PROPER ENEMY DESIGN I

Role of the Enemy

Proper Enemy Design I of V

Role of the Enemy - Making your enemies noteworthy.
Roles of the Enemy - Making your enemies diverse.
Power of the Enemy - Making your enemies fair.
Strategy of the Enemy - Making your enemies smart.
Allies of the Enemy - Making your enemies work together.




Preamble

RPGs are about combat. This is pretty much a fact about both the genre and the people that seek it. Graphics, music, story, and interface are all equally important parts of the cake - Graphics entice people to try your game, Music plays with their mood, Interface immerses them, and Story makes them wish to proceed. These are all well and good, BUT.

Combat is the single most important part of the standard RPG - it consists of the vast majority of the player's decision-making, and most of the systems in a given RPG are geared towards granting the player more power in combat. When you have a crafting system, what do you craft but weapons, armor, and items? When you introduce some sort of food system, you make it generate stat-ups for the player. All of these are applied to combat, the central focus of the game. Combat does the heavy lifting of making your game fun - it challenges the player, gives them an immediate goal (kill these dudes), gives them a set of choices from which to determine an optimum (which of these dudes to kill first), and creates the hazard of failure that creates entertainment when avoided.


The Role of the Enemy

The star of the show in RPG combat is the enemy - the computer-controlled monsters, bandits, fiends, and ghouls that seek to bar the player from advancing further into the game. To this end, they employ damaging attacks, status effects, buffs, debuffs, and a number of other effects to achieve victory over the player. They keep the player on their toes, and force the player to strategize in order to win and proceed. Enemies make the combat in your game fun.

Or... Do they?

Frequently, enemies have the sole option of a basic attack, or a single elemental damage spell, or a heal skill. These kinds of adversaries are inherently boring - they do one thing, repetitively, with absolutely no variation. Even the first enemies in the game don't deserve this sort of treatment. A common complaint about the RPG genre is that "RPG combat is boring" - this isn't the fault of the genre, but of the developers who pay more attention to the fluff and window dressing rather that the meat of the game.

The simplest way to make your battles more interesting it to increase the number of things enemies can do. An enemy that can buff its Attack, debuff your Defense, and then attack you is more interesting than an enemy that simply attacks. A general rule of thumb is that no enemy should have less than three actions - say, an Attack buff, a Defense debuff, and a regular attack command.

Another thing you can do with your enemies is give them notable properties. Something innate to that enemy that makes them noteworthy can stand in for an action. The emphasis is on the severity of the property - An enemy that regenerates 5% HP each turn is not notable, one that regenerates 50% HP per turn is alarming. An enemy that is immune to Fire is not notable, one that is immune to Physical attacks is noteworthy. In these cases, the property drastically affects how the player deals with that enemy, and thus, makes the property worth an action in terms of enemy complexity.

Enemies don't generally need to be too complex - that depends on how much backup they bring with them. A group of eight enemies will need simpler enemies to avoid overwhelming the player; a lone enemy needs to have an intelligent moveset to be able to keep up with three or four player characters, and may even warrant multiple actions per turn to stay interesting. The idea is to strike that delicate balance between spacebar-mashing simplicity and ragequit-inducing impenetrability.


TL;DR

Make your enemies do more than just attack, or just heal, or just buff. Enemies that make tactical decisions can exploit mistakes of the player, and give players mistakes for exploitation. Interesting enemies make for interesting battles.

Posts

Pages: 1
Good read, but I think the focus on adding more attacks is nowhere near as important as creating patterns and predictability.

If enemies randomly perform Action A, Action B, or Action C, i can only hazard a guess at which is coming next. If I know they tend to like attacking, so I use a shield skill but they decide to use a defense debuff skill, I wasted my turn, and this leads into the cardinal rule of standard RPG fare combat:

Damage. Is. King.

Enemy might hurt your part a lot? Kill it before it can.
Enemy can debuff party defense? Kill it before it can use that advantage.
Enemy can heal itself/others? Kill it faster than it can heal.

The way around this is to give the player power to react; if I see the enemy and KNOW it's going to do a massive damage attack next round I can balance my next round with offense/defense/buff so my party can survive and no actions are wasted.

