Stream Pools: Space and Narrative Pacing in Games


Ian here—

I spent the first week of 2017 catching up on things I hadn’t played from 2016. But all play and no work makes Ian a dull boy, so it’s time to get back to writing, even if it’s of the casual sort.

Fair warning: In this post I’m going to dip into some unapologetic formalism as a way of best expressing some otherwise entirely subjective reactions. Obviously, there are pitfalls to this. Formalism puts off some. Unabashedly subjective attempts at criticism puts off others. But, whatever—this is my blog, and sometimes I like to post things that aren’t lesson plans. (Also, a note: I’m going to have fewer of those posted in the foreseeable future. I’ve posted most of my best lessons from past courses at this point, and I’m only teaching one class this term, one I’ve taught before.)

Below the fold, I play with some vocabulary, and offer thoughts on three more interesting games of 2016. These are short takes, and it is quite likely that I will be writing more on some of these in the near future.

Introducing stream pools

It’s been about 13 years since Henry Jenkins’ “Game Design as Narrative Architecture” was published.[i] Since then, it’s been republished in subsequent anthologies, as well as photocopied and scanned for god knows how many college courses. Considering its popularity, it is shocking how little work has actually followed up on it.

Oh, don’t get me wrong: the intervening years have seen plenty of writing on videogame space. We’ve seen articles and even full-length monographs on the subject by figures such as Bernadette Flynn, Michael Nitsche, Alison Gazzard, and Christopher Totten.[ii] But, to my knowledge—and I should stress that I absolutely could be wrong here, as my review of the contemporary literature has dropped off considerably since I finished my dissertation—there hasn’t been nearly enough work on the specific intersection of how game designers craft spaces to tell stories.

To offer up a more precise version of my complaint: What sort of vocabulary do we have to talk about (let alone teach) the pacing of games?

If we are discussing traditional literary narrative, ur-formalist Victor Shklovsky gives us the formal vocabulary of delays, stairstep construction, bound motifs, and free motifs to describe the ways in which obstacles and digressions slow down the progression of a story from its beginning to its ending.[iii] And, to a certain extent, we can use these terms to describe videogame narratives. (We can note, for instance, that videogames are fond of readily-transparent delaying tactics, setting up arbitrary roadblocks so that players must take some time to do some action in order for the events to progress. If I get asked to restore the power one more time….)

But we’re missing a vocabulary to specifically talk about how games can use space to modulate the pacing of their narrative. As one small means of correcting this, let me a term to add to such a vocabulary: stream pool.

I’m borrowing the term “stream pool” from environmental science. (That’s right, despite its extreme simplicity, it is apparently a bona fide technical term.) It describes exactly what it sounds like it describes: those deep, low-current pools that collect at certain points along the path of a meandering stream or river. They are often positioned at bends, after riffles, or partially separated from the main path of the stream by sand bars.

Google Earth image of Conneaut Creek, which I was well-acquainted with in the summers of my youth. The fatter bits are stream pools.

I propose using the term in the following way: a stream pool is a geographic nook in a linear videogame (or linear portion of a videogame) where players are encouraged to wander from the otherwise-apparent path forward, and explore some tangential features of a game’s geography.[iv]

Stream pools work best when there is some sort of reward to offer player in the game—say, health items, or crafting materials. Now, the use of rewards to direct player movement into specific corners of a game’s geography is one aspect of videogame space that has been already been analyzed in existing literature (for instance, the works by Gazzard and Totten mentioned above). What I want the term “stream pools” to recognize, though, is that the design of videogame space is not always simply about risk/reward mechanics. It is simultaneously about pacing the narrative experience of a game. It is about allowing the player to indulge in a delay, including whatever optional delays fit their particular play style and pacing needs.

Some of the most archetypal examples of stream pools can be found along the “Highway 17” area of Half-Life 2 (Valve, 2004). During this stretch of the game, players drive a small buggy along the titular highway. At regular intervals, they are interrupted from this drive, and invited to get out of the buggy and further explore the location on foot.

Sometimes, leaving the vehicle and momentarily wandering from the highway’s clear forward path is required to progress in the game. At certain points, players must clear blockage from the road. At other points, enemies must be cleared from Combine checkpoints, and players must discover how to open certain gates (which often require solving a small puzzle).

Other times, though, the areas that can be explored on foot are pure tangents, unnecessary to the forward thrust of the game’s events. Sometimes an abandoned house along the highway will provide only some spare supplies and a vista of future areas as a reward, rather than the opportunity to overcome an obstacle.

Here’s a typical example of the geography of Highway 17, taken from one of the maps in the “Sandtraps” chapter. The wrecked cars and the checkpoint are more-or-less required stops (although it is technically possible for the player to speed right by them without leaving their vehicle, if they’re anticipating events and adept at piloting the buggy). The abandoned house is not. It provides an optional distraction, something that curious players can explore if they need a momentary escape from the game’s otherwise unrelenting forward trajectory. If they poke around in it, they’ll be rewarded not only with supplies, but also a tiny jump scare, when they are unexpectedly ambushed by some rollermines that fall from a hole in the house’s ceiling. It is purely there to add flavor to the game’s world, to give a sense of a horizon of marginal spaces.


