9 Years, 9 Lessons on Horror

It has now been over 9 years since we released Amnesia: The Dark Descent. That is a bloody long time, and feels like we should celebrate that by talking about the craft of horror games.

Horror games are quite a different beast when it comes to the game industry at large. Most other genres revolve around what the player does. In a turn-based strategy you take turns doing strategy:

Into the Breach

 In a first-person shooter you shoot things from a first-person perspective:

Doom

In a Match 3 game you match three thingies:

Candy Crush: Soda

In a horror game, the activity is not at all as important. What is important is that the experience is a spooky one. This makes designing horror games different from designing within other genres. Many times the standard industry tricks just won’t work, which makes one think about game design in a different light.

In the past 9 years we have learned a great deal about horror games, and to celebrate the occasion, I wanted to share 9 lessons we have learned over the years.

That being said, I don’t see these lessons as only useful for horror games. There’s quite a bit of overlap with other genres, especially any games that aim for a narrative-heavy experience.

And finally – this is by no means an exhaustive list. Still, the lessons here are at the core of the craft of making scary video games.

Lesson 1: Horror is not enjoyable

The basic emotion of horror is not a pleasant one – yet people play horror games wanting to experience horror. This is the paradox of horror as entertainment. This paradox requires game developers to be careful in how they deliver the experience to the player.

You could draw an analogy between horror games and rollercoasters. The basic purpose of a rollercoaster is to simulate the sensation of falling. Under controlled circumstances the experience of falling is thrilling and fun (at least for a good portion of people). But if you put someone in a barrel and push them down a cliff, chances are they will not find the experience fun at all. Even if they survive unscathed, the whole ordeal would be a horrible experience.

The same is true for horror games. If you have a game that only relies on jumpscares – figuratively throwing people off a cliff in a barrel – few people will consider that fun. This became apparent in certain maps in Penumbra. We thought it would be good enough for a scary gameplay section to have a maze and some monsters. Instead of becoming mazes of fear, they instead became mostly… annoying. Amnesia: The Dark Descent had similar issues towards the end, where the monster encounters were just that, not supported by any other aspects. At that point the game no longer felt as entertaining.

Well, a familiar face.

Lesson 2: Players are working against you

For a horror game developer, the worst enemy is… the players. Seriously, if we could sit around and make games without having to worry about what the players will do and think when playing the game, life would be so much simpler!

As mentioned before, being scared is not a pleasant feeling. Therefore the players will try to optimize the feeling away, often unconsciously. In the end, the players will ruin the intended experience for themselves.

Take the demon dogs from our first game, Penumbra: Overture. The game takes a bunch of time to build them up as creepy monsters that stalk the dark mines. However their AI has some weaknesses that some people are very quick to catch. Hence the dogs become easy to defeat, and are no longer scary.

Can’t get me. I’m on a box.

And the crazy thing is that the players complain when this happens! They probe the system for flaws and choose to exploit them, yet want the dogs to remain scary. So their behaviour ends up going against their will.

Some games solve issues of player exploitation simply by making the enemies extremely hard (think Dark Souls): they make sure the monsters are just as hard to beat as they look scary. Another approach is to instead skip much of the gameplay (think Dear Esther): if there are no mechanics, there’s nothing for the player to exploit – problem solved, right?

I don’t think either of these solutions is optimal. Instead I think one should aim for a third route: making the players think about actions in a more narrative fashion. More about that later!

Lesson 3: Scares alone won’t make a horror game

Horror is like a spice that defines a dish. You cannot do without it, but you can’t cook a dish solely out of spices either. That would be just gross.

As an example, let’s take three horror movies I consider to be at the top of their genre: Alien, The Exorcist and Ringu. All three movies deal with very different subjects, have different styles, and are overall different from one another. But there is one thing they have in common: they all have very few scares in them!

Instead each movie is mostly about the characters, the discussions, the anticipation of the horror – building up the atmosphere and the dread of things to come. Very little time is spent actually facing the horror.

Let’s get back to our roller coaster analogy. When you think about it, the actual roller coaster ride lasts a very short time. Most of the time is spent doing things like buying a ticket, standing in line, and hearing other people scream. All these actions are not superfluous extras – they build up for the actual ride, and are crucial to the overall experience.

When we first made the study section of Amnesia: The Dark Descent, we implemented a ton of jumpscares. Books fell down from shelves, doors banged, pianos started playing and so forth. But as the map became more complete, it felt like something was off. So we reduces the scares to just a couple, and instead focused on letting the player learn the castle’s mysteries. At first we were afraid this would make the level too boring – but as it turns out, spacing the scares apart made players much more scared than previously.

