Heavy Rain & the Player Character

While it’s easy to get bogged down in the bad writing, Heavy Rain is noteworthy in its treament of the relationship between the player and its playable characters. The game’s success and failures challenge strongly held notions about empathizing with characters though gameplay.

And oh yes, spoilers ahead.

Pushing your buttons

I’m not sure if Hitchcock was the first to promote this idea, but his films are the prototypical examples of the notion that the audience empathizes with a character performing a dramatic action, regardless of that character’s motivations. A viewer is unsure the character will be successful and that success or failure has dramatic ramifications for the plot. The character could be unlikable, a cold-blooded killer, and we would still be on the edge of our seats if the scene is constructed properly.

Jordan Mechner applies this logic to games as well – performing a character’s actions, even if we dissapprove of them, encourages us to empathize with the character. When Heavy Rain works, it’s by applying this interactive corrollary. You perform mundane actions such as tying a tie, diapering a baby or even just waiting, in a police station lobby. The buttons sequences you perform and the actions the character performs are incredibly different, but are tied together by a physicality unlike most games (as discusses more by Jorge Albor at Experience Points). In this sense, Heavy Rain wants to focus on the distance between the player and the character, but ties them together in a different way.

For Michael Abbott, the lack of autonomy affected his emotional connection to the story. The question is how much agency is necessary for this effect? Agency (as defined in “Agency, Reconsidered” by Michael Mateas, Steven Dow, Serdar Sali, and Noah Wardrip-Fruin) is not just being able to take as wide a range of actions as possible. We feel agency when the actions we can expect to take in the game’s environments (material affordances) line up with the actions that would expected given the game’s fictional setting and context (formal constraints).

Heavy Rain further complicates this by providing clear affordances for action via its button prompts in the game world. A line is drawn between material affordances as suggested by the context of the environment, and the material affordances clearly presented by the game’s interface. It attempts to line up expectations of agency by making this gap as clear as possible.

The story does allow for a pretty divergent number of ways the player can affect it – but it’s never clear when and why you have that option, as opposed to the elements you must perform correctly or where failure has no impact. Heavy Rain tries to make most linear sequences visually dramatic in standard suspense movie form, but the mundane sequences have no dramatic thrust by design. They are necessary in contrast – the suspenseful scenes, especially those where you do have a choice, are meant to be more powerful once you’ve gotten through these, but the perception of that gap in agency is highly subjective.

Press X to Jason

The game starts with Ethan Mars playing with his two sons, Jason and Shaun. In the next scene the family is at the mall; Jason runs away from his father so you must search for him. Quantic Dream obviously spent a lot of effort making the mechanics of this sequence match the desired emotional state. Characters stream by as obstacles slowing you down, your vision is limited forcing to turn around repeatedly and look in the distance, which is out of focus). You can press X to yell Jason. It’s mostly freeform character movement and not repeated button sequences, giving more freedom than in many scenes. Mechanically, everything is set for you to identify with Ethan and feel his tension looking for his son.

As Chris Dahlen points out, this sequence falls incredibly flat. Jason has no reason to bolt out into traffic, and is old enough to know better. He runs out into traffic because David Cage wants to manipulate your emotions. All entertainment wants to manipulate your emotions, the failure here is beginning with such a crass manipulation, and using as many tchniques as possible to force the desired emotion. It is the interactive equivalent of plying us with many drinks as soon as we sit down on our date. With no setup of character motivations or development, Heavy Rain does not earn the right to play us for cheap dates so early. David Cage is the one being cheap here.

The failure in the writing contrasted with the alignment of the mechanics of player navigation highlights the naive simplicty of the argument that mechanics are what drive home the emotion or theme of gameplay. The truth is always more complex. No aspect of a game’s attempt at meaningful play can be taken lightly as an attempt to cheat straight to the emotional payoff.

Your Artistic Vision Seems To Be Telling Me to Go Left

While bad controls plague lots of games, Heavy Rain seems to flaunt proven control schemes for player movement. Like The Path, the result often robs scenes of their emotional power. The difference between Heavy Rain and The Path is that in The Path there is no apparent artistic purpose to driving a teenage girl around like a tank. Heavy Rain on the other hand struggles to guide the player around its environments for dramatic purpose (just not very well).

