Friday, 23 September 2016

SOMA - One Year Later

It is now one year since we released SOMA which means it is time for an update on the current state of things.

As always let's start with the thing that is of most interest to people: How much has the game sold? This is always a bit of a tricky figure to nail down as it depends a bit on how you want to count. For instance, we were part of the Humble Monthly Bundle this September which caused a lot of people to get the game, but none of these were "direct sales". Instead, we got one big payment for taking part in the deal. For the sake of simplicity, I will simply lump all of these figures together as a whole, which brings us to a total of a bit over 450 000. Or to phrase it differently: almost half a million units sold!

This is quite good, in fact it is so good that we have now broken even and then some! I think it is worth to stress just how great this is. We spent over five years making our, by far, most ambitious game ever. We also spent quite a lot of money on various outsourcing such as voice acting, 3d models and animations. For instance, to make sure we got it right, we actually recorded a lot of the game's dialog three times. In the past we have just recorded voices at the end of the project and hoped for the best. With SOMA we knew that nailing the voice acting would be crucial, and spent money accordingly. In the end, it meant that around half of our voice recordings were never used. The same thing was true for things like models and animations. We ordered a ton of these and as design changed many of them didn't make it into the final game. On top of that we also spent a lot on making live action clips for PR purposes. Taken together with salaries and all other kinds of expenses, SOMA cost quite a bit to make - well over 10 times what Amnesia: The Dark Descent cost us.

It is important to understand that SOMA was far from a safe bet. While we had the luxury of having already made a successful horror game, SOMA was not an easy sell. The game relies heavily on getting certain themes across to the player, and communicating this proved to be a hard task indeed. When showcasing Amnesia we could just show how you blocked a door with some rubble and hid in a closet and the game's core experience was neatly summarized. But with SOMA things were way harder. First of all, weaponless horror games are no longer anything special and by no means a stand-out feature. In fact, the "chased by monsters"-gameplay was not even a core part of the SOMA-experience. The whole idea with the game was to give the player a first person perspective on a variety of disturbing philosophical musings. To make matters worse any concrete gameplay example of this would be riddled with spoilers, so all discussion had to be made in an obscure "you'll understand when you play it"-fashion.

We were also constantly worried about a backlash based on faulty expectations. Early on we realized that SOMA was never gonna be the scare-fest that Amnesia: TDD was. But because we felt other aspects of what made the game special were so fuzzy, making "from the creators of Amnesia" a big part of the PR campaign felt crucial. The problem with this was of course that this might set up expectations for SOMA being a direct follow-up from Amnesia - with everything that that would imply. We tried to tone it down and make it clear what sort of game SOMA was, but there was still a noticeable negative effect. For instance, many reviews start by saying "Well, it's not as scary as Amnesia" or similar.

Despite a bloated budget and tough sell, here we are a year later having earned back every single dime spent. And not only that; we earned well past the break-even point! The project was a big success and we are able to keep doing games with scope and quality comparable to SOMA. In fact, our goal is to aim higher still.

It is also interesting to compare SOMA to Amnesia. As can be read here Amnesia: TDD sold about 390 000 units a year after release, but worth noting that that was for PC only. SOMA's 450 000 units come from PC and PS4 combined. However, many of the Amnesia units were sold during 75% off sales, a discount rate we have not really had with SOMA yet. On top of this, SOMA also costs 30 dollars compared to Amnesia TDD's 20 dollars. So even just counting the PC sales the total income is higher for SOMA the first year compared to those of Amnesia. And when you add the PS4 sales on top of that, it is clear that, in actual earnings, SOMA has far outsold what Amnesia: TDD did during the same period.

Another thing worth bringing up are the user reactions. Our current MetaCritic score is at 84 and will probably stay like that. While this is a really nice score, what has really blown us away is the user reviews. You hear a lot of people complaining about the Steam reviews and how they get sad when they read them. But for us it is the other way around. Whenever I feel a bit down, I actually go and read some steam reviews and instantly feel better. I mean, even user reviews that have given the game a thumbs down contain stuff like this:
"Amazing game, [...] This game literally changed how i view games. it had an amazing story, the atmosphere was spot on, there wasn't a moment where i thought i was safe and the pacing of the game was magnificent."
And it is really hard to not feel good when even the refund notes contain nuggets like this:
"I love horror. Soma is distressing. There is a scene where I have to hurt an innocent robot to progress and I don't know why. It made me cry."
Currently our Steam reviews have 98% positive calculated short term and 95% counting the total. This makes SOMA the most well-liked game we have ever made. And when people say they didn't like it, it is almost always because of the monster encounters - a non-core part of the experience. When it came to the narrative bits very few people disliked it, which is a wonderful surprise to me as this was by far the most uncertain element. My fear was always that a lot of people would think there were not enough horror and monsters, and the opposite turned out to be true.

While talking about user reactions it is worthwhile mentioning all the great discourse around the game. For instance, the SOMA subreddit is still fairly active, and new, interesting subjects pop up all the time, like this discussion on the future of the Ark. These youtube videos that deep dives into the story are also great, and it is fantastic to see people giving so much thought on our work. There is a load of other user content like this and it's honestly quite overwhelming.

Finally, I want to briefly go over where Frictional is currently at. As I said last time, our goal now is to be a two-project studio and so far it is going really well. One project, which most of the team is working on, is going to start production at end of the year and the other project is mid-way through the R&D stage. Unfortunately I cannot divulge any specific information on these two, and it will be a little while before there will be a proper announcement. However, we do have some smaller, cool stuff in store, one of each we will announced later this year. If all goes well, we should also have another thing for early next year.

So exciting things happen both in the short and long term, and I am really excited for the future of the company.

Monday, 18 July 2016

Thoughts On Inside and Playable Stories

It is easy to view Inside (2016) as "just" a puzzle game with nice graphics. Many seem to overlook how incredibly good the game is at letting you play a story. Inside manages to craft an engaging narrative through gameplay in a way that few other games do. I think this is quite the accomplishment, and I've seen way too little written about it. The playable story is what I love about Inside, so it's what this essay will be all about.