This is how you add strategy; the player knows what's coming, so they can react with a strategy they develop. Otherwise the strategy really just resolves around killing the enemy as quickly as possible and having a certain level of healing spam since you can just completely offset any damage if you heal enough.

Games across genres share a lot, pattern recognition is hugely important in other games that are fun in the combat aspect, example:

Mega Man X: (in specific, but ANY will do)

Sigmas hands are flying to the left and right? Uh oh, time to climb one far wall to dodge the beams he's going to switch from the near edges to the center with. I KNOW that the center is a bad area to be. Does this make the fight easy? No. It makes it possible.

If he always moved his hands to the same spots and randomly did any move, I could end up guessing wrong and getting blasted which is frustrating.

How do I get around that?
1. Lots of healing.
2. Killing the boss before I die.

Both are the standard method of winning poor RPG turn-based systems.

Another example: Classic Mario

I see a platform goes up to this point, then comes back down. I know when to time my jump and where I'll need to go on the next stage. The skill comes in at the execution stage (choosing which path through, and how well I execute it.)

What if those platforms randomly went in the wrong direction? An unlucky guess and you can end up in an dead-end situation.

Yet Another: Final Fantasy Tactics

Tactical systems let you see where the enemy is going, when they do moves other than their usually basic SINGLE attack they have to charge them, and I can see the order so I can react appropriately. Enemies don't usually have a huge range of skills, and they only use skills for healing/restoration when needed, and I can tell when because of their proximity to other targets.

If I was out of potions and I was fighting a goblin and chocobo in the early game (Mandalia Plains~) and the Chocobo would kill me next turn, but I couldn't kill it, I could hurt the goblin badly to force it to use it's heal skill, giving me time to get more offense focused on killing it or escape.
LouisCyphre
can't make a bad game if you don't finish any games
4523
That really touches on II (What an enemy's role or "class" should be; that is, making enemies follow recognizable patterns like "All wisps use Fire attacks") and IV (Making the enemies smart enough to react to the situation; don't heal an ally at full health and don't let poison sit there if you can do something about it, etc.)

This all works in reverse, though, as well. If the enemy raises its attack power and you subsequently lower it, the enemy's turn is wasted. Giving them more actions isn't an end-all, be-all solution. It's just the first step beyond "this enemy attacks, this enemy heals," and hence, it's I of V.

I'm for and against set attack patterns, like an enemy that always casts fire -> water -> fire, for example. Oops! Fire Shield, you did nothing. Oops! Water Shield, you did nothing. Oops! Fire Shield, you did nothing. Patterns based on the situation are definitely prefered - The boss will counter with his attack unless his strength is lowered, at which point he counters with Fire III instead. If you're immune to his counter he'll remove the strength debuff, and so on. It's up to you to explore the pattern and exploit it, instead of being completely obvious what the enemy does.
Ah, sorry, more to come; maybe have the first article briefly over all items and then each after expands on those with more detail and examples?

Otherwise reading just the first is confusing since I don't know what points and if I wanted to use this I'd be going in the wrong direction before the rest are released.
LouisCyphre
can't make a bad game if you don't finish any games
4523
Yeah, I can do that pretty easily. Sec.
comment=29730
I'm for and against set attack patterns, like an enemy that always casts fire -> water -> fire, for example. Oops! Fire Shield, you did nothing. Oops! Water Shield, you did nothing. Oops! Fire Shield, you did nothing.

This sucks for bosses, but I for one would love to see a normal enemy or two with a predictable attack pattern. It would be a fun little gimmick that requires you to think just a little bit. The sort of small thing that helps the game along.
LouisCyphre
can't make a bad game if you don't finish any games
4523
One of the best things you can do for your solitary bosses is give them multiple turns to compensate for being outnumbered. It allows them to get off combo attacks like Sleep -> Death (can't dodge will sleeping!), and it lets them remove a debuff or two and still deal some punishment.
This is quite an interesting discussion. However, I firmly disagree with the point of set attack patterns adding strategy.

A 100% patterned fight is nothing but memorization. You figure out the correct moves once and you will have found a perfect solution to the fight, same way a 100% patterned fight in an action game would just require muscle memory. Except without the part that even requires the muscle memory. Then you just repeat the correct sequence of action ad infinitum (for however much HP the enemy has) and win.