After a few moments of fumbling with the character navigation, turning left when you’re expecting to walk forward, the control scheme almost seems character-relative (as in the first Resident Evil). However it’s typically camera relative a majority of the time. The navigation is impressive solely for how much it obfuscates what’s actually going on, so I can only propose a theory about the problems. The incongrous direction changes seem to be a combination of a very long time period on the frame of reference hysteresis when the camera cuts, and that the navigation guidance splines actually change the character’s frame of reference for control as well.

That’s probably worth some explaining. A standard feature of most games that have cinematic cameras (like the God of War or Uncharted series) is that while the character is camera relative (push left, the character moves left on screen regardless of the direction they’re facing), when the camera cuts the frame of reference changes while the player is still pushing on the stick. So the game keeps interpreting the player pushing left as moving in the same absolute direction, even though the character’s technically no longer moving left on screen. This is typically only done for a fraction of a second, giving the player time to readjust (and when done poorly results in the player turning around and crossing the camera threshold over and over again, leaving you in a particular brand of hell unique to this style of player control).

Heavy Rain uses interior spaces that are much closer to proportionate real world scale. In most third person games, the scale of obstacles (couch, chairs, etc.) is smaller than the scale of a room, to make it easier for the player to navigate. (For more detail, check out Realistic Level Design in Max Payne from GDC 2002 on Gamasutra). Heavy Rain uses more consistent scale to allow its scenes, at varied camera angles, to have the proper visual impact. To allow the player to navigate the space and avoid bumping into obstacles, which also undercuts the drama, the game has some sort of navigation helper splines around obstacles and walls. If you walk straight into a wall, the character will start turning to walk along the wall even though you’re still pushing forward.

My theory is that this adjustment affects the frame of reference for the player’s control, with another long delay on reverting back – with multiple frame of reference adjustments possibly going on, it’s impossible to tell the association between the direction you’re pushing and the direction you’re going. In tight spaces it takes several seconds of playing around with the control to find the correct direction to push.

When mediating between the actions of the player and the possibilities we wish to constrain in our story world, inevitably the game must make some prediction about the player’s goal. A more canonical example is the auto-aim in HALO, used to simplify first person shooting with a console gamepad. If the player is turning via the right stick, the game slows the rotation speed as the cursor goes over an enemy, making it easier for the player to stop on the enemy and fire.

The idea is to only take existing player input and amplify or reduce it to reach the assisted goal. You cannot change the direction the player has input. The straightforward revision to Heavy Rain‘s nagivation guidance would be to NOT change the control frame of reference to turn (if that is in fact what’s going on), and instead only help the player when they turn to walk around the obstacle – you speed up the player’s rotation as they turn towards the proper obstacle-avoiding direction, and slow it down a bit as they get away from it (with limitations on how fast and how large an angle should be adjusted).

The prediction of a player’s goal is inherently frought with problems, whether it’s at the lowest control level or at higher story levels. The most graceful solution is to never direct the player in a new direction towards the assumed goal, but merely speed them up or slow them down to reach the predicted goal easier. If it turns out the prediction is wrong, they still own the resulting choice.

The Curious(ly Aggravating) Case of Scott Shelby

Your interaction with Shelby early in the game is in short scenes, where he questions other victims’ parents. You don’t get much background on Shelby himself, but eventually he teams up with Lauren, a prostitute who is one of the victim’s mothers. They begin to suspect Gordi Kramer, a rich playboy whose father, Charles, they see at the grave of  John Sheppard.

He and Lauren are attacked and dumped in the river inside his car. Here you can save Lauren or leave her to die. I misread the consequences of the onscreen button prompts the first time I played the scene and Lauren died, but I replayed the scene to save her. In an earlier scene, Lauren gets upset at Shelby and leaves the car – I get out and convince her to stay with me for her own safety.

