken care of using Ragnarok, Megidolaon and Black Sun, there's still need to cover healing, single target damage and debuff recovery. I figured out that two characters with Salvation is actually sufficient for this battle if they are the first two characters to act each turn since passing doesn't use up an entire action but two passes in row does. So to account for healing, normally one of the two characters would be able to do it by spending a turn. If the first one is muted, he can pass and the second character can use the passed bonus turn to heal resulting in only one lost turn. If both are muted, there is no way to really avoid spending two turns even if the third character has Salvation. The first character can pass, the second can use a Dis-Mute to cure the first character, the third character can then pass and the first can finally cast Salvation, leaving one turn.

So that left me with one extra slot available for the last character (Argilla) which allowed me to boost her damage potential a lot by adding (I think, not sure) Mind Charge. To cover single target damage I think I gave everyone Last Word which is the only single target almighty magic attack in the game. Finally I gave exactly one character, the middle one (Cielo) Dekunda to clear those debuffs. I left Dekaja out because it can be cast from an item, which I simply harvested lots of. I don't actually remember what I used Cielo's last skill slot for (I'm not even sure if I gave him Last Word or not) but I'm pretty sure it was something important. Could have actually been Rakukaja which increases my defense, and more importantly, forces one of the demons to cast Dekaja (further decreasing damage).

The last piece of the puzzle to reduce my chances against Gaea Rage was Close Call, a skill that allows a character to survive death once per battle. The problem with this skill is that it leaves you in human form* with 1 HP and it's insanely hard to recover from this situation. Besides, there was no space for it, seriously. However, there were two perfectly good candidates for having this skill: the two characters that were in reserve (you can switch active characters during battle). So I set them up with the skill and something else that didn't really matter. Phew, that's quite a lot of setting up but hey, this is the single most epic boss encounter in any game I've played.

Armed with this setup, a calculator, a pen and one sheet of paper I was ready to go. Wait, what? Yeah, because of the 30 turn Recarmdra rule it is highly recommended to keep careful track of demon actions, so I used tally marks. The calculator was for counting Demi-Fiend's hit points, particularly when he was about to enter half HP because that's when one of the demons will cast Mediarahan, healing the entire group back to full HP. This only happens once, but it needs to be timed quite carefully. It also allowed me to know when he was one Black Sun away from dying, which was also very helpful to know. I kept track of demon damage in my head because Black Sun allowed for quite large errors. All this tallying and calculating is why the fight takes a lot of time (even a failed attempt often takes 30 minutes!)

In the fight itself, I would first drop both demons to low enough HP that I could take them out in one turn (don't remember whether I pummeled them to Megidolaon or Black Sun range), and then pound the Demi-Fiend with Last Words until the demons had taken around 25 actions. I didn't want to get too close to 30 because a double mute could disrupt my ability to wipe them out in time. For the first pair of demons I also avoided using Ragnarok because it causes Rakunda along with damage, which in turn causes Cu Chulain to cast Dekunda and leaves Girimehkala to cast Mamudoon (instant death fun all around). Normally most of Girimehkala's turns go towards casting Dekaja after Cu Chulain's Taunt because Taunt debuffs and buffs the player party.

Wiping both out at once was particularly important when preparing for Gaea Rage. When out of demons, Demi-Fiend gets only two turns. One is spent to summon a demon and the second turn goes to the summoned demon, in this case Pixie, who will then cast Dormina. Since it is now my turn I can react to whether anyone fell asleep or not. I got shitty luck and the first two Dorminas in the battle didn't stick so I had to prepare one of my reserve Close Call characters to take the Gaea Rage. I think the procedure was like this: Ragnarok, switch one character out, Rakukaja. This way, when it became the Demi-Fiend's turn and they had 3 actions, one would go to summoning the second demon, and then the demons would spend their turns casting Dekunda and Dekaja. That allowed my measly human form 1 HP character to keep me in the fight.

My tactic also involved a curious healing pattern. If my characters were inflicted with stun I would actually avoid using Salvation and use full recovery items to heal instead to keep the stun. Although stun is slightly inconvenient, being inflicted by it makes characters immune to mute because of the one ailment rule in the game. If I remember correctly it was particularly useful when preparing to cast a Black Sun because that can be easily disrupted by a single mute. All in all, finding the windows to cast Black Sun was hard, but in the end I think it was the piece that was missing from my earlier strategy. In order to safely cast Black Sun all characters need to be in very good condition and preferably free of debuffs.

