Remembering the Encounter: Souls, Advance Wars, and JRPGs

11 min readAug 12, 2019

This week, I played through Xenoblade Chronicles X (the 1.5th game in the series). A few hours after putting down the controller for good, I thought back to all the unique encounters and boss fights I had seen in the game — when I realized that I couldn’t remember a single one of them.

I wrote this article to answer this question: why can I remember Souls bosses from two years ago, but not the bosses from the game I completed today? What makes their design so different? Here, I’ll primarily be concerned with the question of enemy design in encounters in a few different game architectures: action RPG, strategy, and JRPG. Of course, encounters also implicate music, story beets, audiovisual design, and an array of other things — which is why the Manon ship is the most memorable part of Xenoblade, all things considered — but the focus here is more strictly on the gameplay aspects.

Souls: the Art of the Tell

For a particular enemy design to be memorable, it must be unique relative to the rest of the game’s content. For it to be unique, the player must interact with it in a unique way.

Broadly speaking, there are two ways for a player to interact with an enemy: proactively or reactively. Proctive interaction takes its clearest form in attacking, but we can also consider sneaking through enemies’ ranks or running away from them as proactive interaction. Reactive interaction primarily involves defensive measures: dodging, rolling, blocking, parrying. The difference lies in whether the focus is on the player’s actions, or the enemy’s actions.

As a series, Souls is tightly focused on reactive interaction. In combat, you have little choice in how to attack an enemy. You have two attack buttons, and usually only one of them is useful. While you have the option of using different weapons, you generally lock into one choice before combat starts. And “combos” don’t really exist. The most significant choice in proactive interaction I remember from Souls is the Twin Demons fight, where you can choose to attack the demons’ heads for a chance at a stagger, at the cost of repositioning. Contrast this with Nioh, which has three stances you’re expected to hot-swap during fights, along with stance-specific customizable combos and magic/gun attacks that you can incorporate into swordplay. Joseph Anderson has made this exact comparison in his Mario video (timestamped).

In Souls, the player’s reactive interaction with enemies is controlled by enemy attack patterns. This means that the primary means of designing unique enemies in Souls is creating differing attack patterns.

As forecasted in the title of this section, Souls’ main method of designing attack patterns is the “tell”. This refers to the way an enemy indicates what they’re going to do before they do it. They might raise their sword, or start screaming, or pull out a gun; based on the overt display, the player chooses a specific defensive move to counter the coming action; then the enemy follows with a sword strike, or a massive explosion, or a few gunshots.

The key is that the relationship between tell and action should be unique. While you can make unique tells by giving enemies out-of-place weapons like pistols in Sekiro, in a game about duels between Pepos in Armor, there are only so many ways to indicate that an enemy is going to swing their weapon. Rather, differentiation occurs by leading the same tell into different attack patterns. For example, consider the tell “raise weapon over head” in Sekiro. When a certain surprise boss does this, they slowly fall onto you, and if you parry the blow, they collapse onto the ground for a while. When the final boss does this, they pause for a second before doing two lightning-speed strikes; you should circle-strafe instead of parrying. When the hidden memory boss does this, they will normally do a single over-the-head blow, but will instead switch into a circular strike if you try to strafe too early. Tells can lead into incredibly specific combos, like Genichiro’s Flying Duck Strike, which require incredibly specific player responses. In every case, each enemy does something unique with the same pool of tells.

Because you first defend against the enemy’s attack before attacking, tells and reactive play are not avoidable: players must observe and learn them in order to succeed. In doing so, players (are forced to) get a feel for the unique flavor of each enemy.

Strategy: Proactive and Reactive Interaction Together

In Advance Wars, your options for how to attack or defend are basically infinite. Encounters are unique because they restrict your options to a unique subset of playstyles effective against the idiosyncrasies of the enemy and map. Your enemy loves fielding expensive planes: you should build fighter jets and anti-air units. The map has a narrow road surrounded by an open ocean: you might take control of the waters and then quickly deploy battleships to overwhelm your enemies. The victory conditions are “escape to the west”: so you hit and flee.