To start things off, it's important to explain what this narrative aspect of Inside is all about. It's very easy to only see narrative in very concrete things like dialogue, the text you read, or in film-like cutscenes. However, when talking about stories in games, we have to stop thinking like this. Instead, we should consider what sort of narrative we experience as we actually play. After all, that's what interactive storytelling is all about.

To make this distinction clear, let's compare Inside to another puzzle platformer. For this discussion I'll pick Unravel (2016) since it's recently released and has many similar aspects to Inside. I don't aim to prove that Unravel is a bad game by any means - I liked it quite a bit myself. It just lacks a number of the narrative elements that Inside has, and by comparing the two, it should be clearer what it is that Inside does so successfully.

Instead of relying on a connecting story to tie things together, Unravel is built more around "flavored mechanics". You have a range of objects that you can interact with and all of these are somehow connected to the environment you are in. For instance, in a level taking place in a garden you can push apples around and use them to get up to otherwise unreachable places. But while these things are nicely woven into the graphical style of the level, there isn't really any narrative connected to them and they don't tell you anything about the game's story when interacted with. Neither are these objects, or your interactions with them, an important part of the narrative that unfolds as you play. They are simply there to serve the central mechanics of the game.

Inside (and Playdead's earlier game Limbo (2010) for that matter) is very different in this regard. The various objects that you interact with are not only part of the world visually, they are key parts of the narrative as well. Interactions that you take part in at one point in the game will be of great importance to later events.

To make this clearer, let's compare how both games start out:

Unravel (see clip here):
Yarny exits a house and enters a garden. He  is surprised by a couple of butterflies, and then runs out of yarn. He lassos a length of string so it gets stuck on a sundial, and by tugging the rope he manages to tip it over. This lets him gain access to more yarn and he can continue moving forward. A box of flowers is in the way and he climbs over it. Two buckets now stand in his way, and by using some tricks with his yarn he manages to get past those as well. There are more obstacles in his path, and by throwing out a piece of string and letting it attach to some glowing objects, he is able to swing across them. He eventually gets to a tree stump that can only be got past by throwing out strings of yarn and swinging forward.

And the game continues like this. It should be pretty evident how, story-wise, nothing really interesting happens. The game simply presents a sequence of obstacles for the player to get past. Also note that there are a bunch of things that doesn't make a lot of sense. For instance, why are all the conveniently placed pieces of yarn spread out across the world? And why does Yarny simply not just run around a lot of these objects? In terms of playable story, this game is quite weak.

Again, let me make it totally clear that this does not make Unravel weak as a whole, nor does it mean the game is totally devoid of story. For instance, Unravel has a lot of thematic connections between the photos you can find and the art of the levels. But this is also why Unravel is so interesting to compare to Inside. For while Unravel has story in it, almost none of it is playable. And focusing on playable storytelling is what Inside does best, and what this essay is all about.

Inside (see clip here):
The boy jumps down from some rocks and enters a dark forest. He comes across a fallen tree lying across a chasm and carefully balances across it. Further down the path he comes to a blockage with barbed wire and climbs through it. In the distance a truck can be seen and as the boy comes closer he starts sneaking. It is clear that the boy doesn't want to get spotted. Inside the truck he gets a quick glimpse of some weird figures, but before he can make out what they are a man shuts the door and the truck drives off. 
The boy travels further into the forest; mysterious metallic pods are scattered in the background. One of the pods has a light turned on, and next to it stand two men. As the boy gets near, the men hear him and turn on a flashlight. The boy must now hide behind a rock so the men don't spot him. Eventually they leave and he can continue. He now comes across a long wall with barbed wire on top of it. Someone is obviously trying to keep people out. Various junk is scattered around the wall, and by pushing an old fridge up next to the wall, the boy manages to jump across.

And, for most of the time, this is how Inside continues. Unlike Unravel, there're a lot of interesting story events in this sequence. The protagonist is also an active part of the story, climbing past two obstacles, both of which make it very clear that we are entering a restricted area. So not only do these small puzzles give us an idea of the story (that we are closing in on a place we are not supposed to enter), but by being the one that manages to get past them, the player also becomes the one that moves the story forward.

On a purely mechanical level, there isn't a lot of difference between Inside and Unravel, but the framing of their mechanics is vastly different. From the very moment the game starts, Inside paints a vivid narrative and makes the gameplay a core part in it.

Here are some other examples how Inside manages to craft a playable narrative:

At one point the you come to a platform near a large body of water. A man arrives in a submarine and as he climbs up a ladder, you need to hide. Then at the right moment, when he is distracted talking to a college, you can sneak past and steal the submarine. All of this is framed as a puzzle, so the player must be the one that realizes the man is a threat and that the submarine can work as a means of escape. It's not a very complicated setup, but it makes a big difference in terms of perceived narrative. If the next level had simply placed the player in the submarine, without any playable section in between, the player would just have experienced a change in gameplay. By making the transition (stealing the submarine) into part of the gameplay, the player gets to be an active participant in the transition of the narrative. The player gets their personal story to tell: "I sneaked past a man and stole his submarine". This is a lot more powerful than simply having something like: "And when the level started, I was in control of a submarine". Inside is great at crafting these sorts of transitions. A submarine ride is not just another type of level - it is a new part of a narratively coherent journey.
Another great example is the pig encounter. This scene starts as the player walks past a seemingly dead pig, only to see it twitch. The pig then rises up and starts chasing the player. From this section the player will learn that there are parasites that can take control of animals, which is a bit of lore that will be of great importance later on. The player learns most of this through play. The pig's aggressiveness is apparent from the need to avoid it. Pulling out the the larva is something the player does through gameplay, and doing so results in the pig becoming friendly and thereby usable for a puzzle. So the two core ingredients in this lore bit - the different behaviors of the animal, and the worm causing it all - are both deeply rooted in the gameplay.
A final example is the imitation puzzle. Here the player finds themselves dumped among crowd of mindless "zombies". These have been controlled to walk in line and are forced to perform various tasks in order to (or so I at least guess) check that the mind control has been properly imposed. The only way for the player to survive is to act along according to orders, effectively becoming a zombie themselves. Again, this also lets the player discover a bunch of story through gameplay. And not only that, you also get to step in the shoes of these zombie people. In all it is a really neat way of bringing the player into the game's world.
One could argue that this isn't anything new. Point-and-click adventure games have had this kind of gameplay for decades. You are faced with a puzzle that is directly connected to the story, and in solving the puzzle you move the narrative forward. But the big difference (and a big problem with adventure games, I think) is that the player lacks a lot of agency. Most adventure games (especially the ones of the point-and-click variety) thrive on constraining the player both in terms of what actions are possible and how they are carried out. In a point and click adventure game the player merely suggests the sort of action that should be taken, and it is then up to the game to determine whether it will allow it or not. For instance, if the player wants to enter a cave, the protagonist might respond that she doesn't feel like it. On top of that, the structure of these games is such that it can be very hard to guess what actions are possible. Actions that seem possible are often not, and actions that were possible in one situation is not in another. Because of all this, I have never felt that I properly played a story in a adventure game. I always felt more like a semi-active observer.

