Thursday, November 01, 2012

Gasp! Scary!

Now that Halloween is safely tucked behind us, we can get on with the true nightmare season: Christmas. To that end, I'm going to talk about scary games! Specifically, I'm going to design a scary game using some basic principles and techniques that we'll discuss.

First thing first, my pet theory about scary games is that they operate on a steady diet of information imbalance. However, even if you don't buy that line, the techniques included here are used frequently and pretty well tested.

Alright, from that perspective, let's talk about what makes a game scary!

Normally, people talk about setting, monster, hero. I personally feel that's not really a very good way to approach game design, but let's hit on some of the basics.

Dark and foggy, you can hear the monsters long before you can see them. Who knows where they'll jump out from! 

This is a classic example of a very basic philosophy, and one followed by every kind of scary game. Even non-scary games have scary enemies by having them follow these rules even in pretty friendly, brightly-lit surroundings (for example, the hands that reach out from the ground and drag Link back to the dungeon entrance).

I view this as a kind of information imbalance. The setting says that we're in danger, but doesn't give us any precise timing or magnitude. In an action game like Half Life 2, rooms/areas are generally self-contained. When you approach one, you understand that you're likely to face a certain kind of conflict and with a certain difficulty. You can often see enemies before they see you, and, of course, abuse of quicksaves and replays give you even more solid information about the enemy.

On the other hand, a proper horror game tells you that you're in danger... but never tells you how much danger, how soon. The sound of a gently hooting monkey tells you that one of those damn cryomonkeys is around here somewhere, but it's a very tense time hunting for it. The bank of large windows tells you that zombie dogs or birds are likely to burst through, but not when, nor how many. Rooms and areas frequently have no role in delineating challenges - challenges frequently start halfway through a room, or when you're just about to switch rooms.

In other words, the beating heart of the "dark, foggy, hear-em-before-you-see-em" setting is the setting telling you that there are enemies around... but not telling you when or where or how many. This isn't some exact science. If you can hear an enemy in the darkness, you can usually tell how many there are, and perhaps even a vague idea of where they are. If you step into a graveyard, you can be assured that they'll come up from the graves, so you know where they are... but not when, nor how many.

This is the art of giving the player just enough information to feel threatened, but not enough information to actually do anything concrete about it. The best the player can do is hug the walls, creep along, and keep their guard up.

Trapped and powerless, you fight overwhelming odds.

We talk about this when considering setting/hero design, but in honesty this is a mix-up. This conflates a movie setting design concern and a game setting design concern, even though they are actually two different concerns both served by a similar design principle.

In a movie, isolating a weak hero is a matter of making sense. If a killer started stalking me, I'd just pop over to the police station. So you have to explain why the hero doesn't do that. There are more ways to do it than simple isolation. Fight Club has some very scary moments where it becomes obvious that the police are in on it, and Terminator has some scary moments where it becomes obvious the police can't help. Neither features isolation as such, but they both solve the same basic problem: they both deny the hero the chance to get out of the situation.

In a game, that's not quite as pressing. Instead, the focus is on the level of power the hero has over his surroundings, rather than whether the hero can get out of trouble. Protagonists in video games are extremely active, aggressive, "solution-focused" individuals because that's the nature of playing a video game. Therefore, the setting is less about trapping them and more about depowering them.

Scary game heroes tend to move around a lot. In fact, movement is often the primary method a horror game player expresses himself. A game like Battlefield and a scary game might both feature moving, shooting, collecting stuff... but which elements are expressive will be different. A player in Battlefield expresses himself by how he shoots and strafes and takes cover. A player in Resident Evil expresses himself in how he moves from location to location, desperately gathering tiny bits of strength.

This is one reason why the "safe zone" plays radically different roles in games and movies. In movies, the safe zone exists so that the monster can eventually break in and make things horrifying. In games, the safe zone is a part of the player's ability to express himself. If you remove or damage the safe zone, the player loses the anchor of their expressive play. Moving from place to place matters far less if all places are equally unsafe!