The fight was over in an hour and a half. The last 30 minutes when I realized I could win the fight were pretty intense. I screwed up at one point because I lost count of some demon actions and had to take them out "just in case" so I actually had to endure one more Gaea Rage that I had hoped for, but I got lucky (finally!) and Dormina did stick to one character who was able to recover my fighting potential which allowed me to bring victory home. The sight of the Demi-Fiend vanishing was probably the biggest rush I've gotten from a game. I think it was actually my first attempt with the Black Sun strategy. It did take a lot of failed attempts to arrive at this strategy though. I haven't tested this strategy again, and I won't because of the 1.5 hours it takes, but I felt that it was pretty stable - I only had to rely on luck for that one Gaea Rage because of my own screw-up, the other two were dealt with the failsafes.

I'm sad about one things though: I've lost the paper that had the tally marks from the victorious battle.

* Characters usually fight in demon form and normally start fights in them as well. Human-form is a lot weaker, and most importantly you can't use any skills. They have two advantages: natural immunity to light and access to the gun damage type. The reason why it's so bad getting stuck in human form is that it takes one turn to transform and you have to do it with that character. Furthermore, if you die in human form, you are resurrected in human form. Switching is often preferable because you can switch out dead team members for living ones using another character's switch command which allows the new arrival to often act on the same turn. 

Friday, August 10, 2012

Notable RPG Bosses vol 1.

This blog is drifting more and more towards RPG design. Post-game RPG bosses have been a particular interest of mine for a long time. I guess ever since I fought and defeated the two Weapons in Final Fantasy VII. This series of posts introduces some of my favorite bosses from games I have played. In order to get into this list, a boss needs to be exceptionally challenging while remaining fun to fight and plan strategies against. The ultimate point of these writeups is to collect implications on how to design better boss fights, especially considering post-game and hard difficulties. Let's start with a very recent experience that inspired this entire series.

Gilgamesh (Final Fantasy XIII-2) 

Gilgamesh is a classic adversary in the Final Fantasy series. He was first introduced in FFV as a recurring boss and has since appeared in several games in some role. His latest appearance is in FFXIII-2 as a downloadable post-game boss and he is the toughest enemy in the entire game. He is portrayed as a sword collector and his trademark in battle is the use of multiple swords from fake Excaliburs such as Excalipoor to the real thing and other legendary weapons such Masamune. In FFXIII-2 he starts the battle with firearms instead in what can be called a mock battle before the real thing. After going down pretty quickly he discards the firearms and the real fight begins.

As is quite typical for very hard boss fights, the battle is divided into two sections. I'm actually not such a big fan of dividing lengthy battles into sections because usually dying happens only in the later sections and a lot of time on attempts is wasted simply by getting there. It does serve a purpose on first attempts as it allows the player to learn some of the core mechanics in the fight but after that it's just a chore. This is how it is with Gilgamesh as well. It does allow you to benchmark your strategy against his main gimmick. But yeah, in the beginning he is not that dangerous. He hits with attacks from random sword that are not particularly dangerous and sometimes goes into defense mode. This is a good time to stop attacking because he will launch a rather powerful counter attack with damage based on how much he was attacked.

As boss fights in FFXIII and also -2 tend to do, this one also revolves around the use of stagger. This is also where his main gimmick enters play. I guess a quick reminder of how stagger works is in order. Battles feature a chain gauge which has two purposes: it acts as a damage multiplier and also as a stagger meter. Most enemies in the game have a stagger point, after which they enter a more vulnerable state where their chain gauge can be driven up all the way to 999.9% after which you can deal massive damage. However stagger only lasts for a limited time and after that the enemy goes back to 100.0% chain gauge.You can also lose the chain gauge while you are building stagger if you don't attack often enough. To make it a little more complex, certain attacks (mostly, those by commandos) kind of strengthen the chain but raise it only minimally while others raise it more but make it weaker. The length of the stagger depends in the chain's strength.