Side note: This is not the only reason why playable stories are problematic in adventure games, but this essay would go on for too long if I were to delve into that. If you are interested in finding out more about this, I have written about it here.

Inside is very different though. In this game you basically have control from the beginning until the very end. There are a few cases where you lose control of your character, but all of these make narrative sense. For instance, you lose control at one point when the character passes out and so on. Inside doesn't even feature any loading screens; from the moment the game starts until the bitter end, it is one continuous narrative. The game also features extremely intuitive controls and has a really good set of affordances. Because of this, you're almost always sure what actions are possible. This gives you a great sense of agency, strengthening the feeling that you are the one moving the story forward. Combined, these two things make a world of difference in how the playable narrative is perceived in Inside compared to a point-and-click adventure game.

Side note: A lot of the lore in Inside is extremely vague and ambiguous, and one could argue that it makes the game's storytelling worse. However, apart from a bit about premise at the end, I won't discuss that in this essay because it doesn't feel very relevant. While there are some instances of vagueness that help the game a great deal design-wise (such as how some levels are connected), for the most part it is not a crucial ingredient. If the developers wanted, there's a lot of lore and background information that could have been much more understandable, without much difficulty development-wise. It's pretty clear to me that the vagueness in Inside is a conscious choice by the developers. Because of this, it feels better to just focus on what Inside does best: storytelling through play.

One thing that makes all of this come together is the adaptive animation and action system. When the boy is in a place where he needs to sneak, he is animated in a way that makes this very evident. Because of this, the game implicitly tells the player what sort of things to expect, and the player can shift their mindset to match that of the boy's. In a similar fashion, the game also changes what actions are possible to performing depending on the current state of the boy and the game world. The best example of this is when you pick up a torch and use it to keep aggressive dogs at bay. The boy will automatically start waving the stick when a dog is near, and this lack of control feels totally OK to the player because it matches what they would like to happen. Partly, this is possible due to just mentioned adaptive animations. The other big factor is the setup of the situation itself and the previous hour or two of training the player has had in expecting actions to subtly change depending on the state they are in. On top of this the game also creates a credible setup that constrains the torch to a small, closed-off area. This makes sure that the player can't get it into another place where it feels like it could be useful, but where the action is not supported by the game.

Another interesting part of Inside's storytelling is how it handles its "cutscenes". Instead of taking away control from the player, these cutscenes take place in the background, allowing the player to keep playing as they occur. This is a pretty common trick in games, but what sets Inside apart is that these background events can often be dangerous if the player is not careful. For instance, if the boy gets spotted by some people in the background they'll chase and capture him. This provides a certain "realness" to the background events, and instead of just being some sort of decor they become a proper part of the game's world.

In order for the active narrative in Inside to work at all, the game obviously needs some gameplay. To achieve this Inside has settled on using puzzles.  Now it's worth asking the question: Why puzzles?What is it about puzzles that make them so suited for a game like Inside? At first this question sounds a bit weird. Inside is a puzzle platformer, so obviously it has puzzles. But that's only if we see inside as "just" a puzzle game. If we instead approach it as a game with a certain story in mind and with the goal to tell it in an active fashion where the player is always in control, then the question isn't as trivial anymore.

There are two main reasons why puzzles are such a good match. One is that they allow the game to implicitly force the player into taking part in certain activities. The other is that puzzles allow the player to make mental plans, which is a crucial factor in making the gameplay feel good.

Using puzzles in order to "trick" the player into doing certain actions is something I've written about before here and here. In summary the core idea is that a puzzle allows the designer to direct the player's behavior by crafting a puzzle in way where the problems, goals and solutions make the player go through with the desired actions. The crucial part is that the player doesn't feel like they are pushed along a trail, the idea is that the player should feel as if they came up with it all on their own. They should feel as if they chose to act in a specific way themselves, even though it was exactly what the designer intended. When this works, you have, for narrative purposes, a great puzzle. Any time it feels like you need to guess what the designer is thinking, or are being handheld along a set course, you've got a bad puzzle.

In order for it all to feel like you're part of the a narrative, it's vital that the sequence you take part in feel story-like. If all you do is solve one sliding puzzle after another, it won't be a very interesting narrative to take part in. You have to have actions that are story-like. This is a big problem when a narrative game relies on combat for the core gameplay. While combat systems allow designers to set up scenes that make the player behave in a certain fashion, there is only so much story you can tell about taking down hostiles.

Puzzles on the other hand are much more versatile as they don't rely on a central mechanic in the same way as combat does. A puzzle is simply a framed problem for the player to solve. It doesn't rely on a set feedback loop or on particular system dynamics. The player just needs to know: the boundaries that are in place, that the problem is solvable, and what goal state to strive for. As a player you are constantly on the lookout for these things, and this works as the core mental play state, in the same way as "locate enemies and neutralize them" does in combat-oriented gameplay. This allows for a game like Inside to set up all sort of story-like sequences and have that be the core of the experience.