Strategy is, of course, both proactive and reactive: every move you make is both a threat against the enemy’s position and a defense of your own position. Since strategy is so immediately dictated by the characteristics of each encounter (especially the map layout), it’s particularly straightforward for developers to give each map unique effective playstyles.

“Every encounter in this game is the same: you deploy Giga Tanks and overrun your enemy on land.” — not me

While strategy games can push you towards a set of effective playstyles, they can also forcibly restrict your choices. On some maps, you can’t construct Giga Tanks. On other maps, your enemy can construct Giga Tanks, but you still can’t (playing at structural disadvantages like this is incredibly engaging). On the map above, you can construct Giga Tanks, but they’re useless. While land units exist on every map to some extent, many maps don’t have naval or aerial battle. Every unique restriction yields a unique playstyle and a unique experience for the player.

Not Advance Wars

Yet not all strategy games adequately explore this space of playstyles, and the result is combat that feels samey. Here, I’ll indict Disgaea, which can be described as a turn-based JRPG on a grid map. In Disgaea, too much of the space of playstyles is open at any given time, and the result is that you settle into a rut where you approach every encounter in the same way.

How does Disgaea fail to restrict player options? First and foremost, it is a character-based, rather than unit-based, strategy game. Instead of deploying Infantry or Mechs, you deploy individual characters. While there are on the order of 20 character classes, this only really affects the JRPG-style “skills” they can use; there are more like three actual playstyles (melee, ranged, support — but the difference between melee and ranged is negligible). Characters all move about the same range and do similar types of damage and have similar stat distributions; there are no pointed advantage tables as in Advance Wars. Without character differentiation, designers cannot force the player to use different characters in different scenarios. Even if maps are differentiated, player response is not: it is always a matter of sending the strongest units along the straightest path, because that’s always the best strategy.

Valkyria Chronicles is another character-based strategy game — each character has unique “traits” which modify its stats under certain conditions — but hedges the dangers of this by grouping characters into a few classes, plus a tank, which do have pointed advantages and disadvantages. Snipers have almost no mobility and will get insta-killed by almost every enemy, but can shoot enemies from really far away. Recon units have incredible mobility, but stronger ground units can take them out quickly. And your tank costs twice as many action points to use. Disgaea has no such differentiation — all your characters have skills, that do damage, in slightly different AoE ranges.

Second, enemies are not differentiated. Since player units and non-boss enemy units function the same way and are pulled from the same pool (you can recruit non-boss enemy units), this is a direct consequence of the lack of character differentiation. All but two bosses are also recruitable in the same fashion, and as a result they are also not differentiated. Bosses are overstatted, but there is nothing unique about facing an overpowered enemy.

There is one encounter in Disgaea which is unique: Baal (one of the two non-recruitable bosses). While every other map requires you to kill all enemies or kill the boss (who is behind all the enemies), Baal introduces some new mechanics that force you to approach it a bit differently. You win by destroying four crystals — the catch is that destroying a crystal immediately kills the character who did so. But this kind of mechanical complexity is the exception. Overall, Disgaea’s encounters all feel the same because they rarely force the player to rethink their strategies or attack along different vectors.

Persona 5: Almost Pokemon

Persona 5 looks like it should have no problems with enemy differentiation: enemies have unique type advantages and unique movesets! But the experience of playing the typing system is nothing short of painful.

The main issue with typing in Persona 5 is that it’s unintuitive. In Pokemon, you could always guess at the type of a new enemy just by looking at it. Fire-breathing dragon with wings? Try Fire/Flying. But in Persona, type advantages have nothing to do with enemy design (and half the types don’t make sense anyways). In my entire run, I successfully derived one typing of one enemy — it was a plant, and I guessed that it might be weak to Fire. If players can’t recall the unique typing of an enemy, then they will not experience the enemy as unique.

Wait, shouldn’t the eight-headed dragon be strong against Nuclear? And aren’t dragons weak against Ice?