In horror, less is often more.

Lesson 4: Fun gameplay is just too… fun

In a horror game more than any other, the players go in expecting to have a bad time. And as designers we want them to feel anxiety, despair, and a whole array of negative emotions. But gameplay – because it’s so damn engaging – tends to counteract all these juicy emotions.

Let’s use Dead Space as an example. When I started playing it, I was really scared, walking around slowly and peeking around every corner. Then, about an hour in, I learned how to kill the monsters, and what tricks I needed to survive.

Dead Space 2 promo art
All of the fun, none of the horror!

Not only did I get good at killing the monsters, I thought it was great fun! The things that used to terrify me now became a source of amusement. Instead of dreading the monster sounds they now made me excited – oh great, another necromorph to dismember!

So where did the fear go? It was simply overshadowed by the rewarding gameplay.

Us humans tend to have this thing called attention, and we only have a limited amount of it. If the game is constantly engaging the player with thinking about their aim, checking ammo, and looking for loot, there’s no room left for much else. In other words, the players’ brain will lack resources to frighten themselves.

The early designs of Amnesia: The Dark Descent included genre-typical weapons, and even guns. We also experimented with very elaborate puzzle set-ups, everything from swinging chandeliers to redirecting rays of light. All these caused the same issues as Dead Space. They were too fun, and took attention away from what mattered: getting scared.

Eventually we decided to reduce the “fun” elements the gameplay had – and it paid off.

We saw this very clearly when watching Let’s Plays of the Amnesia games. Since players didn’t have things like combat to pay attention to, they reacted to things they might not have even noticed in other games. A vague sound, almost like a footstep, was suddenly a reason to look for the nearest cupboard to hide in. Had the players minds been filled with thoughts of loot boxes, they would have never reacted like this.

Lesson 5: Narrative is a core element in good horror

So if engaging gameplay can be counteractive to the horror, and you need to be careful with the scares, what do you fill a horror game with?

While no silver bullet, narrative is a big part of the equation.

By building up a narrative, us game designers can make game worlds bigger and more intricate than they actually are in-game. We can prime the player into doing a lot of the scaring for themselves.

In order to explain this, let’s take a random image let’s take a random image of a quaint town:

Aww, I wanna go there. 🙂

This feels like a great place for an evening stroll, right?

Now let’s give this image some backstory. Put on some spooky music, like the Amnesia soundtrack, and read the following:

It has been two weeks since a huge storm cut the town from the rest of the world. All means of communication are down.

Today, our emergency services received a call – it just started out as static, a joke that kids would play, but then the screaming started. The screaming of people, then an otherworldly roar, nothing a man nor beast on Earth could make. I had to find out what happened to these people up the serpentine road from us. 

I am now here, yet no one else seems to be. It’s like everyone vanished. But as the cold sun sets down over the mountain, I get a sense of unease…

…And now look at the picture again.

The worst monster of all is leaving the lights once you’ve been murdered horribly.

Not so cozy anymore, right?

A new context leads to re-interpreting the environment based on this information, and get into a different mindset based on it. While you previously admired the view, you are now scanning it for signs of danger.

A big part of horror takes place inside a player’s head. And by fueling their imagination, we can turn a cozy village into a place of terror and despair.

Looking back on which areas worked in Penumbra, this component became apparent. The most loved environments were those where players could use lore and environmental clues to fantasize what happened… and what could happen. The expansion, Penumbra: Requiem, lacked a lot of this background information. So despite us designing some of our best puzzles and implementing interesting visuals, Requiem was received quite badly. Without a strong narrative component, the players didn’t get the experience they wanted.

Penumbra: Requiem, or as we call it, The Marc Game.

Lesson 6: The world must feel real

In order for a horror narrative to have proper impact, the world it takes place in must be taken seriously by the players. But what does “serious” mean? Grey and brown tones with no cartoonish elements? Not quite.

Let’s draw a parallel between real and imagined worlds. If you suffer from nightmares, there’s a trick to that: make a habit out of knocking on walls, tables, or whatever is closest to you. Eventually you will start doing the same when you’re asleep. However, when you knock on walls or a table in a dream, your hand is likely to go through the surface – that’s how you’ll know you are in a dream, and no longer need to be afraid of the world around you.

Making horror games is basically a business of creating nightmares. But it’s hard to be successful when you have a bunch of players (those damn players again!) constantly doing the equivalent of “knocking on surfaces”, simply by playing the game. As soon as they discover some sort of glitch the immersion of a terrifying world breaks, and it takes a long time to build it back up again.

