Sunday 27 November 2011

The Problem of Repetition

Introduction
After having played some adventure and RPG games lately something struck me: repetition in games have almost the same problems as trial-and-error do. This is not really a shocking conclusion, since repeating things in a game is basically what you do when stuck in a sequence of trial and error. But since the repetition is not a direct consequence of being unable to progress, and that not all repetition is bad per se, I figured it was worth looking into a bit.


The Problem
Most of the time the problem arise when doing an action several times causes the same response. Mostly, this does not apply when doing things to dead objects, like shooting a bullet on a wall. We expect that if we shoot the same bullet at the same place twice, the same response occurs both times. However this is not always true. For instance, many games use randomized particle effects for sparks from the hitting bullet. In more complex system, like water splashes, this is even more common, and while it might not be directly noticeable if they repeat, it can unconsciously lead to the virtual world being seen as less "real" (what I really mean is sense of verisimilitude, but more on that later) . So even though it does not constitute a large problem, we do run into trouble even when repeating consequences for very simple interactions.

The problem becomes more jarring when the object of interaction is a supposed to be an intelligent agent. This is very common in RPGs and adventure games during dialog, where the same question generates the same answer regardless of how many times you ask it. Adventure games are generally a little bit better than RPGs and often have NPCs giving a summary instead of the exact same response and more frequently terminate threads of conversation. Even so, a big part of dialog in both types of games have actions being met by the exact same response no matter how many times they are repeated.

There are of course a reason why it is like this. The player might have forgotten some information and need to hear it again, forcing dialog to be repeated. Or there might be some compulsory puzzle that requires the player to trick or persuade a character, which forces the player to redo the same conversation if unsuccessful at the first attempt. I think these reasons expose two problems that narrative focused video games have: reliance of "info dumps" and puzzles as core activities. Info dumping is a form of exposition that one tries hard to avoid in other media, yet is very common in video games (often forming the core storytelling device). It is something that I think needs to be considered more (and I am well aware we have been using it too much in our own games). Puzzles is something I have talked about having negative effects before and this is yet another argument to why we should try and cut down our reliance on them.

Another very common form of repetition is that of having the same kind of gameplay scenario repeated several times throughout the game. Sometimes this can be a core part of the experience, but most of the time it is just a form of padding and an attempt to prolong the time it takes to finish the game. There are tons of examples of this and two that spring to mind are the vent sections of Dead Space 2 and the spirit capturing in Sword and Sworcery. I felt that both of these activities would have been a lot more interesting if not repeating so much. You quickly become very familiar with them and they eventually loose much of their first

There is a deeper reason why repetition is so common in videogames. Many games base their interactions on traditional games and software systems where reproducibility is a corner stone. You do not want to use a paint-tool and not know what expect when pressing a button, and the only way for you to get this knowledge is to is for consequences to repeat themselves. In traditional games, you need to have systems that a human player can keep track of, and thus the consequences of actions must be easy to comprehend. Videogames carry baggage from both of these directions, and thus it is not strange that video games contain a large share of repetition.

As you might have guess I think this sort of repetition can be quite bad for videogames that focus on story and narrative.

The Causes
As I said earlier, the repetition has pretty much the same issues as trial-and-error. Since they are both about doing the same thing over and over, this can feel pretty much self-evident and not worthy of much discussion. However, while trial-and-error elements are more easily pointed out and can be directly addressed, repetition is more subtle and not always as obvious. Many of issues with repetition are also commonly seen as limits of the medium (or at least our current technology) and thus not really addressed. I do think these problems can be overcome though, and a first step is to figure out what give rise to them.

- Mechanics gets apparent
By having something repeated over and over to players, they will quickly start to notice patterns and short after figure out the system below. What this leads to is that the player will no longer focus on what the system is trying to represent (eg. dialog with a person) but will instead see the mechanics that it is built from (eg. the abstract dialog tree). Repetition does not force this onto the player as trial-and-error do (where the player often is required to learn the system in order to continue). But since many of the things that are repeated constitute a big part of the experience, the problem piles up. Like I mentioned above the repetition can include entire scenes and the player might go through a section in a go (ie no trial-and-error). But then when a very similar sections is repeated throughout the game, the underlying mechanics become more and more visible. As an example I think the enemies in our own game Amnesia have this very problem. This problem is very subtle though as it only applies on longer play sessions and can thus more easily slip by.

There is another aspect to this, that makes the problem even more severe. Once you figure out the mechanics of a system it can damage events that you experienced when you did not have this understanding. For instance, if you feel like a conversation is really meaningful, and then later on find this same character reduced to mechanics, it will change the way you view your prior experience. It will be very hard to still feel the same sense of meaningfulness when looking back at the conversation. Your mental construct of an aspect of the game's world has now been reduced to a mechanic and when you later summarize the experience you have had, this can severely reduce any emotional attachment you might have had to earlier happenings. As this piles up, it will slowly degrade the experience and makes you less emotionally connected to the game's world.