The other aspect that makes puzzles such a good way of making playable story is that, if done right, it allows for a lot of planning.  The sort of planning that we are after here is not just grand strategy like you get from a game of Civilization. Rather, the important bit is that the player is able to look ahead and plan their next course of action and then, to a high degree of accuracy, be able to execute those actions. Crucial here is the ability to, by glancing at the environment ahead, forecast what sort actions are possible to perform, and what the result of doing them would be. I wrote about this a bit in an essay about Until Dawn (2015), where it hit me that the inability to plan is a critical reason why many interactive movies don't feel like proper games. This was especially evident in Until Dawn as due to its reliance on multiple characters and a "anyone can die at anytime"-setup, it sometimes allows for quite a bit of planning. And when it does, it has a big effect on how the game feels to play - it suddenly feels much more like "proper" gameplay.

Because of this, if a game aims to have a narrative built around its gameplay, it is crucial that planning is allowed. Puzzles can allow for this, but it doesn't come automatically as it would in a game about combat. Just recall the discussion above on point-and-click adventures; while based around puzzles many of these games lack the feeling of having "proper" gameplay. A fundamental reason for this is something that I've touched upon earlier: point-and-click games often suffer from a lack of consistency. It isn't possible to know ahead of time what you will be able to do and how the world will be affected if you do. These games often rely heavily on constant trial and error as you slowly discover the boundaries for each specific scene. Compare this to a game like Super Mario World (1990). which I think we all can agree has "proper" gameplay. When you come to a new area in this game, you can instantly mentally simulate what courses of actions are possible, even if there are elements you have not seen before. For instance, if a new monster has spikes on, you can be pretty certain that you won't be able to jump on it. Much of the time, point-and-click adventure games just lack this coherency, and the gameplay suffers from it.

Inside, on the other hand, is much better in allowing for planning. When at its best, you can enter a totally new scene and still have a pretty good idea of what options are available to you, what you need to do and get plausible ideas on how to achieve it. This is when the gameplay feels at its best. In other cases a new area can be more opaque (in terms of possible actions) at first. But as you do some simple trial and error you get a grip on how things work, and you can quickly start making plans based on that. This initial trial and error is impossible to get away from entirely, even games like Super Mario have it, and when used correctly it can be for the better. Figuring out how things work is a fun thing to do and for a game like Inside, plays a crucial part in the experience. Problems tend to arise when this trial-and-error becomes an integral part of the moment-to-moment gameplay, or when it interferes with the control of the player character. For most of the time, Inside avoids these issues and maintains the ability to mentally plan ahead and execute your actions. This is central to what makes the game fun to play.

When the story-connected puzzles and the ability to plan comes together, Inside works amazingly well and it feels like you are playing a story. This is the part of the game that I absolutely love, and there are way too few games that manage to do this,

However, things are far from perfect in Inside. While the game is filled with moments of utter brilliance, there are also plenty of times when things don't work. At these times you are pulled out of the game's world and narrative, and the game feels much more like your standard puzzle game. These moments are well worth discussion as it shows what things to look out for, and suggests ways in which we can take playable stories further.

Despite being a fairly short game (around 4 hours) Inside hasn't manage get rid of all the filler material. Time and time again the game throws you a puzzle that is similar to one that you have already completed and that, more crucially, doesn't provide anything interesting story-wise. The reason these parts exist is fairly straightforward: coming up with new and clever puzzles that fit the story is really hard and time-consuming. I doubt Playdead added these sections because they thought they were perfect as is. Instead, they were probably just the, at time, the best and simplest ways to solve certain issues like pacing, gameplay set-up for tutorial-purposes, and so on. Whatever their reasons for being there, they stick out like sore thumbs and degrade the story-like feel of the experience. If you want to make good narrative through gameplay, it's crucial to minimize moments like these as much as possible.

This is far from a simple problem, of course. It's also something I've written about in the past here. The difficulty of building story-connected puzzles stems from the fact that there are so many different threads that need to fit together into a coherent whole. One approach to tackling this problem is to break it into many smaller, more manageable parts. My own suggestion for a solution is based around having a framework that helps you build your moments in a layered fashion. This approach, which was developed together with Adrian Chmielarz, is called called 4-layers. It's just one possible way of solving this problem, and there are bound to be other ways to address it. Whatever the solution is, it's important that any filler material is removed or changed, so you constantly have a feeling of being inside a story.

For most part the world of Inside has a sense of reality and coherence to it. The buildings, machines, and vistas all have an otherworldly feeling to them, but it still feels like a proper place and you could imagine people actually inhabiting the world. Crafting a world like this is key to making a good playable story. The goal should be for the player to mentally represent the game's environments, characters and objects, not just as thin fa├žade, but as actual things. The player should imagine not just what is on screen, but what kind of things might lie beyond, and to consider it all as one big world. It's when the player start thinking like this that they stop being an observer of the story, and instead become a part of it.

In order to pull this off, it must be possible to take the game seriously. For instance, if the player comes across a hatch, they should be allowed to wonder: "why is this hatch here?" and there should be some sort of plausible explanation to be found. Because if the player can't do this, the game stops taking place in a living, breathing world, and instead is degraded to a simplistic play space.  No matter how fancy the graphics and animation are, in the end it is the player's imagination that brings it all to life.

It's easy to believe that we take in reality "as-is" and what we perceive and feel is this unfiltered flow of information from our senses. This is far from the truth. First of all, our sensory organs are heavily flawed and various systems in our brain need to make up for this fact. For instance, our eyes can only have clear focus on a small part of our field of vision and have to constantly move around to scan the full field. Despite this, we perceive vision as if we have a clear picture of everything in front of us. This is all basically due to the brain guessing what it should be like in the blurry patches. So the raw data that comes in is severely lacking and we need to make up for this. The way this is done is by learning rules of how the world ought to work, and then extrapolating any accessible information to get a full picture. It's not just vision that works like this, but every single one of our senses (note that these go way beyond sight, smell, hearing, etc. and also include things like balance and sense of time).