Gilgamesh has a ton of hit points so you will definitely want to stagger him to make the fight go in any reasonable amount of time. Staggering him also has other benefits: he loses his weapons while staggered, lowering the threat he poses temporarily. However, staggering him also has a downside: after recovering from stagger, he heals himself for a big chunk of HP. This means that whenever you stagger him, you must be able to deal significantly more damage than this chunk, otherwise you won't progress anywhere in the fight. This requires two things from the player: optimizing party damage output and selecting optimal paradigms. Paradigms are role configurations for the party. You can have only six paradigms, and you can only use three monsters in them (which means your third party member can only be chosen among three roles instead of six). To maximize damage output, you would need to include a commando, a ravager and a synergist.

So far so simple. Using these three guarantees the best damage output. However, once you get Gilgamesh below half of his total HP, the battle gets a lot crazier. Not only does he get Haste which makes him attack more often, he also gets longer attack chains and a bunch of new attacks that are a lot more devastating. To have any chance of survival, you need to have a monster sentinel out at almost all times which seriously reduces your ability to raise the chain gauge and to deal damage. You won't really have time to cast buffs at all because you'll find yourself juggling between building the chain, strengthening it and healing yourself. On top of everything else, Gilgamesh gets a really powerful attack that is not telegraphed (unlike every other mega attack in the game) which in addition to doing a load of damage puts one character into sleep. If you don't have your sentinel out when this attack happens, your lead character is guaranteed to die and it is your monster that goes to sleep (that is, your sentinel!)

Getting a stagger in is now harder. He does still lose his weapons in stagger but gets to keep his Haste. This is important because throughout the battle he can make an attack that causes Fog and Pain statuses which disable a character's magic and physical abilities respectively and this forces you to spend valuable stagger time to heal them. Since he's on Haste for the latter half of the battle, the chances of getting hit by one of these go up tremendously. You are still supposed to do massive damage on each stagger because of his healing ability.


The reason I like this battle is that it (along with some other DLC fights) gives the game the challenge it originally lacks. Since choices in FFXIII-2 are always limited: three monsters, six paradigms, 100 accessory equipment points (with most great accessories costing over 50 points, that's a big limitation), the player is forced to make compromises to deal with all aspects of this battle. It's also a necessity to develop some really kickass monsters, especially in terms of damage output. Resistance to Pain and Fog need to be factored in, because getting hit by these constantly is a severe hindrance. But the element I like the most is the time pressure during staggers. Your party that is geared towards survival still has to be able to dish out a lot of damage on every single stagger. All in all, the fight takes a lot of planning and subsequent development.

It also makes great use of the game's battle system, forcing the player to utilize every aspect of it in order to optimize damage. Especially in the latter half, situations change rapidly and the player needs to swap paradigms frequently and keep an eye on Gilgamesh's chain gauge to avoid losing it and having to start building it up from scratch again. All the planning and battle performance is due to multidimensional pressure. Gilgamesh has all sorts of nasty tricks that will force the player to adapt their strategy to take all those into account. It is important to notice that the huge amount of damage he deals is there to hinder the player from focusing into dishing out tons of damage and the main piece of the puzzle is the healing ability. As we will see in future parts of this series, having multiple sources of pressure is the key to interesting boss fights.

Appendix: How I defeated it

My strategy was far from optimal, since I took an hour to defeat it. The target time is 24 minutes. I'm aware of at least one better strategy that involves using different monsters, but I wanted to try my own first. This section is not particularly explain-y, I'll be using game terms with abandon, explanations in footnotes. I was using the following paradigms:

COM-COM-COM (dish out the hurt during staggers)
RAV-RAV-RAV (build chain gauge rapidly)
COM-COM-SEN (keep the chain gauge alive while shielded)
SYN-SYN-SEN (cast defensive buffs while Gilgamesh is defending, was quite useless)
MED-MED-SEN (for healing obviously)
SAB-SEN-SEN (for taking the biggest blows and casting debuffs at the beginning of stagger)

I used Lightning as my commando for her high damage output and rather lengthy feral link* (to make use of chain gauge locking**). I chose Valfodr as my ravager because he's from the same pack as the other two which gave me a damage boost from Pack Mentality for Lightning and Valfodr. He also has exceptionally high damage for a ravager. Finally I used Snow as my sentinel. He has the highest HP in the game and should have no trouble shielding Serah and Noel. He also has a highly damaging feral link attack which could be used to boost damage during staggers. I equipped Serah with a fog resistance accessory and Noel with a pain resistance accessory.

