# Enemies, What They Do, and You

So, I was reading another topic, and a reply from @Xelnath had a link to his blog. Since I’m kind of bad at Game Design, I figured, “sure, I’ll read it.”

One of the posts deals with his first job of designing the raid boss Attumen the Huntsman in Karazhan…a raid I remember fondly from Burning Crusade. In it, he talks about the issues with Attumen’s kit, and while those are interesting in and of themselves, it’s the underlying process that I feel has real value. It takes the form of five questions:

1. Is this ability clear?
2. Does the player care?
3. Does the player have a response?
4. Is this response satisfying?
5. Does this make sense in this situation and fit the theme?

Given that AI enemies are finite state automata (for instance, a mage-type enemy with the pattern, “If the player is 5 tiles away, throw a fireball at them”, “If the player is within 1 tile, use Ice Nova to freeze them then run to 5 tiles away”) I think this pattern is a great way to take what is otherwise a boring mathematical construct and look at it through an emotional lens, not only providing our players challenge, but also an enemy that does what enemies are supposed to do: lose to the player in an entertaining way.

So, a question - what other approaches do my fellow designers enjoy? I’m sure there’s a few other ways to do enemy design.

4 Likes

I don’t specifically focus on emotion or math. I kind of build a little virtual being and then ask myself “now what does this thing do? What can it do? What are its limits for sensing the player?” and so forth.

This produces a list of behaviors maybe roaming, collecting and / or building are its default behaviors. Just like people and animals in real life something triggers a change from these states. Could be time or even belly full, collected something now take it wherever. And, of course the part you’re talking about… could be heard a sound, sees something in the distance, is damaged by something, etc. All of these latter are events making it aware of players presence.

Then it’s how to react which is basically 3 possibilities. Fight, flee or ignore. Depends on the personality of this virtual creature. If attack or flee more options open up. Prefer attack from distance or up close? Depends on options available to it and the situation. Maybe weaker appearing foes are better dealt with melee and stronger appearing foes are better dealt with at a distance. If fleeing, is it just to evade creating distance between itself and the threat or is it to find safety in a certain kind of environment or is it to find help.

That kind of thing is what goes into my thinking.

EDIT: Obviously, I mean for games that have these kind of enemies. If it is something like my last game it is slide across screen til either past edge or destroyed. And if equipped… shoot at player if chance to hit is above certain point allowing for misses.

2 Likes

Let me waste 10 minutes and try to give you information that people will likely ignore, because of course they will.

And be fucking warned, I am going to talk about game design right now. I know.

1. Everything the player does is in response to some rule that the game imposes upon them.
2. No matter what the player does, the game should not be “stumped”.

So in the case of artificial intelligence, the driving force behind the game’s moving objects, the job of the A.I. is then to be able to always have a move to make no matter what the player does within the rules of the game.

Perhaps the approach is too god damned cerebral, taking into account feelings and emotions. 'Cause remember, everybody has different emotions and feelings and so you can’t really design for that. What one person finds to be a scary part of a game, a million other people think it’s annoying and dumb. What does work is a functional approach based on rules.

An example:

Good AI:
In Rock, Paper, Scissors… the A.I. always has a valid move to make.
In Tic-Tac-Toe… always has a move to make.
WoW… always has a move to make.

If you don’t try, you will lose. If you try, you may win.

In a FPS game, if you can stand in a certain spot and the enemies stop attacking you, then the A.I. has failed and the game fails.

I’m not clear what the question was, but this is the answer to what the purpose and function of A.I. is related to the player.

1 Like
1. Is this ability clear? It’s good to know what’s actually happening.

2. Does the player care? It should hopefully create a conflict in the player.

3. Does the player have a response? You do usually want to resolve conflicts.

4. Does this make sense in this situation and fit the theme? Avoid ludonarrative dissonance and all that.

5. Is this response satisfying? This one has some decompressing to do. The worst case scenario is that the response has become rote and trite. The player is responding to the exact same conflict with the exact same response. The player is responding to a single mechanic with their own single mechanic. The player should hopefully be juggling multiple conflicts and finding the times where they can resolve a few of them at once.