After the attack, Shelby goes to Kramer’s mansion, kills many of his bodyguards, to interrogate him about the killer’s identity. Why does Shelby risk his life to do question a man he knows is not the killer? Even though Kramer’s son is revealed to have killed a boy in a perverse copy-cat killing, Shelby seems to dismiss this. If you fail the assault on the manion, all the red herrings involving the Kramers are never explained, leaving an entire subplot in tatters. Shelby then sends Lauren out of town (for her own safety as he describes), kissing her goodbye.

I eventually achieved the “good” ending, where Shelby is killed by Jayden – leaving Lauren (and myself) to wonder why Shelby had feelings for her after murdering her son. The scene stands out as a poorly constructed filler to deal with the gap in accurately describing Shelby’s motivations. Did his motivations as the killer change as a result of interacting with Lauren? Was he acting out of a guilty conscience? Did he actually enjoy a sick pleasure of gaining Lauren’s confidence as the murderer of her son? None of these have any support in the writing whatsoever, leaving Shelby an empty shell of a character in the end.

On the Permanence of Rubber Bands

Overall, it’s clear Quantic Dream took much care to structure scenes in ways that would allow multiple outcomes with minimal impact on future scenes, to support as much player variation as possible. David Cage compares this storytelling structure to a “rubber band; the player can stretch or deform the rubber band through his actions, but whatever he does the backbone of my story is always there” (from 1Up’s Indigo Prophecy post-mort via Grand Text Auto).

Ethan Mars is put through several trials by the Origami Killer. At each point, the killer asks Mars to do something  in order to receive more information on his son’s location, like cutting off his own finger, killing another man, or drinking poison. In each scene, the choice presented is powerful in the moment. Going through the motions of cutting off your own finger leave you shaken, or you’re torn between killing a criminal who also is a father.

Whatever you choose, the killer keeps presenting his trials. Even if you refuse the last one, Madison figures out Shaun’s location and can call Ethan to tell him. While you’re affected by the scenes in the moment, at the end of the game they feel hollow. The permanent, long term consequences of most of those acts do not line up with any reasonable expectations.

This structure is used to minimize the work of building content to support variance on every path. It’s not clear that supporting the permanent impacts of those choices required a significant amount of additional work though – a line here, perhaps an extra scene there. The failure to explore those interactive plot elements is similar to the plot holes in the traditional narrative, and adds to the feeling of reduced agency.

Separate from global agency, the ability to have a large impact on the over-arching story world, an action’s long term and short term consequences have to in line with the player’s perception to create agency. But the elements of a gameplay choice that would lead to perceiving long term vs. short term consequences, and the corresponding design problems, are still a rarely discussed area.

I Win?

One of Heavy Rain’s design goals was to allow the player to continue playing even when there were life-ending consequences. As Jorge Albor points out, death has no meaning in Heavy Rain. Except for the meaning you apply to death.

Replay in interactive drama is often perceived as a negative because it assumes a single success state. Why would you have to replay a series of scenes over and over except to get at the final win state? This is not only limiting from the point of view of the player’s agency, it removes the player’s desire to improvise with the story (which is the ultimate goal of most interactive storytelling).

I found myself replaying a handful of scenes to get my desired ending – even though the game supported multiple endings, and even though the ending I wanted ended up feeling awkward (because of the identity of the killer ended up being different than I suspected). Even though Heavy Rain supports most permutations of the main characters being alive or dead, the game failed to convince me to improvise with it, or more accurately, failed to communicate when improvisation was encouraged. It fails to do so by not allowing apparent choices, by not supporting smaller permanent consequences, and by not being consistent in its communication of those consequences.

Playing Characters

While there plenty of other criticisms to be leveled at Heavy Rain, such as the exploitive sexuality of Madison Paige (as described by Denis Farr at The Border House) or the questionable portrayal of race (as summarized by Scott Juster at, yet again, Experience Points), there is much to be learned from its manipulation of the player-character gap.

The dissonance between player and character is often one of the major arguments against integrating storytelling into games. For many years it has been conventional design wisdom to strip personality from playable characters, so that they serve as empty vessels for players to graft on their personality. This gap, and how it is widened and shortened over the course of play, isn’t why games can’t tell stories – it’s precisely what makes storytelling in games interesting.