- Decrease in Verisimilitude
What verisimilitude means is basically how real and truthful the fictional world feels. This does not mean how well it replicates the real world we live in, but how much a it feels like it represents an actual place. In most narrative media, giving a strong sense of verisimilitude is really important. As I said, this does not mean that everything should be "just like in real life", but instead follow the fictional world's internal logic somehow. What this means in games is that when encountering a virtual element, such as a character, we do not need for it to behave exactly like in real life, but simply to behave in such a way that it evokes feelings of verisimilitude.

This means that we can tolerate dialog selection and similar, while other things are instant deal breakers. I think one of these deal breakers is the repetition of a responses. If a character repeats the same sentence over and over, it is very hard to see them as nothing but a simplistic automaton. We can quite easily disregard our knowledge that there is not a sentient mind
shaping the responses, just like know something is not really happening in a movie. But when the information that the experience is feeding us (in this case the repeated voice response), the very thing that is supposed to support the view of an intelligent being goes straight against its purpose.

Not only dialog is affected by this but plenty of other aspects. For example, whenever you have to go about clicking on the same hot-spots over and over in an adventure game, it also significantly reduce the feeling of verisimilitude.


- Decrease in effectiveness
This point is almost identical with what happens in trial-and-error. Certain scenes and events simply does not do well when repeated. For some events it is simply that they are very emotional, and it will be hard to feel the same way once again. You will grow desensitized and less prone to reacting to it. Just compare a movie filled with gory sequences to one with a single visceral scene. The latter will pack a much harder punch. Other times it might be that the event or scene is set up like a magic trick - it only works when you are not expecting what will happen. Finally, it might simply be that the passage is too boring, sensory intense or similar that you cannot bare to take further viewings. Other media rely on things like these hard-to-repeat moments a lot, but since games are so prone to repetition, they are much harder to put in and/or to have the same emotional value.


The Cure
So how do we overcome these issues? I think there are a few things to keep in mind when designing that makes them a lot simpler to avoid:

  • Not a approach the experience as a competition. The less goals we set up for the player the less likely we are to need to repeat things for the player or to make them repeat their own actions.
  • Make sure that the story is understandable without the need of info dumps. If the player is required to have story related information repeated to them, then I would consider that bad narrative design. The story should emerge simply out of playing.
  • Skip the notion that players need to learn a system. I think this is mainly historical baggage from how software works for more practical application, where mastery of the system is essential. Creation of narrative art does not have this requirement though, and I think we should instead make the player focus on the representations (graphics, sounds, etc) that the system provide.
  • We must demand more of the player and give them more responsible. We must teach them them live in our virtual worlds instead of trying to beat our game systems. As most games reward players for combing the virtual world for goodies this is not the easiest of tasks though. Our goal must thus be to undo this and reward roleplaying instead.
These small rules does of course not solve everything and there is a lot of hard problem connected with this. For instance, conversational responses is an incredibly tricky problem and the same is true for narrative devices in games.

Still, I think just a little change in our thinking can take us a long way and simply recognizing the problem is a big step forward.


Tuesday 1 November 2011

Thoughts on Heavy Rain

Introduction
It is very easy to talk bad about Heavy Rain. One can say it is just an interactive movie where you press buttons at certain key moments, in rare cases changing the outcome of the story. One can talk about the hole and cliche filled story and the weakly developed characters*. One can talk about this and other negative aspects of the game and I would fully agree. But if one only focuses on these areas, there is plenty of really interesting aspects that are missed.

Despite all these flaws I really enjoyed playing Heavy Rain. Sure, the quick-time-events (QTE:s) really got me worked up on more than one occasion and a lot of other issues bugged me, but on the whole it was quite an engaging experience. There are some truly tense and disturbing moments in the game that work great. For example the scene at the mall, while lame in many ways, managed to capture the protagonists sense of panic and that in an environment and setup I have never seen in a game before. The game also features great graphics, nice music and not too shabby acting (for most of the time anyway, and once you get used to the uncanny valley feel). The game also lets you be in situations that I have never seen outside of Interactive Fiction.

What really made the game interesting though was not the things that I liked, but the things that are slightly broke. Because of the way that QTE:s work, being a quite fragile system in terms of immersion, it sort of exposes your own usually hidden thought processes as you play the game. Also, the game's filmic nature and focus on a branching narrative makes it a virtual smorgasbord of game design theory to try out. This is what truly makes Heavy Rain worth playing.


