Procedural Storytelling

(I’ve been out of town for a bit, but a couple of weeks ago I discussed some of the initial PR around Valve’s upcoming co-op shooter. I’m going to turn this into a series discussing what I’ve learned about procedural storytelling, hopefully an area of interested and one of my passions and areas of game research.  I’ve unfortunately had little luck finding many experts in this field, so much of my work has relied on science techniques, trial and error, and analysis.  Forgive me if these thoughts are stale and don’t hesitate to point me in an stronger direction.)

This week, I saw this interview with Doug Lombardi:

“A lot of co-op games become very predictable and static after the first or second time you play through it. The AI Director is generating the population dynamically each time you play through the game. It’s also sort of monitoring your success rate and scaling the difficulty, based on if you’re doing well or if doing badly it’ll turn up or turn down, sort of, the action. And it’s also looking at pacing, which is something we found to very important in the creation of the Half-Life games. Trying to avoid too much combat and making the player feel sort of battle fatigued, if you will. Where they need to put the game down, instead we’re trying to schedule those breaks in the game… keep the game dynamically tailoring to the player’s experience whether it’s that difficulty setting, the battle fatigue and scheduling the pacing, etc.”

And I thought the AI Director was a great idea before! 2 thoughts: One, kudos to Valve for actually finding a way to do this. I’ve got a number of friends of there and it’s neat to see them working on this stuff. I’ve found procedural pacing control is still in it’s infancy, and in my experience most developers aren’t willing to take this kind of risk. Second, pacing is a word I haven’t heard in PR before. Why are they using a word associated with storytelling when story doesn’t seem to be the goal at all?  On some level, Valve wants to control the entire player’s flow through the experience, something that very few non-competitive games have done.  I have to admit my surprise to see this coming from the co-op multiplayer sphere rather then the single player realms that I have been focused on, but on reflection it makes sense. Why? I haven’t played Left 4 Dead yet, but to answer that I’ll have to compare it with the last comparable situation I can think of: Diablo 2.

In Diablo 2, a game designed for team play, there was a main game and then a replaying game.  In the runs through the main game, the experience was deliberately planned out, just as most single player games are. Even though it used dynamic spawning, it carefully followed the “easy-easy-hard” difficulty model to create dramatic arcs within the game and give the game a controlled pace and a tense experience. Co-op play increased the challenge a fixed amount, keeping the overall challenge static.  But once players had beaten the main game, they would continue playing by teleporting around and replaying the areas that would give them the best rewards, the “replaying game”. Since the fixed challenge throughout the main game is now too easy, the replaying game flow is designed to become player-controlled to accommodate the array of co-op team skills and character levels, despite leading to “farming” and losing all sense of story.And it was mighty fun! In fact, just like with World of Warcraft today, many fans don’t believe Diablo 2 really starts until you have beaten the main game. Diablo 2 showed that  variety, challenge, and pacing in the main game were not just the basic components of its narrative experience but key components of the game’s replayability, even after the designed narrative experience functionally ended. Players deliberately manipulated challenge, variety, and pacing to maximizing their performance, even at the expense of crafting dramatic arcs. Players, to wit, seem to prefer mechanically rewards, ie leveling up, rather then narrative rewards, but there are manipulatable controls that link the two.

This is significant because it shows a mechanic experiential benefit for storytelling beyond story, a game-y justification that players and designers are already engaged in. Diablo 2 players were taking control over challenge, variety, and pace. In a sense, these players were also creating their own player narrative alongside their maximally fun experience.  It was just a narrative the designers had to let go.  So if we choose to maintain these components it should keep the game replayable and cooperative and could also through a stronger narrative into the mix!  Left 4 Dead sounds like it’s taking a stab at this by addressing the following direct corollaries:

  1. Players are bad at forming dramatic arcs in their play, so we should create the structure of them for the player.

  2. Players will happily do things that destroy dramatic arcs, so we should not reward them for doing so.

And we are back to Doug Lombardi.  These goals both depend on pacing control.  Game structure is heavily based on controlled pacing – in that it requires some control over sequencing.  Rewards are directly linked to pacing, as seen in the psychological study of reward schedules.

Left 4 Dead appears to be tackling these goals using what I call “Encounter Scheduling”. On one hand, they are tracking the schedule of the players, their current narrative arc, and attempting to manage that using different numbers and difficulties of enemy spawns.  While the narrative here is probably quite primitive, it accomplishes the main goals, namely, building a cohesive line of play for the players, and providing challenge, variety, and likely tension as well.  In my research, I’ve found the narrative limitations here appear to be actually limitations of the game mechanics themselves, not the narrative form, and if the player has bought into the game this style feels quite comfortable.  Multiplayer deathmatches are a good example of similar simple narrative arcs and flows, bounded by their inherent mechanical depth.

Additionally, Left 4 Dead seems to eschew the traditional mechanics that break narrative flow, and consequently co-op replayability.  Level grinding seems absent, for example, as well as (hopefully) backtracking through “cleared” rooms to pick up items and player-created lulls caused by confusion or unfortunate design that suck the air out of most combat games. In the trailers, the game seems to constantly push you forward, encouraging you to engage with the narrative and the game mechanics it intertwines rather then “game” the story.

Narrative and the game mechanics it intertwines.  This startling statement is why I think we find the idea of procedural storytelling so compelling, why I think is likely to become a major component of future game designs and game art, and why I think Left 4 Dead could be so significant.

It’s worth pointing out that this doesn’t seem to be dynamic difficulty as we traditionally understand it, contrary to the press interview, and it brings with it a fair amount of innovation risk, so I’ll make this my starting point next time.  If the difficulty, a function of challenge over pace, is automatically being controlled by the procedural narrative, the “Encounter Manager”, what satisfaction of accomplishment is ultimately being earned by the player?

2 thoughts on “Procedural Storytelling

  1. Dan,

    Great question to finish on. I think you’re right about that sense of achievement. I play only Versus mode against humans now because I know if I win then my team did everything right and we deserved to get to the end. (It’s also cool playing as the infected.)

    Against the CPU, you can finish the level in normal mode as long as you don’t do anything stupid.

    That said, advanced mode against the AI is very different than in Versus mode — it’s a bit less frustrating and reinforces the team spirit much more. (In Versus mode, people quit and drop out more often than in Campaign.)


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s