Secondly, the information that comes in from our senses is not very valuable in its original form. At its very basic level, vision is just a list of dots and a value of how bright each one is. On its own this doesn't tell us anything. In order for it to be useful it needs to be processed. The human visual system first breaks the information down to things like borders and shapes. When this is done, it can start recognizing certain patterns and eventually figuring out what sort of objects are in front of us. Before we get a conscious experience of what's contained in our field of view, all sorts of work has to be done. This work uses data not just from what we can see, but also takes into account all sort of other related information. A simple example of this is the checkerboard illusion:

Squares A and B are the exact same color (check here for proof), but seem like they're different because one of them is lying in shadow. The brain has taken into account the green cylinder and its shadow when evaluating the squares, and similarly any other related information will alter how the final image is perceived. And do note that we're just talking about the "pure" act of perception here. When you want the player to feel as if they're in a living, breathing world, that's something which takes place at an even higher level, relying even more on correlating information. The end result is that everything which you put into your world affects the final perception of it.

As stated earlier, most of the time Inside does a good job at this. But there are a few moments where it doesn't work very well. For instance, at a couple of places there are pressure switches that open doors and control various machinery. Given what sort of people are supposed to work in the building and the level of technology they have, this doesn't really make sense. This makes it very hard to take these pressure switches seriously. Instead, you end up thinking of them as abstract puzzle devices. This is okay from a gameplay perspective, but if you want to make a playable story it's quite problematic. The player can't form an interesting and coherent narrative from using these devices. They have to go from "I am living through a story" to "I have to solve a puzzle that the developers placed here" and they are pulled out of the narrative aspects of the game. The world is no longer "real", but just a convenient place for puzzles to take place. The boxes that can be triggered to shoot into the air pose a similar problem, and there are a few others like this.

In order to have the best possible playable story, it's crucial to keep these situations to a minimum.

Another related issue is the trial-and-error gameplay of Inside. Now, in Inside, just like in Limbo, having the player killed a few times before completing a puzzle is key to how the game is supposed to play. So it doesn't feel entirely fair to point out a very intentional core feature of the game as a problem, but nonetheless, if the goal is to make a good playable story, trial and error pose a problem. The reason for it being a problem is rooted in how the player perceives the game. As I stated earlier, the final perception of a game is the combination of a lot of different data. The goal is for the player to think of the game as "real", but a designer will always face a problem: the game's world is an imperfect simulation. The trick is to never let the player notice this fact and it's here that trial-and-error becomes an issue.

There is one important rule all magicians have: never show the same trick twice. Why? Because on the second go, the audience know what's coming and is much more perceptive to all the tricks they'll be using. Almost all magic tricks rely on either false premises or misdirection, and it's rare that these will work as expected twice in a row. Games work similarly to this. For instance, the dog that's chasing you in Inside is not a real animal - it's just an animated mesh controlled by some relatively simple code. But if you show it to the player in the right way, they'll see it as real. However, every time the player gets to replay a section, they'll notice more and more discrepancies. The player will have additional information about the underlying mechanics that drive the dog, and this information will feed into their final perception of the whole scene. Repeat the same sequence enough times, and the dog will go from a "ferocious, dangerous beast" to "gameplay object that needs to be passed".

On top of that any fiction requires a certain suspense of disbelief. The player must be a willing participant in the events that unfold and have a certain level of roleplaying for it all to work. This is active work for the player's part and if they choose not to do it they will see the experience as "just a game". Trial-and-error wears down on the player's stamina; their ability to concentrate falters and they will become increasingly unwilling to roleplay. Not only life-or-death scenarios have this effect, but puzzles can too. If a puzzle is too unclear, or just too complicated to execute, the player can be forced into a trial-and-error loop that erodes the immersive qualities of the game's world.

Overall, Inside does a good job making sure that this doesn't happen very often, and that's despite the game's focus on trial-and-error scenarios. It seems to me like the developers have been more conscious about this issue than in their previous game Limbo as most challenges are fairly easy to complete on the first few attempts. But from time to time you do get trapped in trial-and-error loops, and since otherwise Inside is so good at crafting a playable story, it's extremely interesting when it happens. Pretty much all games have this issue, but in Inside it is especially evident as to the sort of narrative issues that it causes.

It's worth bringing up that trial-and-error is not always bad. Sometimes this type of gameplay is crucial to get the right type of behavior from the player. It needs to be clear to the player what sort of things to fear and having a life-and-death puzzle challenge is really good at this. Forcing the player to redo certain actions until they do it right also teaches them what sort of play styles this game favors. This can often be crucial to the player's enjoyment. For instance, if the player doesn't understand when to sneak, and can just sprint through stealth sections, a large part of the experience is lost. Finally, trial and error can also force the player to take the world more seriously. For example, the above mentioned background cutscenes are made more palpable by having certain gameplay effects on the player. The player learns to view things happening in the background as not just fluff, but crucial to their survival. In the end, this makes the world seems more real and increases the sense of being inside a life-like world. Keeping a fine balance here is key, and more than most, Inside does an excellent job.

Another important thing worth mentioning is that Inside doesn't have any superfluous actions. Pretty much any action that you can do in the game, every box, lever, button, bridge and so on, has a gameplay purpose. This has a huge benefit for the puzzles, as the player will always be aware of what the constraints of any situation is. They can just ask themselves, "Can I interact with it?" and if the answer is yes, then they know it will be needed in order to solve the puzzle. This allows the game to have lots of "out of the box"-type of puzzles without coming off as overly frustrating. Since the constraints are so narrow, the player is bound to come across the solution sooner or later. This makes the aforementioned trial-and-error-loops less likely to happen and you get a better narrative.

However, all is not perfect with this approach. By making the game's world all about gameplay, it also removes a certain sense of authenticity from it. As explained earlier, what creates the final perception of a game is all of the aspects taken together as a whole. Inside doesn't let the player freely explore the world, and because of this you lose a whole swathe of information that could have improved the overall perception. For instance, the player could learn more about the inner workings of the machinery or get a better understanding of its various inhabitants. Inside's interaction focus also forms a player role which is all about overcoming whatever puzzle is in front of them. While the player, in a certain sense, does explore the world of Inside, this is more of a side effect. The mechanics of the game almost never encourage the player to explore for pure, intrinsic reasons. Apart from a few scattered secrets Inside constantly revolves around solving the current puzzle at hand. This gives a certain sense of the whole world revolving around the player, and takes away the feeling of the place having an agenda of its own.