Immersion as an essence
By far, the most important realization I got when playing Heavy Rain is how interaction is not mainly about giving the player interesting choices. When playing the game I never felt the need to make choices on the basis of seeing what would happen, instead I simply wanted the characters to act in certain ways in order to confirm to my expectations of how I thought they would (and should) be acting. What I think happens is that as we interact in a videogame, there is feedback loop between us sending input to the game and us getting information back from the game (in the form of visuals, audio, etc), which builds the basis of us feeling present inside the game's virtual world. The better this loop works, the more we feel as a part of the experience.

Heavy Rain is an excellent example of this process at work. When there is flow in the controls (which is usually in the scenes giving you direct character control, such as the early mall sequence), there is a very satisfying feeling of being one with the character. Then suddenly some weird QTE pops up and you either fail at completing it, or it simply does not give the result you expected, and once again you are pulled out of your sense of presence. The game is littered with moments like this, pulling you in and the throwing you back out. When Heavy Rain manages to sustain the belief of you having agency over the character, that is when the game is at it best. These are the occasions when there is a very strong loop of interaction going on and you are the most present inside the game's world. When this loop is broken, it does not matter what kind of interesting choices you might have at your disposal. The game immediately becomes less engaging the moment the loop of interaction breaks down.

In this light of thinking, QTE events make perfect sense. It is simply a rudimentary system for trying and sustain a feedback loop during various types of scenes. It is not about setting up a competition for the player, it is just a very blunt and unreliable system to sustain a sense of presence. I really doubt that QTE:s is the way to do narrative art in videogames, but it does gives us invaluable information on how to proceed.

What all this seem to indicate is that a videogame that wants to tell a story, should not use interaction to deliver a multitude of choice, but instead to reinforce the feedback loop of immersion. This might entail having choice, but the choices in themselves are not what is of the most importance, giving a very sharp focus on how to design the mechanics. It may actually be that the very future of making artful games with focus on narrative is to focus on this interactive loop of immersion. There is a lot more to discuss on this subjects and there are other things that also points in this direction. I am hoping to devout an entire post on that subject soon, so consider this a taste of things to come.

A final note: This "interaction as a means to create immersion" does not imply that the future of videogames are incredibly linear interactive cinema -far from it. In many cases a non-linear and open game world is essential in order to support the feedback loop.



The importance of determinism
In most games you have a pretty strong sense of what the protagonist will do when a button is pressed. Not so in Heavy Rain. Apart from direct movement and a few repeatable actions (like be able to shout your son's name in the mall scene), most of the time icons just pop up with vague hints on what the input will achieve. Sometimes you will learn what action might happen (such as that an up-arrow at a railing will mean that you will lean against it), but this takes a bit time and requires that a similar action has already been carried out.

In many cases this has a drastic reduction on the sense of presence. For one, it makes you unable for you to form plans. Simply by surveying an environment you cannot determine a course of actions (even if you know all trigger spots), and during action sequences it gets even worse as QTE:s may up at any moment in pretty much unguessable form. Making up plans is one of the basic corner stones of human intelligence, and possible the reason we developed a conscioussness, so not having the option of doing this is a hard blow against the sense of agency. Another reason it reduces immersion is that your character might not act in the way you intended. Before picking an action you almost always makes some kind of assessment of what will happen, but it is quite likely that this will be dead wrong. Thus the character your are supposed to feel a connection to, ends up performing an action that you did not intend. Of course, it is very hard to feel as a part oft he game's world when this happen.

This system stands in stark contrast with how Limbo works, where you are pretty much always certain of exactly what will happen. I think this is very much connection in the level of immersion Limbo manages to have throughout (unless you get stuck in trial and error of course), and how Heavy Rain stumbles through the entire experience. One should not be too hard on Heavy Rain though as the space of interactions that are possible to perform throughout the game by far outnumber those in Limbo. The real challenge for the future is to coming closer to multitude of actions in Heavy Rain, but still having the determinism of Limbo.


The understanding between Player and Videogame
Another big problem in Heavy Rain, which is related to the point above, is that the game sometime seem to work against you. It might seem obvious that this is a dealbreaker in terms of immersion and I have already discussed the problem of camera control in Dead Space Extraction. The issue can be a bit more subtle though and Heavy Rain serves as great example of this. For instance, in one scene I had made a plan of actions: to first bandage an unconscious person and then to poke around in his stuff. There really was nothing hindering me from doing so but instead the game removed my ability to interact directly after caring for the person. The game interpreted me wanting to help the guy as I also did not want to poke around, thinking that they two were mutually exclusive actions. Of course I thought otherwise and considered it no problem at all to do some poking afterward.

There are plenty of situations like this and it makes it quite clear that you should never move ahead on a bigger outcome from a choice without being certain that this is also what the player expects. I also see this as a problem of having major choices the player in a game that lack a high level simulation (like Fallout for example). Just the simple action of walking out a door can have many different meanings to a player, and one needs to be careful and make sure that most players have same idea of what it means. Once you throw branching paths into the pot, it gets a lot more complicated and clashes between player and game is much more likely to happen.