Anyway, the game side of the design - helping the player express himself in how he travels - means that scary games need to have a very nuanced travel system. You can travel within a room - for example, skirting around the other side of a kitchen island, or sticking to the top of a rafter so the milling zombies won't get you. In that case, travel is tactical. You're choosing a path based on how well you can keep up your guard and whether there's any loot to be had over there.

You can travel between rooms - walking through gateways, opening doors, etc. This is strategic play, where you are deciding whether the timing for that room is right, or whether to hold off and explore other, safer areas and gather more power first, or maybe whether to return to your safe room... just in case.

You can also travel between zones, such as changing from the house to the back yard, or from the factory to the underground labyrinth. The biggest difference between zones is the chunking factor. Each zone essentially resets your strategic options, since you have to rediscover the safe rooms, map out the connectivity, and judge each room for safety all over again. Backtracking is typically not scary unless the zone has significantly changed since the last time you explored it... because you've already explored it. You've already done the sweaty-palms risk analysis.

As to my pet "information imbalance" theory, exploration of the areas is about gathering what information you can and trying to plot a safest possible course through it. There are many times where you'd prefer to tackle a dangerous room over one that looks safe, just because you can clearly see all the dangers in the dangerous room and the safe room has some information still hidden away.

Winning or losing? It's always losing.

In most scary games, a big part of the game is the idea that fighting an enemy is always a loss. Even if you win, you've expended valuable resources. This leads to radically optimizing your playthrough. For example, mastering the melee weapon so you can kill the minor enemies without losing any resources (although with a risk of getting injured).

One of the fundamental principles you'll always hear bandied about is that you need to have a powerless hero. However, I don't believe that to be true for a video game. A movie, sure. But in a video game, rather than the power level of the hero, the dynamics of conflict are more important.

Basically, a scary game's fights are always losses. There is no way to win a fight such that you come out of the other end better than you went in. Even boss fights - the reward isn't more power. It's just an open door. This is in direct opposition to action games, where the enemies drop ammo and health and experience points.

You gain power by scrounging and scavenging from your surroundings. You spend power on the enemies you fight. That's the scary game dynamic. The core of this is the tension - not knowing whether you'll be able to scrounge up enough to get past the enemies.

Risk, pressure, and regret.

A common line of thinking is that there's got to be risk in order to be scary. There's certainly some of that, yeah. But you need to be clear about whether you're talking about pass/fail risk (dying, failing the mission), story risk (can you defeat the oogie-boogie man?), or attrition risk (can you keep things from getting worse?).

The scariest games are scary because of attrition risk leading to pass/fail risk. A game where you have to restart the mission if you're noticed isn't going to be scary, it's going to be a mechanical aptitude test where you replay the mission over and over.

Instead, what's scary is when things get worse. If the player has unlimited save/load capability, this attrition needs to be quite slow so that he doesn't feel any given conflict went badly enough to make him reload. If the player doesn't have much save/load capability, it doesn't have to be as slow.

Things getting worse include mostly losing health, ammo, and maybe sanity points. It is possible to add in a lot of other "getting worse" factors if your game is going to be unusual, but in general attrition simply makes the next wave of enemies feel more dangerous because you are closer to being unable to deal with them.

Of course, the balance here is quite difficult. What one player may find a grueling last-ditch effort to squeak by a zombie using your last bullet, another player might be able to cruise by with just their fists. Nothing says this clearer than Thief, where getting in melee combat was supposed to be a knuckle-biting "you will die!" experience... but some players are good enough they can just gut everyone they see.

Because this is so hard to balance, a lot of designers lean heavily on story beats. They introduce "story risk" like "the house is on fire! Run!" Story risk will resolve the same way for every player, so it's a known level of pressure. Of course, that known level of pressure is much, much lower than the pressure of a properly balanced attrition situation.

Often, the story beats are also used to 'reset' the attrition-based power level, essentially capping how absurdly overpowered a skilled player can get. Another big limit is inventory management, which keeps really skilled players from collecting a massive stockpile of resources.