Side note: Adrian Chmielarz has a great article on how the most immersive games around all treat the player as an intruder. It is well worth a read as it goes over a lot of the issues that you can find in Limbo. Read it here.

A particularly big problem that arises because of Inside's straightforwardness is that the game's spatial representation suffers. When you go through a game, or any location really, you continuously build and update a mental model of the place. At first this mental model will be very sparse and lacking, and you won't really have an intuitive sense of the place - it just "exists". But as you traverse the world you start to understand how everything is connected and your mental model goes from being very simplistic to actually becoming a virtual representation of the place. It's when the mental model becomes sufficiently detailed that a world starts feeling real. As an example, compare quickly riding a car through a town and actually living there for a few days. In the former it's just scenery; in the latter it's an actual place. Not having this mental model is big problem in Inside. You always quickly pass by the various locations and never get to know them spatially. As such, the game's world never forms a cohesive whole in your mind. And without a proper mental model of the game's spaces, the final perception of its world suffers.

Again, just like the trial-and-error design, this is a bit of an unfair critique. The walk-forward-to-progress mechanic is part of the core of what kind of game Inside is. But that also causes issues to arise and it becomes easier to pinpoint narrative shortcomings.

Finally, another thing that that I think is worth bringing up is that Inside never properly introduces its protagonist. Early on there's a bunch of stuff done to get better a feel for the character we are playing, such as animations and the deadliness of the world. But there's never a shred of information as to where the boy is heading and why he wants to go there. This becomes problematic when your brain tries to weave a cohesive narrative through the game. At some points there are concrete urges that push you onwards, such as trying to escape some danger. In these instances, a story-like sequence will form in your mind and the journey feels like a proper narrative. But at other times there's no intrinsic reason for you to push onwards, and you only do it becomes the game tells you to. Once again, the makes the game go from playable story to a "just" being a puzzle platformer.

It's obvious that the developers intended for the boy's background to be vague, but I think they could have provided some more information. At least Limbo has "Uncertain of his sister's fate, a boy enters Limbo" as a description, which at least gives us a clear goal: "find your sister". Inside has "Hunted and alone, a boy finds himself drawn into the center of a dark project," instead, which lacks a clear motive. And playing the game, I can't remember that I ever felt drawn towards something from a narrative perspective - the game itself compelled me to move onwards. It's fairly obvious that the developers intended the premise to be vague, but I think it's damaging to the game as a playable story. When we have a proper premise we can latch onto it is much more easily to create meaning around the various events. It also gives the player directions on how to roleplay and as a direct continuation of that helps enhance the perception of the game as a whole. A good example of this at work is the opening to to Last of Us (2013), where first a cinematic and then a playable sequence gives us a great setup for the character we are about to play for (most of ) the remainder of the game.

Note that Inside is far from a failure in this aspect. As I mentioned earlier, the animations and general setup helps a lot in setting up the character. I just felt the game left a bit more to be desired. And I find it annoying how many games skip over the the intial setup and start the game without any real sense of who you are playing. Interactive storytelling relies heavily on roleplaying, but without any defined role to take on this is very hard to achieve. Because of that I thought it was extra important to point out.

And that sums up my  thoughts on Inside and playable narrative (for now at least!). While I might have come off as a bit harsh at the end of the essay, it's worth making it clear that I still think Inside is an amazing game. In terms of being a playable narrative Inside is really great - sometimes even genius - but that doesn't mean it's the best way to do things. As I've hopefully outlined in this essay, there's a bunch of stuff that can be improved. But it's also evident to me that Inside is on the right track. If we want to make proper interactive stories, Inside shows the way and is one of the best examples that currently exists.

Tuesday, 14 June 2016

How We Hire At Frictional Games

Now that we're in the process of hiring a new gameplay designer / programmer, it feels like a good time to talk about the hiring process here at Frictional Games. We're a small company, so don't hire tons of people, but we've still taken on quite a few over the years and have constantly evolved our strategies for doing so. Hopefully this article will be of help both to people who intend to hire others, and for those who want to apply for a job (with us or with anyone else).

First of all, you've got to figure out what to write in the work ad. Usually ads tend to be very specific with requirements such as "At least 3 years of work experience as junior programmer". They also tend to inflate the skills needed in a certain area, e.g. "World-class skills in linear algebra". Normally these are meant to set a high bar for the applicants; the assumption is that very few people will meet the demands, but that people will apply anyway. We tend to be a bit softer in our approach, and be a bit vaguer in the skills we want. So, for instance, instead of saying "good knowledge of 3D maths," we just set a requirement saying "has worked on a 3D game".

The idea is to make sure that people who might be unsure of their skills still send in an application. The same goes for any requirement about years of experience. While, of course, it's great if someone has worked on commercial games for years, we don't want miss out on people who have been working on hobby projects for years. In fact, because of how we work at Frictional, having worked on hobby projects can sometimes be more important than working for a commercial company. I'm particularly emphatic about this as I myself had zero proper work experience when starting out on our first game - Penumbra Overture. I've also collaborated with lots of other extremely talented people over the years who've never held a proper job. So take that as advice: work experience is far from everything, and make sure you apply even if you're not sure that you match the criteria!

When the ad text is done, it's time to find places to post it. Over the years we have found that, by far, the best place for this is our own Twitter and Facebook pages. Posting on Gamasutra tends to lead to a lot of applicants, but over 90% of them are from outside of Europe, and we can't employ those. Pretty much all the other sites we've tried haven't been worth our time or money. The one exception is Polycount, which is a great - and free - place to look for artists. In fact, all of the modellers we've hired over the past five years or so have seen our ad on Polycount.

Once the ad is out, it's time to wait for the applications to pour in. Usually we get 100 - 200 of these over a period of a month or two. This is when we begin phase one of the elimination process, which is just to remove anyone that's obviously not right for the job. This includes people who live outside of Europe (and have no plans to move here), people have the entirely wrong skillset (e.g. programmer applying for an art job) and people who simply can't write a coherent sentence. About 70% of the applications are rejected at this point.