But what if the enemy might use one of three possible actions next turn? The player does not have the 100% knowledge of how to make the perfect preparations to neutralize the next enemy attack before it even happens. They must now anticipate all three outcomes and formulate a plan from the options they have, none of them being a certain "trap card", so to speak.

The player now has choice.

Choice is vastly important. Both choice in setup to the fight as well as adapting to the flow of the battle. Perhaps the boss has a dangerous composite attack (a great way to add depth to the system - one action need not do just one thing) called Blinding Blaze that deals both heavy fire damage and blinds the target. Do you block the damage, or the status? Which are you willing to deal with? Or do you forsake both and go for a stat twinked setup?

Let's use an example about adapting to the flow of the battle with an unpredictable enemy party - let's say you have two characters hanging on at lowish health and two dead. Do you revive, hoping to gain more actions to your side to deal with the situation with the risk of the revived character getting wiped out before they can act? Do you heal the living, with the risk of getting overwhelmed with just two acting characters? Do you go for a suicidal blitz, hoping you have enough firepower to win the fight before you succumb to the enemy's offensive? Do you status out the enemy, instantly making the playing field more even but not putting you closer to winning the fight in terms of raw HP numbers? The trick is to figure out the action that will not be "wasted".

Contrast an action game boss that is not perfectly patterned. You must mind your position on the battlefield so you won't put yourself in an inescapable situation no matter what move is used next and figure out the correct time to attack, as well as dodge.

This is the element of choice.

In fact Final Fantasy Tactics is an interesting example, given the party and the enemies play mostly by the same rules. Now, vanilla FFT enemies suffer from a lack of competency and choices to make, but let's use a hack such as LFT where the enemies are on more even ground as an example. If you go in with all guns blazing with a pure attack+heal mindset, you will lose any difficult fight. The AI of FFT is near unparalleled since if it is given a sufficient number of different actions to use. Wounded warriors will retreat to the back ranks behind the meatshields to receive healing. Mages will buff and debuff, nuke and revive appropriately. You will have to consider any possible actions that can be used and adapt to the strategy used by the opposing force.

There is always a better strategy that singleminded damage damage damage, because the non-direct damage options are good enough. This is an important pitfall of design - no one cares about your enemies using a 10% defense buff or a poison status that shaves off 1% of your maximum health per turn. Make those skills matter.

And now to respond to a few lines specifically:

An enemy that is immune to Fire is not notable, one that is immune to Physical attacks is noteworthy.


This comment makes the assumption that Fire element is not a notable part of the party's offensive output, and that Physical attacks, in turn, are. Why does this have to be so? Systems should not be afraid to stray from the norm, a comment such as that should not be absolute law.

Otherwise the strategy really just resolves around killing the enemy as quickly as possible and having a certain level of healing spam since you can just completely offset any damage if you heal enough.


This is an issue completely unrelated to unpatterned fights and makes a number of assumptions about the (flawed) system used.

1. Enemies can be blitzed before they can do anything noteworthy to your team.
2. Healing can be spammed indefinitely.
3. Enemies cannot disable the healer(s).
4. Healing is as powerful as to basically be a reset button.

Sounds like this hypothetical game has an overly powerful defensive safety net. Valkyrie Profile is notorious for this - enemies can nuke your party from orbit for damage that overkills them all three times over and it will not even be threatening. Why? Because there is a high chance passive skill that keeps the character at 1 HP instead of dying if they had over 10% before the attack, auto-item, which automatically has the survivors use revival items on dead characters as a free action, and as a THIRD layer of immortality, Angel Feather accessories which automatically revive your character and have a 30% chance of breaking afterwards. And you can equip three at a time. Indeed, end- and postgame foes will overkill one or more party members every turn, but you just don't need to care.

Well, that sure was a longer post than I intended. As I said, interesting topic!
Very interesting discussion. :)

First, knowing what your opponent is likely to do on the next turn doesn't always require the same reaction; each round can differ and the strategy differs based on what you're fighting.

A round where:
* Enemy A is about to nuke
* Enemy B is going to defend their entire team with a shield
* Enemy C is going to heal because Enemy A is hurt

opens a lot of options; what if you have a method to take down a shield? Maybe you've buffed up the speed of one character allow them to nullify shields before the other characters go. Or your game has a mechanic where you decide player order.