Some story beats feature a choice where you can do either A or B. We still stick with this idea as if it were actually a tense situation, but 99% of the time it's pretty boring. In order to make it tense, you have to do two things.

1) You have to make the choice actually hard. Not simply the same choice over and over again. IE, not "good vs evil" or any other 'A vs B' that repeats. Instead, each decision must exercise a different kind of choice. Good vs evil. Honor vs glory. Love vs good. Health vs authority. Whatever you can come up with.

2) The choice has to matter a lot, for a long time. The player has to regret that they didn't make the opposite choice. That way, when the next choice comes along, they are more crippled by it.

Screwing with the player's expression: the mastery of pacing.

The player expresses himself by how he travels in the game. This is where a player's delicate sense of risk, pressure, skill, and reward come together to express the player's personality. It is the paintbrush with which the player paints.

Therefore, it's also a major part of the scary.

Enemies are scary, sure. Especially when there is no way to win, just ways to lose less badly. But enemies are adrenaline-scary, not the kind of creeping, continual terror you feel when properly put into a scary situation.

You can get that by screwing with the player's primary means of expression. How do you screw with the player's travelling? Well, you can do it by brute force or with incentives.

A brute force example is when you can't travel freely. The floor collapses from beneath you. Enemies block your path out. You can't move over to where the enemies are, but they can still strike at you. You can also do it with some level of subtlety - fires you take damage from if you get too close, water that slows you to a crawl, ladders that leave you exposed while you climb...

Incentives are when there's no fundamental issue with the level, but their choices are suddenly much more difficult. For example, if you hear a cryokinetic monkey hooting, you'll want to hunt it down without exposing yourself to potential blasts. If there's a monster hunting you down, you'll want to run perfectly without getting caught. If there's an ammo cache on the other end of the room, you'll want to go there even though it's a wide open space...

Both brute force and incentive systems require a sense of pacing. If every floor falls out from under you, you're going to start treating it as a system-level rule and incorporating it into your basic play. So enough time has to pass between floors falling out from under you that you aren't worried about that particular problem any more.

Scary Challenges

A lot of scary games revolve around "how do I do this thing" puzzles. The most famous are Resident-Evil-style "puzzles" like "find the octagonal handle for the cupboard inside the demon statue's left foot, then gather up some slug resin so you can combine it with a banana and glue it into the cupboard so you can open the cupboard and get a triangular key".

These puzzles are derided as being very silly, but their opacity is on purpose. If you had a challenge where you knew the solution and were making concrete progress towards it, you're not going to be as tense. You know where the end is, you can measure your exact progress... not as tense.

These puzzles are everywhere, but usually they aren't quite so silly. For example, having to swing a catwalk around so you can cross a gap is the same kind of puzzle. It's not completely opaque, but you have to figure out that it's even possible.

These challenges are most brutal when you add time pressure in. For example, having to figure out how to rotate some statues while a boss monster continually chases you around. These will frequently result in death, so carefully play test them for balance.


There's a lot of talk about NPCs and their role in horror games. Some scary games extensively feature NPCs in both major and minor roles. In others, you never talk to a single living soul.

One of the problems with NPCs is that they are generally reassuring. If there's someone to help you, then you have a safety net. This is why, in general, NPCs in horror games are either evil or burdensome.

Putting aside evil NPCs, burdensome NPCs are a mixed bag.

Having something to protect is very powerful... but you have to want to protect it! Frequently, burdensome NPCs are just plain annoying, to the point where you start to hate them.

You could theoretically fix this by making the NPCs much more intelligent (AI-wise), but this would fundamentally be making them less burdensome, and it would take a lot of effort.

Two easier solutions are to make the NPC remote, or to make the player choose to take them. If the NPC is remote, you can simply hook triggers into the game world. If X time passes, they get injured. If you collect Y prisms, they give you a better gun. Remote NPCs don't have any AI worth speaking of, and basically serve to put a human face on in-game challenges.