Emotional Simulation
An interesting aspect of Heavy Rain that I have not seen (at least not this directly) in any other game using QTE:s (or normal mechanics for that matter) is to trick the player into feeling certain emotions. The way it works in the game is that the player is forced to hold down a lot of buttons at the same time, while often also moving the stick around. This creates an uncomfortable and demanding way to hold the controller in, which is meant to simulate the way the onscreen character feels. While it might sound a little dodgy, it works quite well in many cases, especially in a scene containing self-mutilation.

The research behind this kind of response is actually very well established and designer Chris Pruett has hypothesized that the effect is probably a reason why many unforgiving horror games turn out to be extra scary (a design decision that comes with other problems though). The way it works is that we humans often do not know why we are feeling a certain way and unconsciously project it onto something else. For instance one experiment had people thinking that arousal due to their fear of heights was due physical attraction instead.

All is not good with this design in Heavy Rain though. Because the inputs you perform are not fluent (as it is prompted on a situational basis) and non-deterministic (as explained above) you are mostly very conscious of what you are doing with the controller. If the controls where more transparent (like in Limbo) you would be less conscious of your input, and any uncomfortable placement of the hand is much more likely to be projected into whatever the protagonist is doing. I think this can be very potent stuff if handled properly and let the player get immersed in experiences that would be hard to simulate in any other way.



Trial and error
Heavy Rain boasts that it does not have any game over screen, but it still manages to have is massive amounts of trial-and-error. This time the forceful repetition of events does not only occur in death threatening situations though. In Heavy Rain it often happens during extremely mundane actions like brushing your teeth and taking a shower. It is an extremely good example why this sort of design is so immersion destroying. From believing that you are playing an actual living character, the sudden requirement to repeat an event pulls you out from the experience directly. It is so obvious that you go from trying to become present in a virtual world to just trying and overcome a very mechanical task.

I think the biggest problem is that Heavy Rain is very sensitive in how you complete the QTE sequences. Let go of a button for a micro seconds and it results in an instant failure. When the game gets rid of so many other stigmas of old game design, it is sad to see it stuck in this one. I think the way it should have done it is to become a little bit more relaxed and to allow some more failures. Instead being competitive-like and very strict in the actions, it should instead check if the player tried enough to do something. As long as the players are playing along, I see no reason for punishing them. The game should have tried to keep the illusion of an interactive-feedback loop alive for as long as possible, instead of simply breaking it at the slightest incorrect input.



Some misc points
Now for some shorter stuff that I found interesting:
  • When done right, the direct and free control method is by far the more immersive. However it also puts a lot of pressure on the character reacting in a proper way. Quite often, the character I was controlling ended up acting like a moron, walking into walls and the like, even if I really tried hard to control him properly. The constrained events do not suffer this problem, and have the characters act much more lifelike, but at the same time they do not have the same level of interaction required for deep sense of presence.
  • Heavy Rain is at its best when simulating tightly space and time-wise bounded scenes. At these points it was much easier to give me a sense of having agency and to let me become one with the moment. The scenes self-mutilation, pushing through a crowd, escape from bench in cellar, etc are all great examples of this. Judging from what seemed to have worked best in Amnesia, I think a lot can be gained by taking this design further.
  • The game is a great test bed for a game that has decisions with big ramifications, such as the death of main characters. My own conclusion from Heavy Rain is that all of these choices are probably unneeded and did not gain me much except the sense of missing out on the story. Interestingly, Heavy Rain feels quite different in this regard from a game like Fallout (with the, as mentioned, more higher level narrative simulation).
  • Achievements (trophies on the ps3) really suck in story-centric game. Having gone through a scene and then getting a sort of grade, really removes the ability to make up your own mind of what just took place. It is quite similar to the "understanding between player and game" problem, as achievements has a high risk of going against the player's intentions (and does not really help gain anything).


End notes
As I think this post shows there are many reasons why Heavy Rain is a really interesting game to play. It does a lot of things that other videogames do not even dare to consider, and while it kind of fails on a lot of it, just attempting it is an important step on the way. If only more mainstream games were like this.

Also, after playing through Heavy Rain I have come to wish that there were more games like it. By that I do not mean more games with QTE:s (which I really hated much of the time) but games that allows the player to always progress and focus on a rich narrative experience. In most other games I either have to endure annoying puzzles or have to become an accomplish in a genocide. Given the high scores the game has gotten (from press and the public) I do not think I am alone in this. Please do not see this as an urge for people to copy Heavy Rain though, but instead to use the game it as a step towards something that truly makes use of the medium.



*Emily short has a really good essay on the story of Heavy Rain. Check it here.