Right now I’m trying to monkey around with always making it so that the player is never doing just one thing at a time, like making elemental attacks push resistances to elemental status effects, so a fire attack could make freezing easier. It’s shaping up to be a very different kind of battle system.

2 Likes
• Is it clear? (worth repeating - goals, feedback, and mechanics must be clear or you’ve already failed)
Does it follow the interest curve? (peaks/valleys leading slighty up to the right)

• Does it promote risky play? (pickups in space shooters that are far away, or clearing all enemies in a wave for bonus Pts)

• Does it encourage replay? (varying rewards, fun mechanics, or part of a meta)

• Is the difficulty balanced?

Gigi

4 Likes

I feel that few games nail replay value these days…

Which ones do that for you and how?

When i was a teenager i could play the same game everyday for months, like e.g. Battletoads, that unbeatable StarWars-Game on Nintendo or those RPGs on SNES (Lufia, Secret of Mana/Evermore/Time, Terranigma). But today there is simply too much to play (well, employment also takes its fair share of time) e.g. i bought “Bioshock 3” and “Soma” on Steam during sells, but i haven’t even started them yet. When i have time to actually play something, i always go back to pvp-games where i get some progression out of playing, as of now it is WarframePvP and Overwatch, two year ago it was GuildWars2. Whenever i play other games e.g. Gungeon, Hyper Light Drifter, Moonhunters or one of the hundrets of android/iOS games i downloaded, it is more out of research and inspiration on gameplay, pixel-art and what kind of games are already out there.

I like enemies that respond at least borderline sentient.

If you check out games like Hyper Light Drifter you see that the enemies have some kind of randomness while approaching instead of pathfinding and highly conditional attacks (look how the grunt is strafing into the wall ( youtube.com/watch?v=t7XDnGmNTig&t=319 ) but all in all the combat was challenging and fun in HLD. The Creator even said somewhere that they tried harder AI but battles become insanely hard as soon as there are more than 2 foes.

I see the same thing in my project (top-down-arpg):

• When there is one enemy that dodge rolls your melee-attacks once every 2s and apart from that heals 2 brainless melee-grunts → 5 star difficulty.
• Or when i tried 2 archers with slow projectiles but aim-leading → you will get hit, even when the projectile has 1s traveltime

So i limit my foes to 2 different attacks and 1 trait like “DodgeMelee” or “AimLeading”, whenever i exceed those and have more than 2 foes in a room the fight becomes boss-level difficulty but of the unsatisfying kind. So I try to make basic and predictable foes. The player wants a challenge but a beatable challenge where - in hindsight - he could see HIS mistake. Every game should be beatable in principle (yeah, i heard Candy Crush has some unbeatable constellations, but i don’t know).

But another game “Moonhunters” takes this too far, the enemies are braindead. It is so easy to outsmart or kite them because the have no pathfinding at all and only one attack, but i only played it for 5 hours or so, maybe they amp up the fight later on.

TL;DR: Group-encounters: make enemies dumb; Individual-encounters: make them smart

1 Like

Gotta say something for that, really AI that is predictable isn’t fun to play with at all, humans are good at detecting patterns even unconsciously, a mathematical construct can sometimes include a random factor that allows unpredictable behavior which I -as player- find it more enjoying -regardless of win/loos- than dealing with same pattern over and over.

2 Likes

I am very flattered to see my approach considered so carefully here.

Wanted to jump in on the topic of “Creature Predictability”.

This is a great topic, because its an area of great creative expression. The way to handle a monster (the response to it) can be three different kinds:

• Reactive - You see it and you respond to the results of its actions

• Proactive - You pre-empt its actions, by shutting it down

• Adaptive - You handle it in the moment, as it acts.

Moon Hunters is a great example of a game that doesn’t trigger much of any response at all - because their ability to threaten you is very mild. Dark Souls is a great example of a game that demands a mixture of all three - plan ahead for what you’re about to face, handle its attacks in the moment, then recover from any damage or bad positioning caused by finishing it off.