I was controlling Serah for this fight. The start of the fight was spent mostly in the first two paradigms, building stagger and charging feral links. To maximize damage output during stagger I used the following process: immediately upon stagger I switched to the last paradigm to quickly cast Deprotect and Deshell. They stick on Gilgamesh on the first casting during stagger so then I immediately switched to RAV-RAV-RAV and ordered Serah to attack and then activated Valfodr's feral link so that Serah would reach full ATB while the gauge was frozen to attack again and push the chain gauge to or near 999.9%. Then I switched to COM-COM-COM and pounded the boss for two full ATBs, using items to heal pain from Lightning or Noel if necessary. A quick switch to COM-COM-SEN and, utilizing ATB cancel***, ordered Serah to attack and immediately activated Snow's feral link (again to allow Serah to reach full ATB while the gauge was locked)

Again, immediately after Snow's feral link animation ended I switched back to COM-COM-COM for some final pounding. For Serah's last attacks I switched from magic to physical attacks and timed it to hit when Scourge would be activated (magic attacks can't use Scourge) and finally activated Lightning's feral link to allow Noel to get in more attacks with Scourge. Especially during latter half of the battle I sometimes had to visit more defensive paradigms briefly to stay alive but this process was mostly repeated as it is presented here for every stagger.

During latter half of the fight I mainly stayed in COM-COM-SEN to keep the chain gauge alive and to do some damage. Every two full ATBs I switched to RAV-RAV-RAV for one quick burst of chaining (using ATB cancel) unless I suspected a Divider might be coming (that's the non-telegraphed mega attack). I also took his telegraphed mega attack in SAB-SEN-SEN to reduce its damage. After Divider I typically had to go to MED-MED-SEN for a while to heal (I dropped the gauge a few times because of this actually). Whenever Gilgamesh went into defensive mode, I switched to SYN-SYN-SEN to cast some defensive buffs, particularly Veil which reduces the chances of getting hit by status effects. Although Gilgamesh can dispel these, Serah usually got to keep her buffs because she was nowhere near Snow (who took practically all the hits in this battle).

After an hour of constant juggling between paradigms he was down. It didn't feel like such a long fight because it demanded constant attention and although my stagger process sounds a bit repetitive there was a bunch of timing issues to constantly stay aware of. Of course I didn't take him down on my first attempt, no, it took several just to figure out this strategy and then a couple more to perfect implementation.

* Feral links are unique special attacks monsters have and work like limits etc. in older games of the series. They can be used at intervals and have varying effects (most do damage though). 
** Enemy chain gauge is frozen during feral link animations but rest of the world moves normally, allowing more attacks during a stagger and also more hits with Scourge, an attack that activates when the enemy has very little stagger time left and deals a lot more damage than normal attacks. 
*** Whenever you have used to full ATB gauges before switching paradigms, you'll start the new paradigm with full ATB gauge instead of empty. This accelerates damage output (among other things) a lot. 

Friday, August 3, 2012

Persona 2: Innocent Sin

I've been a fan of Atlus since I played Lucifer's Call (that's Nocturne for the US crowd). Since then I've played all of their RPGs I have gotten my hands on to and I've never experienced much of a disappointment. Well, the first Devil Summoner aside, but even then they corrected their mistakes in the sequel. Atlus has a tendency to do two things really well: deep game mechanics and intriguing plot. Storytelling on the other hand was a clear weakness before Persona 3. So um anyway, enough with the context. Although I knew from the start that Persona 2 would be in no way comparable to its successors I think that a lot has gone wrong with this game.

Actually, the only thing the game really has going for it is its plot which is primarily hindered by the f-ing long and numerous dungeons in the game that have random battles bursting out at approximately five second intervals. Talk about old school. These are pretty common errors and not worth a blog post really, so let's take a look at some of the more specific screw-ups in Persona 2.

1. Demon contacting gone horribly wrong

This has been a stable in the Shin Megami Tensei brand for quite some time. Especially in the main series. Normally contacting is done to obtain new demons for the player's party. The core mechanic is quite simple: the player chooses a communication skill to use and the target demon reacts to that. The reactions are predetermined and often predictable because they are based on a fixed set of personality traits. All this makes contacting less of a random selection and more of an actual decision. The mechanic is overall just fine and in the main series it actually demands decisions from the player because recruitment skills take space from battle skills.