Let’s look at an example from Penumbra again. In Penumbra we want the players to imagine that the demon dogs are “real”, implying all the traits (demon) dogs possess. So, we want players to be worried about encountering a dog, and hiding from it. However, some players “knocked on surfaces” by messing around with the environments, and figured out that the dogs can’t reach you if you camp on top of a box. So, whereas a real dog could jump up on the box and chomp the player up, the AI dog cannot. Therefore the fantasy of dogs as “real” is lost, and the game loses a bunch of its scariness.

Penumbra: Overture by IsisMasshiro on DeviantArt
The intended reaction when encountering a demon dog

Because of this effect, game developers have to be careful about how they construct environments, and what tools they give to the player. There should be enough things to do to make the place feel real. But not so many as to aid players in breaking the illusion.

Lesson 7: Keep it vague

You know creepypasta and scary photos you can find on the internet? Almost always the thing that makes them scary is that they leave a lot to the imagination. Seeing a silhouette and glowing eyes out in the corner of a photo feels threatening. A close-up glamour photo of the same monster does not.

AAH! What IS that?
Oh, just our good friend Terry bringing us a gift. (by ThiccBoiMyers on Discord)

As mentioned before, much of the horror comes from simply not being sure what the hell you’re looking at. It’s when there is a gap in our knowledge, a certain amount of uncertainty, that horror can really shine. This is especially true when you combine it with some sort of danger element.

It is quite common in games to make sure the player understands the systems in place as clearly as possible. This often results in some really daunting tutorials. Of course for some games, like fighting games, it’s important to have in-depth knowledge about the systems to be able to optimise the game. In horror games we actually want the opposite!

A vague and uncertain game system is like a creepy photo. You can make out enough to get an idea of what’s going on, but there’s still room for the imagination to go wild. Let’s use the health meter in Resident Evil as an example. Internally it is an analog property, a decimal number from 0 to some value, but the player will only ever know that it has “three” states. This strikes a great balance between giving information and being vague, and helps crank up the tension.

The sanity system in Amnesia: The Dark Descent is similarly vague. You know scary things – whatever those are – lower your sanity, and bad things – whatever those are – will happen if it drops too low, so you don’t want to risk it.

This was not always the case. We started out with a pretty straightforward gameplay system, hoping players would play along with it. However, people either game it or got frustrated by it. When we tweaked it so it was much less clear how it worked, it sparked player’s imaginations and it was much more enjoyable.

Alex isn’t looking so good.

Lesson 8: Players need a role

All stories are driven by the characters that are contained within it, and how a plot plays out is determined by the characteristics of these characters. Just imagine how different Jurassic Park would be if the annoying lawyer guy was replaced by Judge Dredd! So, in order to get the most of any narrative, it is crucial to establish roles.

Games are no different. The role that a player inhabits will determine what actions they have at their disposal, what their goals ore, and so forth. Knowing the character is a vital component in order for the player to be an active part of the story.

Yet this is one of those components that many horror games forget. You are often thrust into a story as some generic character. Often the thought behind this is that the player would “play as themselves”, but this is not how any narrative really works. In order to properly parse a story situation, you need to understand what kind of person is dealing with it.

Say that you come across a corpse. You are playing as Sherlock Holmes, a corpse means a case! You will want to search for clues and try to solve the mystery of how this person died.

Now imagine you’re playing as a flesh-eating ghoul. Now the same corpse is suddenly dinner – yum!

An alternate universe where Daniel is turned into a ghoul. Bon appetit!

In most areas, horror games are well beyond your average game in terms of narrative. But for some reason, a large portion of horror games just fail to set the player role properly. It’s strange, relying on a narrative backbone, yet losing so much of the atmosphere by not defining the player role.

Another big reason for defining roles is that it can help with some of the issues addressed earlier. For instance, it can limit the number of actions the player feels is rational to take. For example Penumbra’s protagonist Philip is a physics teacher, so while he could perhaps fight some demon dogs, it would be more logical to run and hide from aggressive humanoids.

This lesson we clearly learned in SOMA. At first we thought about having a non-speaking Simon with very little character. However, this made player distance themselves from the events. Things got a lot more personal when they played as a character who was reacting to what was happening. While players previously wouldn’t ponder the strange events in-depth, Simon pushing them in the right direction it worked much better.

Lesson 9: Agency is crucial

When I talk about agency, I’m not talking about the CIA. What I mean is agency of the free will kind. A game that has a lot of agency lets the players make decisions and feel like an active part of the narrative.

