Thomas Grip thinks a lot about monsters. As the creative director at Frictional Games, developers of horror games including Amnesia: The Dark Descent and most recently Soma, he has substantial sway in how his studio's creatures look and behave.
Grip shares his fascinating approach to creating the unique monsters of recent existential nightmare, Soma. Not everything went as planned, or came out as expected...
What were some guidelines you had in mind when designing the monsters of Soma?
The first thing about the monster design was that they should relate to the theme somehow—that the monsters were put there for a reason. They should make you think about the story. We also wanted to create monsters that put players on edge, and that boosted their imaginations a bit.
Those were the design cornerstones, but the ideas about the monsters did evolve from the start, because we weren’t really sure how to do the monsters at the start. We weren’t sure how the experience would flow, because there was a lot of testing going on. What we ended up with are monsters that are thematic and put players on edge. That set the framework for the monster design.
Thomas Grip creating horrific monsters
Going through the playtesting, how did playtesters react to the different iterations of monsters, and how did you adapt to that in development?
One of the first ideas that emerged with the monster encounters--around alpha time, or even vertical slice time which is about three years ago--was to get players to feel really uncertain in monsters' behaviors, as to whether the monsters were hostile or not.
There are some floating robots that we spent weeks on that had this behavior, but they only showed up for a minute or two in the game. [laughs] The idea was that the players would be curious enough to approach monsters, but also be a little bit afraid—that they might keep their distance and watch how [the monsters] reacted.
But it turns out that players didn’t react like that at all. Either they ran straight up to the monsters, which triggered this aggressive behavior, or they’d stay away from them and not really notice them at all.
"It's that ambiguity that I wanted to convey more in the creatures. But it turned out to be a harder problem than we thought."
So you didn’t have any of this interesting middle ground where people started to second-guess the monsters at all. So that was all in vain. We then switched gears on the monster design so that they were either really docile or pretty aggressive instead. We had to sort of polarize enemies a little more than before. I’m not sure what went wrong, but I think one thing we missed out on is the narrative bit--that maybe the monsters should’ve spoken clearly, that the monsters should’ve had more foreboding behavior so that players knew what to expect.
But yeah, that was one of the big findings from playtesting, that we had to polarize the monster behavior.
Whether or not the monster design turned out as you initially intended, it’s interesting that you considered prodding a feeling of curiosity about the monsters. I think a lot of times, as a player, you obviously want to stay away from the monsters, or obviously kill them. It’s rare that you’re drawn in by a monster out of curiosity.
Hitting the grey zone of how players approach monsters is a very big part of the game. Not just from an enemy standpoint—which didn’t turn out as good as I hoped—but from a story standpoint, an ethical standpoint, we wanted players to think, “this robot is screaming…am I going to care about it? Or am I ok with that?” Or if you do something that hurts a robot...should you keep doing that?
It’s that ambiguity that I wanted to convey more in the creatures. But it turned out to be a harder problem than we thought.
You said before that the way your team did monster AI in this game was a risky approach. Why was it risky, and what would’ve been a less risky approach?
One aspect that we talked already about was ambiguity [in theme]. But another thing that also has to do with ambiguity is ambiguity in behavior.
The idea was that if you have fast enough behavior that still follows some strict—yet still fuzzy—rules, the player is going to project a lot more intelligence and intent behind the monsters than what actually exists.
We wanted to make sure monster mechanics weren’t like the ghosts from Super Mario where [encounters] became very obvious gameplay objects, in which you try and optimize your path through them as much as possible. Rather, we wanted players to be constantly unsure whether or not they’re doing the right thing.
"If it’s too mechanical, it’s a very uninteresting, systemic feel. And if it’s too random, monster interactions will be a haze that the player can’t make sense of."
Doing that turned out to be really, really hard! Because the balance you have to have is that if you go to far on the mechanical side, they become gameplay objects. And that’s no fun anymore, because the fear of the unknown is something you want to draw a lot of your horror from. On the other hand, if monster behavior becomes too unknown, then the player has no idea what’s happening with the monsters, and they become an annoyance.
“Can the player make a mental model of them?”: that’s what you always want to be doing, and you always want the richest possible mental models of a monster in a player’s mind as you can have—meaning you want their imagination to be as wild as possible.
But if it’s too mechanical, it’s a very uninteresting, systemic feel. And if it’s too random, monster interactions will be a haze that the player can’t make sense of. So that was the risk.
Some players had one good encounter—some had all good encounters. And it varied which monsters were considered “good” [experience-wise]. Overall, we didn’t hit a good amount of players getting the right kind of experience, even though some players got a perfect experience all the way through.
The thing I’m thinking of is how to fix this. We could’ve made the monsters a little bit better. But then I also think that if we went a safer route, the players who said they had a perfect [monster experience] might not have had such a good experience.
It’s very hard! And it’s an interesting design problem. Am I good with 30 percent of players having a 10 out of 10 experience, or am I ok with no one having a 10 out of 10 experience, but having 60 percent have an 8 out of 10 experience? That’s the risk we took.
Something at the end of the hallway...
How much of this player data that you collected even applicable to future designs?
We're a bit unsure. One thing that we relied a lot on was to get a lot of gameplay from 'nothing.' Just hearing the noises, being in a room wondering where a monster might come from, and seeing the player react told us whether or not we had enough behaviors, enough narrative stuff going on that that would be exciting for players.
Some [long-time horror game] players liked this approach, and some thought, “Auughhh, this is getting boring for me, hiding from monsters is getting old for me!” Whereas other players who've not played a horror game since Amnesia thought, "This is awesome!"
"You always want the richest possible mental models of a monster in a player’s mind as you can have—meaning you want their imagination to be as wild as possible."
Going forward, I think it’s time to leave this “wait-and-run” behavior a bit behind and see what more we can do instead.
You’ve done horror with no combat—do you think you could do horror with no monsters?
Yeah, we kind of already did in Soma. [Ed. note: A certain level in the game plays with this concept.]
Yeah, though I wonder if that’s something sustainable design-wise throughout an entire game, because the fear is still of monsters, although in the anticipation of monsters.
That level was inspired by a staircase simulator, in which all you do is walk down a staircase—that’s the entire game. There’s something about just strolling down that makes it really creepy, especially if you read about the SCP and you know that something’s lurking down there.
But I had trouble getting to the end of that game because nothing happened. My anticipation just kept building and building and building, and it was one of the few games where I’m like, “Fuck, I am not sure if I can handle going down more.” It was just a stupid staircase!
So [in Soma], the idea was to do something similar...where you’re going forward instead. That level was another risk—it was a lot [more difficult] before. From early playtesting we found that players either thought it was the best thing in the whole game, or they hated it. It was almost 50/50.
You brought up that maybe it’s time to move on from this “hide-and-seek” horror game design. Then where does it go?
I have some ideas right now. It’s about expanding upon the things we tried out but that didn’t work. I think we can make them work. The important part is that you don’t want to make things too gameplay-focused.
There are two things that you can do: One, you need to have monsters that still have a lot of unknown to them, but still rely on the player being an active [participant]. The player needs agency in how the encounters play out.
The other thing is that players should have more options in how they go about the way they play. You see it a bit in games like Alien: Isolation where you have crafting and tools to lure the alien in. I think more stuff like that, where the player isn’t just reactive to the horror, but is able to make plans, and have a lot of ways to move forward, [is a design path]. So basically, the players are active in encounters in a way that’s not just about hiding, but also actively encountering monsters—though without fighting them and looking the monster in its face, in order to preserve an element of the unknown.
It’s about giving more options in these encounters, rather than doing the old hide-in-the-closet trick.