Unique enemy movesets are more interesting, but also fall prey to the same problem. In Pokemon, characters primarily learn moves associated with their types, which are, as established, strongly linked to visual design. However, they also often learn moves from semantically nearby types (pick a Water-type Pokemon and teach it Ice Beam!), as well as moves that are appropriate to their visual design (you should not be surprised that Gyarados can learn several Dragon-type moves). In Persona, an enemy’s attack types can’t be guessed from its design, which means that players can’t intuit enemy movesets, and therefore can’t perceive enemies as unique in terms of their movesets.

Persona 5 has the potential to make enemy interactions unique, but flubs it because the visual design can’t support the mechanics. If we think about enemy interactions in terms of tells, I should be able to guess what kind of attacks the enemy will use — and therefore what kind of response I need to take — just by looking at them, without having to consult the wiki every other turn. Pokemon does this. Persona 5 doesn’t.

Note the focus here on typing: typing is interesting because it forces player interaction. You need to use types that are strong against the enemy, and you need to defend or switch out team members who are susceptible to the types of enemy attacks. What would happen, then, if we removed typing? Enemy movesets would cease to be unique because the player would no longer be required to respond to them in different ways. This is where I feel most JRPGs fall, and it’s where Xenoblade falls.

Xenoblade and the Lack of Differentiation

In Xenoblade Chronicles X (XCX for short), enemies are almost completely undifferentiated. Whether you’re fighting a little pig or a volcano stegosaurus, you follow the exact same combat loop.

What contributes to this overwhelming sameness? First, you’re not allowed to interact with enemy attacks. XCX is a real-time action RPG “like” Souls, so you would expect to be able to dodge or block enemy attacks. But when an enemy uses a targeted attack on you, you cannot dodge it; it will hit you no matter where you are. There are no blocking mechanics. Attacks are the only way the enemy interacts with the player, so there’s nothing else to discuss. This game has no reactive interaction: you can simply ignore whatever the enemy is doing, because there’s nothing you can do about it.

Second, you never have to change your strategy based on the enemy. Supposedly, enemies have type resistances, but I couldn’t find any of this information, and you usually have type-specific builds, so changing your attack type is rarely beneficial. Also, while you can attack the “appendages” of large enemies, in most cases this only results in a stagger when it breaks, which makes appendage-bashing a generic approach you use for every enemy larger than yourself.

There is, though, some potential in the combat loop. While a first glance at the system makes it seem like an MMO where you spam combos over and over, there are a few unique features. Every now and then, one of your teammates will call for you to use a specific sort of skill (melee/ranged/debuff/buff/aura), and if you do so within a few seconds, then a boost is applied and you both recover some health. This is the main method of healing in combat. Also, if you leave a spell off cooldown, it enters a “secondary cooldown” which, when complete, grants significant bonuses to its power. Thus, you have to balance combo spamming with healing and secondary cooldowns. The only problem is, none of this interacts in any way with the enemies themselves. You spam undodgeable attacks, enemies spam theirs.

So, while the game does have a wide array of proactive interaction options, none of it is dependent on the enemy you’re fighting. As a result, every enemy eventually starts to look the same: it does damage to you over time, and you do damage to it over time, and there’s no interaction between its idiosyncrasies and yours.

My impression is that JRPGs with irrelevant or nonexistent typing systems often fall into this rut. But I’ll save the grand indictment for after I’ve played a few more of them.


I consider unique and memorable encounters a mark of good game design. It’s regrettable, then, that XCX is so sorely lacking in them, especially given how amazingly unique the visual design is. Its environments are beautiful and filled with incomprehensible ecologies, breathtaking and inspiring only as long as you continue to tiptoe around it all. The game wisely tricks you into doing just this by placing overlevelled enemies in every area — until the very end of the story, you’re constantly surrounded by alien animals who could easily one-shot you, but who aren’t interested in you. The magic of the game is precisely in not fighting — because the moment you pull your gun on one of these strange aliens, it turns instantly into a sponge.

A question remains, then: is it possible to make a bestiary as developed as XCX’s, but also make player interactions with each enemy unique? Monster Hunter has the uniqueness, but not the scale. Is it possible to put the two games together? I’d like to see a game that tries.




Software engineer, epic gamer, and Touhou fangame developer.