This is closely tied to the previous lesson. Not only do we want to give players a role, we also want them to own that role. They need to feel like they really inhabit the character they are supposed to play. A game can achieve a lot by combining agency with keeping things vague – and letting players decide to take uncertain decisions.

Say that you are faced with a dark tunnel – dark tunnels are pretty scary!

Now imagine that the game explicitly tells you that your goal lies beyond the tunnel. There’s no choice, you gotta go in. And if the game forces you do something, it will also make sure you do actually have the means to complete this quest – in this case get to the other side of the tunnel.

What’s the worst that could happen? 🙂

But what if entering this dark tunnel was voluntary, or at least presented as such? The game vaguely tells you that there might be something important there – but you don’t know, and might also be a certain death. All of a sudden the tunnel feels a lot less safe. By adding agency and making entering the tunnel an uncertain choice, all sorts of doubts pop up in the player’s mind.

There’s also a number of other ways to add agency. Say the player needs to do something unnerving, like Amnesia’s Daniel drilling into a corpse to get blood out. In the game it is clear that there is no other option. Overall reactions to this was not very strong.

Just petting a guy’s head while the drill drills a hole in it.

Compare this to similar moments in SOMA, where intended course of action is much less clear. Here players are forced to actually think through what they need to do, and get emotionally involved in the process of it.

While SOMA did do this part better, it also had its shortcomings. In Amnesia: The Dark Descent, the game was divided into hub maps, so there was no one path or right order to do things. These choices increased anxiety. Whereas maps in SOMA were way more streamlined, and we noticed a considerable drop in scariness due to this.

In closing

And them’s the rules! As said before, these are not the only ones, but I believe these come out on top when listing the most important ones. You could also go into them with a lot more depth, but I wanted to keep this blog concise. A lot of my previous blogs in the design tag dive deeper into related subjects.

Finally, I want to close by saying that, because of all these special requirements for horror games, I don’t think you can approach them like other games. Instead of “finding the fun” and iteratively building upon that, horror game design needs to start with some strong principles.

When designing a horror game, you want to hone into what you’ve chosen as your core principles, be it atmosphere, theme, or something else. Then, as you progress in development, you don’t want to evaluate the game on how “fun” or “nice” it is to play – but in how well it fulfills its set core principles. And a cornerstone for being able to do that evaluation is to keep the above lessons in mind.

This in itself is a huge topic of its own, and will need to be dealt with in some future post. Stay tuned for more!

3 Replies to “9 Years, 9 Lessons on Horror”

  1. I agree with written above. However, I must also add some of my own thoughts.
    For some reason, Amnesia felt like a much scarier game than Soma. I believe that to be from variety of factors, like a setting for example. Something technological and futuristic is not as scary as something old, ancient, ghostly.
    I never actually finished Soma because it felt like there is not much to do, so I stopped (in Amnesia at least there are tinder and oil – limited resources that I need to search for and manage). As for Amnesia, I finished it and really, really enjoyed it. Such a great game! As you mentioned above, there are some shortcomings at the end, but it was still great. I would actually add one thing- Amnesia felt much scarier until I lost my life few times and noticed that not only I lose NOTHING by losing life, but the game even becomes easier. I don’t think checkpoint system would work in Amnesia, but in other games (like some FPS games) I really feel fear when I play for some time, and am about to lose my life and ALL my progress since the last checkpoint. This can add to fear significantly, but if I lose my life too many times it becomes frustrating and I end up rage-quitting.
    I hope this can help at least a little in developing your next games, wish you all luck in developing them! I am very excited about them.

  2. Lesson 9 should also be applied to open world games. It’s much more satisfying to discover things on your own instead of blindly following the quest marker. Sadly developers nowadays are obsessed with the size of their worlds wich makes them impossible to navigate otherwise.

  3. Toally agree with Milos, yet I’d like to make some supplement about lesson 5–from the perspective of a non-English speaker.
    I think SOMA has done a much better job in narrative than ATDD. This is partly because the plot is more clear, but the main reason for me is that the localization of SOMA is done decently. Back at ATDD, the incomplete localization results in most players in my country have no idea about the plot (and the short story collection is not translated at all), but this doesn’t stop ATDD from being considered the most scary frictional game in the community. I’ve asked my friends why they were scared by a game they couldn’t understand, they said it was the unpleasant experience of playing hide and seek with grunts and brutes. So narrative can be easily destroyed by poor localizations, but an unpleasant feeling can be delivered accurately without language. I really agree with lesson 1 and lesson 4, just keep the atmosphere tense as long as possible, and reduce the players’ chances to take a break.
    BTW, an official simplified chinese of Amnesia Rebirth would still be very appreciated: )

Comments are closed.