I prefer to put similar layering into my creatures, after determining which bucket they belong in:

1. Fodder - exists to be killed. Doesn’t serve an immediate threat, but will provide attrition over time if not cleaned up.
2. One trick - exists to add a singular, specific threat or tool to the game. The suicide bats in Metroid 1 or the Laser Beetles in Axiom Verge are great examples here.
3. Durables - they exists to provide a persistent threat, such that you need to wear them down and pay attention to them periodically, but not constantly.

If you think of all fights in all games, they become combinations of these three. Durables require maintenance, One-tricks demand a specific reponse, while Fodder taxes your attention.

Bosses are just combinations of these three concepts.

4 Likes

@Xelnath - Thank you for writing your blog. It’s really helpful to people like me who are trying to get better at game design. It’s more than mere flattery - it’s the very subject matter we’re all here to discuss!

A lot of this ‘is my creature boring?’ seems to fit into your point #2 - Does the player care? Enemies that you can just exploit the pattern of more or less cease to be threats A) unless other creatures are around that prevent you exploiting the pattern, and B) unless the environment itself prevents you from exploiting the pattern.

Of course, if you can’t exploit the pattern, you may then have a balance problem, because the enemy is now in a tactically superior position, and may not be beatable, unless the player can then change the circumstances they engage that challenge under. Of course, as the designer/implementer, we would try to avoid those situations…but they can slip in despite our best efforts.*

*: This happened to me a lot in The Hero’s Journey, and was a reason some of my levels were just too hard to complete. And it only took me three years and a few posts by an industry veteran to figure out. Sadface.

This seems like a good train of thought. But all design is about looking ahead down forking paths - as many forking paths (broad) or as far down those paths (deep) as possible, reasonable - or as few as is relevant.

For example:

Why is it a problem if the player can exploit the monster AI? It might not be.
Why is it a problem if it is a problem? The exploit path might not be the most fun route to beat the game. (See standing out of reach of monsters on BWL’s Nefarion fight)

in Game Design you are min-maxing for a goal. In Dark Souls - it’s the pressure of a challenge that rewards incremental refinement and persistence. In WoW - its the pressure of coordination, collaboration and self improvement. In League - the same thing.

Know what you are min-maxing for first. That will define your goals.

The Wolf Among Us is not at terribly challenging game, but the story sequences and cinematic combat with losses support their goal of achieving connection with the main character.

3 Likes

I watched the sample play of a boss fight from the Dark Souls board game. It was interesting because they had to break down the boss’ AI so it could be played simply. Having the deck set a pattern that the player can learn, but then change it when certain conditions are met so the player needs to learn a new pattern was, to me, an ingenious solution.

Here is the Kickstarter link. The first video on the campaign tab (about 1/3 of the way down the page) is the one I was interested in but they are all interesting to see how they broke a digital game into a simple step by step board game.

2 Likes

Ahah, that’s great!

I start by asking “what belongs here?”

Since enemies serve the needs of the fantasy you are trying to create, it’s important to place them in the context of the interest curve, character development, difficulty curve, and narrative structure.

Conflict can pick up the pacing from a more relaxed state to keep things interesting, it can also be used to teach or test the player on an important skill. Opposition can serve as a foil against the character to motivate or resolve character development in the narrative.

Enemy design can have a lot of leverage when it serves the experience of the moment that you place it in, if that makes sense.

I think about this more in terms of what is it going to make the player feel, what reaction are they going to have, and how will that motivate them?

I also try to keep in mind that games do not really exist. Actually the entire experience that’s going on when someone “plays a game” is entirely within their own mind - their inner experience. It may seem they are engaged with a bunch of images on a screen but all the processing and feeling and reacting and thinking and experiencing is happening inside the person. So then it’s really a matter of psychology… what’s going on inside the person. So putting stuff on the screen needs to consider what effect it’s going to have on the person or how it will interact with their mind and emotions. What kind of character or personality does the enemy have. What kind of relationship does the player have with the enemy. Often I think enemies in games are far too robotic and that comes down a lot to the AI/programming being too simplistic or repetitive. It’s so easy in game development to ‘save time’ by being extremely repetitive or using the same code, same patterns, same behaviors, same appearances, etc many times over. I like to get a sense that each individual enemy is its own unique character or variation on a theme and that it’s not so easy to just say it’s like all the other instances of it. That I think makes it more personal and challenging.