With phase one over we move on to phase two. Now we discuss all the applicants and consider whether they could be a good fit or not. This usually means that we send people further questions in order to clarify their situation. For instance, sometimes it might be clear that someone might not have access to highspeed broadband, or only wants a part-time job, and so forth. Sometimes these kind of issues are negotiable, and in other cases it forces us to reject them. If needed we also ask further questions to probe their knowledge a bit. This is never of the "solve this puzzle"-variety, but just going a bit deeper into their previous work. Normally fewer than half the remaining applicants move past this stage.

Now we're on to phase 3, which is the most important step of them all: the work test. This is a pretty big test, usually demanding 40-80 hours of work, in which people are given a package with our tools and some of our assets. Depending on how busy the applicant is, we're pretty lenient on how long the test will take and how they'll split up their time - it could be over a few weeks, for example.

An important part of the work test is that everybody doing it always gets paid. We feel it's bad form to ask someone to spend a week or more of their time on something without getting anything back. I'm especially proud of one guy who applied for a job, got rejected, but then used the work test money to start his own indie studio. That made it even clearer to me that paying people for doing work tests is the right thing to do.

In this test they need to create something based upon a few vague directions. There are a lot of things that we get out of this. One thing is to see their general work ethics. All the people doing the test need to log their time - we can see how conscientiously they do this, how much time they spend on each task and so on. (And if they lie about the time-keeping, this is very easy to spot.) It also gives us a good sense of how creative the person is, how good they are at using unfamiliar tools and so on. Normally everyone in Frictional will have a chance to review the work tests if they want to, so we have plenty of feedback and a good consensus on the candidates that stand out. Again, fewer than half of the applicants move on from this phase.

If we consider that a work test has been passed, it's time for the final phase: an interview. We used to do these interviews before the work test, but found that we usually didn't get much information from them - but once a work test is done, we have something to discuss and a basis to ask questions. I also wouldn't want a "bad hair day" to be a reason for us to reject someone. I want people to show what they can by doing good work, not by being good at selling themselves. It's also worth noting that we never ask any hard question of the "explain Quicksort in 30 seconds or less"-kind, as we're unsure how much that really says about someone. I know I would have trouble doing simple multiplication  if I was nervous during the interview. So, instead, this phase is about getting to know the person and when we ask questions it's more about learning how this person thinks, and about getting a sense of how they might gel with the rest of the team.

It's now time to decide who we want to offer a job to. There's usually lots of deliberation at this point and it's never an easy choice. Finally we send our chosen person a mail saying they got the job, and check to see that they're happy to accept it. At this point there might be some salary discussions, but we try to settle the basics up front so we know hiring a person is within our budget.

When the person eventually starts, we try to ease them in. In the past I've made the mistake of giving people too much too soon, and this often makes it all work very badly. It's important that people feel at home when starting to work; if any issues arise early they are often hard to fix. Because of this, we make sure that the new employee has a lot of basic learning to do. The idea is that both parties decide when it's time to move on to actual work. We set up a number of features they must test and get used to and the newly employed can then work on those until they feel confident in their skills.

When that grace period is over, they're a proper part of Frictional! Mostly, at least. We give people 1 - 3 months of trial period in order to make sure they like the job and that they're a proper fit for the company. Usually it only takes a week or so to see if all is okay, but it's good to have plenty of time to make sure.

And that sums up how we hire at people at Frictional Games!

If you think it all sounds exciting, please consider applying for the gameplay programmer / designer position that we currently have open (more info here).

Friday, 10 June 2016

Hiring: Gameplay Programmer / Designer

Frictional Games, the creators of Amnesia, SOMA and Penumbra, are looking for a gameplay programmer / designer to join our ranks. This is a full-time employment, working from home.

At Frictional our goal is to create innovative, narrative-focused experiences, and explore how to use the potential of the videogame medium to the fullest. Penumbra was a unique combination of horror, physics and adventure gaming. With Amnesia we redefined what horror games can be. In SOMA we explored deep subjects in new unsettling ways. We aim to continue this trend.

Working with us, your core duty will be to collaborate with other team members in order to design and craft immersive worlds and experiences. At Frictional we do not simply rely on the core mechanics to carry most of the game. What makes or breaks our experiences is the care giving to the all specific level elements. Making sure these are top notch, both in design and implementation, will be your main duty. This includes everything from level layout and pacing to designing and implementing puzzles and smaller events.

Here is a rundown of some of a few more specific tasks you will find yourself doing:
  • Design the layout of a level based on a high level summary.
  • Come up with ways to convey certain story moments in the most efficient manner.
  • Design and implement puzzles so they fit with gameplay, world and story.
  • Script background events that sets the tone for a level.
  • Implement gameplay systems, such as an inventory.
  • Tweak AI to make it behave like you want.
These are far from the only things you will be doing when working with us. We like to see people dabbling with things outside of their expertise and to constantly learn new stuff. The more areas of development you are able to take part in, the better.

Of great importance is that fact that you will be working from home. Frictional Games does not have an office, so it is crucial that you are able to plan your day, and work without strict guidance. Because of this, we allow an amount of flexibility that you rarely find in an office environment.

The most basic requirements are the following:
  • You are fluent in English.
  • You have played a major role in completing at least one game.
  • You have worked on a game that uses 3D environments.
  • You have a fast and stable internet connection.
  • You are well-versed in C++, C#, Java or similar.

Additional (non-essential) requirements:
  • You live in Sweden or are prepared to move here.
  • You have an interest in narrative games.
  • You are good at writing, drawing or both
  • You have an interest in horror and sci-fi.
  • You have a Windows PC you can use for work capable of playing recent games (SOMA is a good test case!)

If this sounds like the job for you, send your CV to now!

Monday, 30 May 2016

Frictional Merchandise Is Finally Out!

For a long time we have been thinking about doing merchandise, but we never really felt that we had the time nor the proper partner. But recently we managed to free up some time and locate a nice partner in the form of Gametee. And after a few months of setup and work, the goodies are finally here!