I'd counter that 3 possible actions doesn't broaden your choices, it limits them. If you don't know what exactly is coming, but know what potential actions are, you need to simply use a strategy that accounts for all of them or at the least the most damaging of them. If you can't account for an average of them all you run the potential for an unavoidable game over because of a choice you couldn't have known not to make at the time.

One method to make combat more predictable for example: a hate system. By having one character focus an enemies hate on them, you can help ensure the enemy attacks that character. This allows you to predict the enemies actions (you know who they're going to hit) and allows you predict when they need to be healed vs. having enemies randomly decide to strike on your healer, making them spend their turns healing themselves and wasting any time you spend buying good armor or defensive skills for the first character.

Your FFT example with the advanced AI is also good; if you know the enemy will do something smart on the next round, for example retreat, you can use that as a weapon to help yourself as well. If I'm being overwhelmed, I can use damage to force some of the more troublesome foes to retreat so I can focus on the others and give myself time to recover. The more intelligent the AI, the more predictable it actually gets (higher chance it'll do exactly what I would do on the next turn; take out the weakened enemy there, or pull back to recover, etc.)

Predictability does not mean an enemy always does Fire 1 > Fire 3 > Ice 2 > Heal 3 > repeat.

Predictability means you can more accurately control the flow of battle as you wish using a range of abilities you have.

It's taking that ability to use your actions to change the flow of battle that exists in so many other genres that needs to be brought more to RPGs; in a Mario game I can take a different path to avoid a dangerous spot, in Mega Man I can run under a jumping enemy to attack them from behind, in Fallout 3 I can use a dart gun to take out the legs of a Deathclaw so I don't have to deal with it's excessive speed.

In each one I can choose to do things differently based on my own level of feelings for risk vs. reward; but I never feel that choice in RPGs because the 'random' element needs to be planned into the combat system

Player feedback: "Oh, this enemy is too frustrating because two unlucky turns can end up with a character dying no matter what I do!"
Developer: "Okay, toned down the enemy."

I think a good combat system is one where I can win by thinking and concentrating to keep myself playing on the edge, and a poor one is where I'm forced to use safer strategies because that's the only level of control I have.
Will just have to agree to disagree there. I absolutely despise the "tanking" mechanic present in MMOs. It's the kind of thing that further makes systems offense-based, as characters beyond the one meatshield don't largely have to care about their durability at all, because they rarely get hit and are getting oneshotted if they are hit anyway. Which simplifies the choice of which stat to focus on.

What ideally controls the flow of the battle in the way you mention other genres do is various non-damage abilities. A silencing ability to take care of a magical threat, speed debuff to ground the multiturning enemy, and yes, even a taunt to temporarily (this is key) make enemies concentrate their attacks on the character you want.

The "random element" or RNG is an inherent part of RPGs - how much it affects the game is up to the system. (I am fine with most RPGs in this regard, but absolutely staggered by how luck-based tabletop RPGs seem from a glance.) If you are looking for a system that entirely removes every random element? It is an interesting thought process, but probably very hard to make interesting at all.

And some invidual picks:

First, knowing what your opponent is likely to do on the next turn doesn't always require the same reaction; each round can differ and the strategy differs based on what you're fighting.

A round where:
* Enemy A is about to nuke
* Enemy B is going to defend their entire team with a shield
* Enemy C is going to heal because Enemy A is hurt

opens a lot of options; what if you have a method to take down a shield? Maybe you've buffed up the speed of one character allow them to nullify shields before the other characters go. Or your game has a mechanic where you decide player order.


Ultimately, if all of these actions are predetermined, you can retrace the steps to produce the exact same encounter and result with these three enemies, unless there's some kind of random element involved. Memorization.

I'd counter that 3 possible actions doesn't broaden your choices, it limits them. If you don't know what exactly is coming, but know what potential actions are, you need to simply use a strategy that accounts for all of them or at the least the most damaging of them. If you can't account for an average of them all you run the potential for an unavoidable game over because of a choice you couldn't have known not to make at the time.