What really makes the stream pool design of Highway 17 work is that first-time players will rarely be entirely sure whether getting out of their vehicle to explore a given pool will be necessary or not. Roadblocks and ambushes are often well-hidden enough that they can take players by surprise, and so it quickly becomes learned behavior to get out of one’s vehicle whenever one sees a building around, and approach trepidatiously, looking for supplies, enemies, and a lever to defuse an as-yet-unseen trap. This subtly encourages a stop-and-smell-the-roses approach to these chapters of the game, modulating player behavior so as to modulate the pace of game events.

These days, is especially common these days to see stream pool design in games that emphasize survival, while still being linear enough to avoid falling into the sandbox survival genre. Games such as Metro 2033 (4A, 2010), The Last of Us (Naughty Dog, 2013), and, for something otherwise entirely different in tone, Shelter (Might and Delight, 2013) foist a scavenging mindset upon their players. Stopping to search for scarce food, ammunition, or crafting materials will frequently take precedence for players over moving toward the next story-mandated point on the map. In their cultivation of cautious and sharply observant players, such games are able to slow down their pace without throwing an endless stream of manufactured obstacles at their players. Player curiosity about an unexplored bit of the map turns out to be a more elegant delaying device than asking them to fix a generator for the umpteenth god damn time. (Which is not to say that you don’t also have to fix generators in at least two of the above-mentioned games. Because of course you fucking do. I’m just grateful that you don’t have to fix a generator in the game where you play as a badger.)

Alright, so, that’s the basic gist of stream pool design. Now, I’d like to use this term to talk about three games from 2016. Again, these are just short takes, possibly to be extended in future writings.

35MM (Sergei Noskov, 2016)

Ah! The impact of Andrei Tarkovsky’s Stalker on Eastern European post-apocalyptic games continues. It really is interesting to see both the American and Eastern European post-apocalyptic genres mature and diverge, with the Americans creeping away from the starting points of Matheson and Miller, edging towards McCarthy, and the former Soviets digging out their own territory between the Strugatskis, Tarkovsky, and Glukhovsky. And my god, was the Tarkovsky angle strong in this one. When I first heard the flute lines of Dmitry Nikolaev’s score, I could have sworn I was listening to Eduard Artemev’s opening theme for Stalker.

Just as Tarkovsky took much of the conventional action out of the Strugatski’s book and transformed its skeleton into an opportunity for religious contemplation, Noskov’s 35MM seems like an attempt to re-work Metro 2033 as a much more sedate affair. The scavenging mindset of the latter persists (as well as the occasional need to disarm a deadly trap set by previous occupants), but much of the combat has been removed. In its place, there is walking through rainy forests, and taking photos with your character’s titular 35mm camera.

35MM also takes stream pool design and sharpens it into something frightfully pure. In Metro 2033 or The Last of Us, scavenging an abandoned house or railway officer’s quarters is a short respite from the combat that the march forward is otherwise sure to bring. In 35MM, scavenging is just a break from your long walk. Walk along the road, scavenge in the nearby buildings. Repeat. Some might find this to be monotonous, but I found it to be admirably pure.

The clip below captures this flow in its most elemental form. I walk along the road with my partner. He calls out a nearby house, in a move that gives players permission to stray from the path and explore, if they want. Nearly all of the houses along the road are open, and contain some sort of item to scavenge. Your partner will wait patiently as you do so. If you continue walking forward, though, you’ll discover that there’s no fuel for the railcar. (A “fix the power source” moment!) This is a correction: You’ve been going too fast. The game wants you to slow down. It wants you to explore every nook and cranny. It’s back off the path for you. The story that’s being told is a slow one, and you’ve been rushing it too much. Time for some delays.

I’m intending to write some further thoughts on this game later. [UPDATE: I didn’t write more, but I did create this.]

Firewatch (Campo Santo, 2016)

Okay, so let’s get this out of the way first: Firewatch is not perfect. In particular, the writers overshoot while playing up the conspiracy thriller/psychological horror angle in the game’s second act. I found the gradual ratcheting up of paranoia to be superbly executed, but there’s simply no denying the fact that they tie up these narrative threads far too abruptly at the game’s finish, and fail to adequately deliver on what came before. For as much as I like the game’s ending—and I do, in fact, think that it’s thematically rich in some ways, putting a nice coda on some of the game’s threads of characterization—it simply cannot perform the tasks it needs to, and reveals some structural weaknesses earlier in the game that weren’t previously apparent. It gave me a few moments of “fridge logic” after playing, which left a bad aftertaste in my mouth after what had otherwise been a highly enjoyable four-and-a-half-hours or so.[v]

That said: Firewatch is very, very good, and frequently outright stunning. One of the reasons that its storytelling sins only become apparent at the end is because the rest of the game is so well-paced, hitting a near-perfect balance in its first few hours between delivering necessary character exposition, allowing for moments of player relaxation in the game’s gorgeously-rendered environments, while at the same time gradually increasing tension, one that is at first amorphous but later becomes fever-pitched.