First out is a t-shirt and a hoodie with Amnesia print:

Order them here:

Our plan is not to stop here but the goal is get our merchandise related to our other titles.

Let us know the in comments what more you would be interested in!

Thursday, 24 March 2016

SOMA - 6 Months Later

It is now a bit over 6 months since SOMA was released, so it feels like it's time for an update on how everything has gone so far.

The total number of sales, across all platforms, is currently at a bit over 250 000 units. This is pretty good; it'll only take 20k - 30k more until we've earned back our entire investment in the project. Given that the daily sales are still solid (about 125 units a day) and we have regular boosts from various sale events, this is bound to happen well before this year is over.

While this is a good result for us, it's by no means earth shattering. For instance, Firewatch (which has quite a few elements in common with SOMA) sold over 500k in just a month, so there's obviously room for SOMA to sell a lot more. It might seem weird, but this is actually very encouraging for us. SOMA was a really ambitious project which took 5 years to develop, used a load of external help and had a big chunk of money spent on a live action series and so forth, making it a very costly affair. Yet SOMA is well on the way to becoming profitable after just 6 months, despite not being a runaway success. This makes us a lot less worried about making another game of similar scope.

Still, it's interesting to ponder what kept the game from selling even more. One stand-out thing that we've identified is that the game falls between two genres: horror and sci-fi. What this means is that the game might feel a bit too sci-fi for someone looking for a pure horror experience and vice-versa. While we think the mix works very well for the game, it seems quite possible that this has put off potential buyers. I'll discuss this in more depth later on.

User created custom stories was (and still is) a big part of the Amnesia community. So far almost 450 Amnesia finished mods have been released. This is despite the game's mod support being far from good. So with SOMA we wanted to make sure we allowed even better mod support, so we would hopefully get as many mods made as we did for Amnesia.

Unfortunately the modding community around SOMA hasn't really taken off. So far only 5 custom stories (2 on moddb and 3 on steam workshop) have been released, and while it's amazing that people spend time making mods for SOMA at all, we expected that there would've been a few more. Just about everything in the game is controlled via script and modding allows you to replace any file, making it much more powerful than in Amnesia. Because of this, we'd hoped to see people do really crazy things with mods, but apart from Wuss Mode and a location tracking Omnitool there isn't much out there. Both of these are very cool modifications, but considering that the game could have been changed into an RTS or a racer, we'd hoped to see more experimental stuff.

It feels worthwhile to discuss why modding hasn't been as successful as it was with Amnesia. The first and most obvious answer is that SOMA is simply not as popular as the mega-hit Amnesia: TDD. Secondly when we released Amnesia there weren't many other similar horror games around, and as a result many of Amnesia's mods got played by popular streamers. This gave people a huge incentive for completing their mods. Thirdly, both the level creation and scripting is a bit harder in SOMA, making it more of a hassle for people to get things up and running. And finally, you can quickly create gameplay in Amnesia by just placing a few basic items, whereas SOMA requires more setup and lacks easily reusable elements. All of these issues combined probably explains why fewer mods are being released compared to Amnesia.

But we haven't given up on modding. Far from it. There are lots of interesting things in the works coming from the community (for instance, a very fitting SCP inspired custom story) and we're discussing what we could do to give people more incentive to create and finish more mods.

I think the most surprising part of the player response is the depth in which SOMA's story and subject matters have been discussed. For instance, there have been really interesting discussions as to whether the game's (semi-)antagonist, WAU, is evil or not. Patrick Klepek at Kotaku wrote up a nice summary on this that can be found here. While WAU was designed to not be really evil, it was very surprising to see some people seeing it as the good force in the game's world. It made us look at the story in ways we'd never thought about ourselves.

Another interesting discussion has been the coin-flip 'controversy'. For instance, here is a long discussion that brings up various sides to the argument. This was another thing that we didn't figure would be very controversial, but ended up spawning tons of intelligent arguments. It's been a great deal of fun to see discussions like this. There's one aspect of the game that we've only seen mentioned once, though, which we thought would be a much bigger issue. What that is we'll leave as an exercise for the reader to figure out.

It's also been great to see all of the real-world connections people have made with SOMA. Here's an article that goes through a few of them.

By far the most surprising reaction we've had yet has got to be one guy mailing us saying that the game inspired him to fly to the US and propose to his girlfriend. We've always seen SOMA as a rather bleak game, and it was really interesting to see how some people actually found it uplifting and inspiring.

In all, we couldn't really have hoped for a better response. People report still thinking about the game months afterwards, and that it's made them think deeply about subjects they haven't considered before. This was what we were after when we started the game all those years ago, and it's incredibly satisfying to see that we managed to reach that goal.

I mentioned above that a problem with SOMA is that it lies between two genres. Not only has this probably led to lost sales, it's also most likely the reason why SOMA cannibalized the Amnesia sales. The moment that SOMA came out, sales of Amnesia: The Dark Descent went down too, and has stayed down ever since. We saw the same happening when we released Amnesia: A Machine For Pigs, but since SOMA is in many ways quite different from Amnesia, we thought it wouldn't happen this time. But it did, and the reason seems to be that people lump both titles under a "Current Horror From Frictional Games" label.

In order to combat this issue we're thinking about differentiating the games we make a bit more. So if we make another sci-fi game, we'll probably tone down the horror elements and make the sci-fi narrative more prominent.  The reverse would be true if we made a new horror game. The idea is that this'll not only let us reach a new and wider audience, but also minimize the risk that people will mix up our games, and instead they'll see them as separate entities. With SOMA it feels we've made it clear that Frictional Games is not just about pure horror, and we want to take advantage of that and diversify the experiences we craft.

Related to the above is our new internal development strategy. For the first time in company history we're now developing two games at the same time. This will require non-trivial changes in how we manage the team, but in the end we're very sure it'll be worth it all. By having two projects going at the same time, we can release games at much higher frequency. In turn, this let us be more experimental as we don't have to rely as much on each new game being a big money generator. We're still in the early phases of this transition, but it's shaping up really well so far.

This also means we might do some recruitment in the near future. Watch this space for more news on that!