Making the player choose to take NPCs requires an adaptive/open world, since the player might choose differently at any given moment. A player that chooses to do something will generally be a lot more accepting of the difficulties involved than a player who is simply forced to do it. Choosing to do an escort mission is a lot more acceptable than being forced to... even if the player is eventually more or less forced to choose to do an escort mission. The act of opting in is powerful, even if there's really no way to opt out.

Fumbles and bad controls.

A lot of scary games have very bad controls. The reason for this is fourfold.

Firstly, very bad controls are a source of weakness, making the hero less capable of dealing with the enemies. Since dealing with the enemies is always supposed to be a losing proposition, bad controls cement the deal.

Secondly, bad controls are equalizers. Very skilled players shine with responsive, tight controls. While skilled players will still be better than unskilled players with clumsy controls, the difference will not be as large.

Thirdly, bad controls respond very aggressively to player panic. If a player is scared, their ability to use your shitty control scheme drops much faster than if your control scheme was really great. Therefore, if the player is scared, they will get even more scared because now they are shit.

Fourthly, it radically increases the pressure by making it harder to gather and react to information. Awkward cameras and controls don't just make combat harder - they make it slower and more difficult to gather and react to all sorts of information, including things like room layout, hidden items, NPC interaction, inventory management...

How is all that information imbalance?

I've explained a few of the above things as being based on information imbalance, but others I've left unexplained. And, of course, my pet theory is just my pet theory, so I didn't want it to intrude much.

Some of the things above - such as equalizing skill between players - aren't based on information imbalance. I don't think information imbalance is every little thing that could ever be scary. But I do feel that information imbalance is fundamentally scary, or least very high-pressure.

For example, when you are playing poker, there's a desperate push to read your opponent's faces, to count cards, to control your own expressions. The pressure of playing poker, even just with free chips, is tangible. And it's all thanks to the information imbalance. The high-level play of poker is almost entirely about gaining the edge in information.

There are a lot of different kinds of information imbalance. Let's mention some of them.

Information gaps are when you are given some information, but there are gaps. These gaps create pressure. For example, if you hear a hooting cryomonkey, you know that there is one... but not exactly where, or when you'll run into him. If you look into a room and see a cupboard, you know you can open it and there might be stuff inside. But you don't know how much stuff, or whether there's going to be a trap along the way.

The gaps create pressure not just to fill them in, but also to nullify them completely. If you hear a monkey hooting, you can try to fill in the gaps by hunting down the monkey on your own terms... or by skirting the area and blitzing through, trying to make it so that the encounter never happens at all. Similarly, if you suspect there is a trap in a room, you might try to navigate such that it will trigger when you are in the perfect position to weather it, rather than stumbling in as the designers intend.

Information scavenging is a primary method to try to fill in gaps or decide to go around. This involves skirting the edge of the situation looking for insights before approaching. It's valuable both as tension and as pacing.

Information delays are when you have information, but either too far in advance or not far enough in advance to be comfortable with it. This isn't the same as information gaps, because you have more or less complete information. An easy example of this is when monsters burst through a window. Yeah, you can see them. You know what they are, where they are. But can you act on that information rapidly enough?

You can go the other way, too. If you know right away that the final boss of the factory is Glouberk and he's in the manager's office, there can be a lot of pressure as the knowledge builds. You discover fragments of information that make Glouberk more menacing. You see the manager's office window from the factory floor and it's covered in goop... there's no surprise - eventually, you walk in and fight Glouberk. But even though there's no surprise, the delay is uncomfortably long, long enough for the pressure to keep building up. This only really works if you can keep flicking the player with reminders and additional bits of pressure - otherwise, they'll forget all about it pretty quick.

Bleed control is when you are the information source, and you are trying to limit how much you reveal, or perhaps even distribute misinformation. It also may involve trying to goad others into revealing more than they want to.