Ideally you can approach the situation from a number of different angles, provided the actions are sufficiently close in overall harm done and not unfairly punishing. BOLTKILL/Fire 1/Ice 1 makes it an obvious non-choice, while BOLTKILL/FIREKILL/ICEKILL can be a stupid guessing game where you guess right or lose. What angle you choose and what move is actually used, what is your response, the enemy's response, the battle can unfold in a number of ways if well designed.

hate system


Rather than making it a waste of time to upgrade the first character's armor if characters two, three and four take some hits, I see it a total waste for character 1 to soak up all the hits making character 2/3/4's armor completely superfluous.

Predictability means you can more accurately control the flow of battle as you wish using a range of abilities you have.


This I agree with. Just the methods seem to be different!

In each one I can choose to do things differently based on my own level of feelings for risk vs. reward; but I never feel that choice in RPGs because the 'random' element needs to be planned into the combat system


A patterned game like Megaman holds no risk, in my opinion. You either do the pattern right or you don't. It's really hard to do right, granted, but it's still largely memorization if your reflexes don't get it right the first time. Random element forces me to adapt.

Player feedback: "Oh, this enemy is too frustrating because two unlucky turns can end up with a character dying no matter what I do!"
Developer: "Okay, toned down the enemy."


Depends. How much of a setback is a character dying in this case, anyway? Tester feedback is very valuable but going with every suggestion is just going to lead to contradictions as different testers will have different approaches to fights. If everyone's saying a mandatory boss is utterly curbstomping them, then you're most likely doing something wrong obviously.

I think a good combat system is one where I can win by thinking and concentrating to keep myself playing on the edge, and a poor one is where I'm forced to use safer strategies because that's the only level of control I have.


I definitely agree with the first part, but...what exactly do you mean by "safer" strategies? I am confused. Is encouraging defensive play a bad thing in the land where damage is king?
LouisCyphre
can't make a bad game if you don't finish any games
4523
where did this come from Oo

*reads*
LouisCyphre
can't make a bad game if you don't finish any games
4523
Alright, read up. I'll just throw in the point of there being better ways to handle multiple actions than just a roulette. Enemies need to be smart enough to stand a chance against a fully applied human brain, after all.

I think I've got enough to go on to part II sometime this week. If I don't, remind me.
l think you're right; we have the same ultimate goals but the middle-ground process is different for each of us.

For example, each of your mentioned methods is something I'd consider part of a prediction-based model:

1. Temporary taunt (how is this different from hate? I don't know any 'permanent' hate systems, most MMOs use a temporary taunt system as well, FFXI, WoW.)
2. Silence effect - These are nothing but methods to help you lead an enemies actions - a mage that is silenced doesn't likely have a high-level Holy Sword skill and high melee enhance stat. (And if they did, why bother with the status effect?)
3. Speed debuff - Again, you're working to assign the enemy to a slot location more advantageous that you KNOW so you know you'll get characters off first; if the speed debuff was random between 5% and 80%, would you use it? If you don't know if your character will go first, why bother debuffing speed?

I haven't played a lot of the tabletop RPGs out there, but D&D does reward and focus heavily on predicting/controlling the actions of your opponents; status effects to stop them from acting, forcing them to move to various areas or zones, or heavily penalizing them for walking through areas or taking actions, and many ways to force enemies to fight a certain "tank" or to ignore certain classes.

I don't think the 'risk' of the unknown needs to be so pronounced in RPGs; it's lead to a number of serious downfalls in the basic combat model:
1. Status effects usually being useless since they don't always hit and the most useful places (bosses) are usually immune
2. "Safe" tactics - Attack + Heal = Win - most secondary skills and effects are ignored, and if you require people use them to win, they have to guess how you wanted them to play (Hitman series anyone?) once they have guessed the ideal strategy... it's just memorization.

I think trying to get that feeling from Mega Man to an RPG Combat system would be a great ideal to reach for; you spend your time knowing 'what' you need to do, but doing it correctly depends on your focus and ability, not on random luck.

I get frustrated when I screw something up myself repeatedly, but I get angry when I didn't do anything really wrong yet I lose.


Also, curious about what I meant by "safe" strategy, the best example:

The damned table floor-boss in Persona 3 - up to this point the game rewarded me for being smart, I took the right people with strong elements into fights, and I was able to proceed with no grinding and really under leveled.