Persona 2 turns the occasional demon negotiation into a grind. Instead of immediately gaining new allies, the player now has to collect tarot cards from demons which can then be spent to summon personas in the Velvet Room (a blue room with creepy people in case you're wondering). This system effectively means that to obtain new personas you'll be contacting the same demon over and over again to get that bucketload of cards you need. What grand fun. Before that though, you'll spend some time figuring out what the f all those contacting skills actually mean regarding demon personalities. The protagonist for example has the amazingly useful skill of imitating the sound of construction equipment and can also initiate a discussion about manliness with the demon. Based on my experimentation, if a demon is "wise" and "forceful" it will be intrigued by a  good talk about being a man. Mind = blown.

Although this contacting business is quite the grind there is actually a silver lining. Once you get a demon to become eager, the battle immediately ends after you get your cards. This typically takes a fraction of the time it would have taken to actually kill the demons so it's a good way to skip lots and lots of battles. I found myself doing this a lot in the longer dungeons.

2. Let's drop this demon fusion business...

Hi. What? In case you're unaware, one of the biggest hooks in the entire Shin Megami Tensei brand is the ability to raise demons (or personas) and then fuse them together to get more powerful demons while inheriting the more useful abilities from the "parents". The beauty of this mechanic is that every raised demon serves a purpose and often the lineage of end game demons is amazingly long. The fusion mechanics have all sorts of interesting rules and properties which people have written very long guides about. So what's the genius move in Persona 2? Oh they dropped demon fusion altogether (actually they might've done this in P1 already, I haven't played it).

Basically they kept everything else which makes about as much sense as keeping the stuffing and throwing away the bread from a sandwich (on a second thought, the analogy makes more sense). Because personas are leveled and higher level personas learn more powerful skills, the player needs to upgrade their persona arsenal after pretty much every dungeon. The player also cannot summon personas that are 5 levels higher than their characters. A newly summoned persona starts with a grand total of one ability and another can be added by throwing a skill card into the summoning process. Personas open new abilities by gaining ranks which involves using them in battle. At least they had the decency to show which skills a persona will learn.

Since you need experience and levels to summon new personas, and you also need to use personas to make them actually useful, you're faced with a choice in each battle: grind for cards or grind for levels/ranks. That means double the grind, yay! Not to worry, the game gives you a lot of time to do both because the number of battles is overwhelming. All in all, this system is just incredibly stupid. Every persona you raise is essentially worth nothing in the long term. Sure, you get a small gift for releasing a maxed out persona, but they are very rarely worth the trouble. Second, a new persona starts out really crappy but you are forced to use it if you ever want to make it better. Completely and constantly replacing something the player has spent a lot of effort into developing is a fucked up mechanic.

3. Fusion spells

This is the only "new" mechanic in the game that actually makes sense. Well, almost, and it comes with a cost. If certain spells are cast in succession, their effects are replaced by a fusion spell, which is typically more potent than the individual spells would have been. To make this possible, commands to characters are given in advance and then executed by hitting "run". The player then needs to decide whether to start a fusion spell or cancel it. Starting a fusion spell cancels the individual spells as characters wait for the final participant to cast the final component and then the fusion spell itself is unleashed. Figuring out the most effective persona configurations and battle order to effectively utilize fusion spells is the best part of this game. Sadly the combat UI is a bit lacking and makes this often rather tedious.

UI problems are nothing new though, they are pretty common in old RPGs. What's amazingly stupid about the fusion spell system is that the player has to discover the recipes by trying out combinations of spells. Oh my god, what were they thinking? I have never ever been a fan of "try everything" puzzles or systems. Including such a system in the game's already tedious combat mechanic is just plain wrong. Hidden information that is in no way hinted is an annoying trope in Japanese RPGs and I don't really find it excusable ever, but it gets a several magnitudes nastier when applied to the one core mechanic that makes the game's battles go in decent time. Fortunately for us modern gamers fusion spell lists are nowadays available on the web.


The theme in Persona 2 game design seems to be "everything is a grind". I didn't find the game particularly enjoying to play. It was playable though, partly because for an Atlus game it was ridiculously easy. I played on hard difficulty and I got a grand total of one game over. I would usually complain about this, but with all the screw-ups made in the game's design I was just happy to get through it in good pace. I'm not certain if this is just a quirk with the PSP re-release though. Despite all this thrashing, I'm still going to play Persona 2: Eternal Punishment when it comes out in Europe. It uses the same system, so at least I will be familiar with all the quirks. I'm also dying to see how the story develops. The quirky humor is also a bonus.