Chance is when you know roughly how likely something is, but you don't know how the dice will land. For example, you can release a chandelier and it'll crush anyone beneath it. You decide to use it on a heavyweight enemy in the room... but it's difficult to line up. There's not really any missing information, here. There's just a chance you'll hit them, and a chance you won't.

Chance is a bit iffy, because it adds a lot of chaos into what is typically a very carefully balanced experience. However, if you're making a less carefully balanced game, chance can play a very good role. Just remember: this is about having all the information and choosing to gamble, not about randomly getting screwed over.

Information pressure is when you have information, but if you don't act on it, it will either lose value or things will get worse. Pressure doesn't necessarily mean time pressure, although that is the most common.

Information swamping is when you have too much information and the challenge is to figure out which bits matter.


So, if we take the information imbalance thing as a solid idea, can we come up with any other game mechanics that might be scary and aren't on that list?

Well, the format of modern scary games is such that some obvious mechanics aren't used. Chance, information pressure, and information swamping are all not very common in scary games, because scary games are fundamentally tightly scripted.

If we move to an unscripted scary game, we can radically amp up the roles these things play. For example, the game could revolve around high-stakes betting. It's a very popular subgenre in anime these days, sometimes revolving around a specific game (mahjong) or a random assortment of anything-goes high-stakes betting.

We could set our game in a town or school. Something high-pressure has happened - the town has been flung into a new dimension, your school grades depend on bids, etc. The result is an open-world game where all the NPCs have challenges you can accept.

In order to make it scary, we don't have to make it foggy, and we don't have to make traveling our primary mechanic. Instead, we can create the feeling of pressure through the less commonly used mechanics.

A big part of making it scary is to make the world wear down on your resources, and make the resources only gainable through betting. So there is a heavy attrition as the world gets more and more difficult to deal with. NPCs change as time goes on - the things they can provide, the things they are willing to do, the gambles they offer. The fear is, on the surface, because of the human flaws of these people who don't always cooperate to survive. At a deeper, mechanical level, the fear is because of the need to constantly make dangerous gambles to survive another day.

So the core play is deciding which gambles to take in which order. The resources you gain from gambles might be resources required to survive the world's attrition, but sometimes can be used to make other gambles easier or clearer. For example, if you win an easy gamble with someone for the right to have them cooperate with you, then you'll have an easier time with a poker game because they are signalling you while looking at your opponent's hands.

The game is about taking risks. To a lesser extent, it is about gathering enough information about any given risk to know what the risk actually is. Also, whether to expend valuable resources on making the risk less dangerous.

The game is about an information flood. You have to figure out which of the townsfolk are offering what kinds of deals, and glean out which gambles are worth taking and in what order.

The game is about information pressure. If you don't take a gamble soon, it's going to change as things get worse. David won't always be offering a simple high-low card pull wagering his help vs a can of beans. Maybe the gambles he offers will get more dangerous. Maybe less.

And of course, the game is about how the NPCs react to the grinding attrition of the world. This will vary not just by NPC, but by the various things you might do to help them (probably in gambles). For example, it may be that losing a gamble will allow an NPC to survive another day...

That game could be pretty scary.

It could also be tense but not scary if you allow for saving/loading or make the world's attrition level too low or make the NPCs too cooperative.


Random_Phobosis said...

One thing I've always found scary is responsibility, which can be viewed as high level information imbalance.

In System Shock 2, there are lots of things you can do, lots of places you can explore and lots of resources you can use, but any decision is scary, because you don't know if it's the right one. Save/load isn't particularly harmful, because many decisions have long term consequences.

New Xcom is scary as hell with saving disabled. Same thing here: lots of decisions, and any one could potentially turn out devastating.

I guess the longer the need to make hard decisions (or at least illusion of it) persists, the scarier the game.

Craig Perko said...

I agree. That's why saving and loading is such an irritating issue.

In order to make the player have responsibility and also allow the player to save and load, you need to make the responsibilities evolve very slowly, rather than being simple either/or choices.

You have to make it so that even if the player loads, it's just a miniscule amount of the total he can redo.