Suddenly floor-bosses got a new ability: Megi-spells.

These spells hit your whole team, can not be blocked or reflected, and damage can only be reduced by reducing their attack/accuracy stats, no one has defense against the spells. The ONLY viable tactic was to use a broken skill that nerfed their accuracy really hard-core (Ness mentioned she used this method to for most of the early game to make it all really easy) or to be able to out-heal their damage. Otherwise it was only a matter of time until the boss did the attack enough times in a row to completely wipe your team. (I got REALLY close once, but he used the spell 6 turns running!)

I could extend my life by bringing team members strong or immune to his other damage type, fire, but he would always eventually use it 3+ turns in a row and that was that.

I finally beat this using the old tried and true RPG method: grinding until I got the second level group-healing spell, the boss fight became a lovely slug fest of me healing my team with that skill every turn until it died through the rest of my team chicken-pecking it to death.
Ah. "Safe" as in unimaginative strategy. Gotcha, that point confused me quite a bit. A safe strategy sounds more akin to defensive turtling to me instinctively, but that's because your definition of "safe" tends to be very suboptimal in the combat systems I like.

But yeah, plenty of options that are worth using = more fun combat.
I think it's not only a question of how unpredictable enemies are, but also in what way they are unpredictable. Let's say we have a boss which each turn stuns one of the characters. With a standard RPGM four member party, that means only three will be able to act on a given turn and if the target chosen is random the player can't predict which three it will be that are available. This makes that element very unpredictable, but how this plays out changes drastically depending on if the boss stuns a character at the beginning of a turn or at the end of a turn.

If the boss stuns a character at the end of the turn, the player will know which characters that can't be used for the current turn, but she/he can't know who will be targeted the next turn. However, if the boss stuns a character at the beginning of the turn, the player won't even know who can act for the current turn. I prefer to have it at the end of the turn.

Making it at the end of the turn means the player has to form four different strategies depending on who is stunned. Making it at the beginning of the turn means that the player have to form one strategy that accounts for all four outcomes. Of course, other circumstances can also force her/him to adapt. How healthy the characters are and whether or not the boss is about to launch it's strongest move will affect the strategy. Even so, have the stun happen at the end of the turn creates four different situations for every single situation there would have been without the stun gimmick.

Whenever you give monsters a skill and/or AI setting you should be conscious on how it affects the strategies the player has available. The options will add and remove strategy. For example, taking an enemy who only has a physical attack and giving it a spell as well will remove some options. Inflicting blindness, boosting party defense or debuffing enemy strength now only has a 50% chance of doing anything at all. In theory the player can now also inflict silence, boost party magic resistance and debuff enemy magic, but in practice you will rarely see players use strategies which only will do anything at all 50% of the time. This is however not necessarily a problem. There are still other strategies available. Sleep and paralysis for example doesn't give a damn about what kind of attack the enemy uses. Removing some strategies can mean that the player has to search for other strategies.

Consciously changing which strategies are useful and not useful between every enemy can force the player to adapt and ensure that not all encounters are the same. However, this requires that you don't leave a catch all strategy that will always work like the "kill it fast" strategy Anaryu described. Also, for the best result you both want to close of a lot of strategies and leave a lot of strategies available simultaneously. If you don't effectively close of a lot of strategies the player can pick two general purpose strategies and they will then work for 95%, or more, of the battles. One the other hand, if you don't leave a lot of strategies available you're not really letting the player formulate it's own strategy, the player can only use the strategy you devised or fail.

What I think would be even better is if enemies closes and opens up strategies depending on what move they use. That way, while the first turn against three dire wolves will always be the same, subsequent turns will be fought differently depending on what moves the enemy chooses. The stun gimmick I described earlier is one such scenario although that particular one may not be very appealing. In any case, my main goal is to find ways to accomplish exactly that.
The stun move thing reads like the main argument for why I consider traditional turn-based should be left behind and only used for games such as Pokemon where it's more than a gimmick to frustrate the player.
I don't know about all of you, but my enemies (in WeaponBirth) will actually have to have AI and do tons of things even for the simple act of trying to hit the player. In fact, lots of realtime RPGs I've played have enemies process state machines and the like.
Pages: 1