Part of what makes this work is the design of the game’s space, which is generally based around a path-basin-path principle. That is: players will often be constricted to a fairly narrow path leading from one location to another, only to have that path open onto a larger basin, that allows for some more wandering and taking in of the sites. After awhile, though, they’ll discover that getting to the next location requires them to travel down another fairly closed-in path.

This helps spice up the game’s linearity with occasional bursts of the feeling of openness (whether or not that feeling is really warranted). The game is especially effective at making the transition from basin back to path not feel arbitrarily restrictive.

The clip below illustrates this well. It’s somewhat indulgently long, but that fits with the leisurely pace of the game’s opening hours. In it, I am attempting to make my way over to a plume of smoke that has appeared in the distance. For the first 45 seconds of the clip, I’m in a canyon, and so there’s no real way forward other than to follow the stream out to the mouth of the canyon. By the 1:30 mark, things have opened into a meadow, and I’m given more of a chance to plot my own way. You’ll notice that I attempt to take the most direct route possible, only to discover that it is blocked by an outcropping of rock.

This doesn’t feel restrictive because that’s how natural landscapes work. Hikers who get overly-ambitious and try to take more direct routs than offered by the paths will often be thwarted, and discover that the path takes the direction that it does for good reason. The game is admirably consistent in its presentation of Henry’s effectivities, and the rope-breaking scene early on does a good job of establishing the fact that he resides within a recognizably human body, capable of getting injured. Little cues like this help us buy the obstacles of the landscape as realistic details rather than designer-imposed restrictions.

The landscape design impacts the pacing not only of the player’s exploration, but also of Henry and Delilah’s conversations. Henry’s calls to Delilah are always prompted by the player, often by placing a cursor on something in the immediate vicinity. Players wishing for a slower, chattier experience are thereby encouraged to explore around a bit more, looking for things in the environment that Henry can ask Delilah about. (What really makes this work is the fact that the dialogue unlocked in these moments is so rich. Some of the moments in the game’s script I laughed aloud the most at came in the form of these completely optional interactions.) Delilah’s calls to Henry, on the other hand, tend to happen during the “narrow path” moments of the game, when the game not only knows exactly where the player is, but also where they’ll likely to be as they walk along the path during the duration of the conversation. This gives the game a nice “path=talk,” “basin=silence, unless the player wants to investigate certain things” flow, that moves the narrative events forward while allowing the player not to feel rushed.

Not all games successfully circumvent making their players feel rushed … as we’ll see below.

Virginia (Variable State, 2016)

During my first hour of playing Virginia, I found myself wondering why it was leaving me cold. I couldn’t seem to muster any explanation. I have a well-documented love for Brendon Chung’s Thirty Flights of Loving, so one would think that Virginia‘s expansion of Chung’s experiments with in-game cinematic editing would intrigue me just as much. I love Twin Peaks, and the amount of hours I spent obsessively scouring the details in Deadly Premonition (Access Games, 2010) shows that I also have a taste for games inspired by Twin Peaks. I’m generally on board with experiments in interactive narrative design.

Now, don’t get me wrong: there are parts of Virginia that I felt were absolutely fantastic. Given your tolerance for prescriptive, “you will feel this emotion now” program music (and, admittedly, mine is exceptionally high), I think there’s a case to be made that it possesses the finest musical score of any videogame ever made—besting Austin Wintory’s score for Journey (thatgamecompany, 2012). I found the story’s unexpected dive into the moral turmoil of a woman of color working as an FBI internal affairs investigator to be not only a stunning direction for a videogame narrative to go, but also deeply affecting (spurred on, no doubt, by that prescriptive score). On the whole, I would say that I found its final-act turn into unreliable, impressionistic, “what-if” storytelling to be refreshing and bold.

I’ve decided that I have to mull the game over some more to come up with my final thoughts on it. (This might be a future post.) Here, though, I do want to address that niggling feeling throughout the first half of my playthrough, that by all rights I should be enjoying the game more than I was.

At first, I thought that I was being held back by the degree to which Virginia did not meet my generic expectations. I knew it was going to be a narrative-heavy art game, inspired by cinematic and televisual form. Before going in, though, I had also assumed that it was going to be more of an adventure game than it was. Adventure games have been tackling narratives of investigation at least as far back as Deadline (Infocom, 1982) and The Curse of Crowley Manor (Adventure International, 1981). Such investigatory yarns have provided grist for some of the most ambitious games of the past twenty years, from Shenmue (SEGA AM2, 1999) to the aforementioned Deadly Premonition. I was expecting Virginia to take bold risks, and to plot new directions forward for videogames as a narrative art form. What I wasn’t expecting was it to throw out all of the strengths that mystery-centered adventure games had built up over the past few decades. I have to admit: every time I encountered another “click to progress to the next vignette” moment, I felt a pang of desire to do some good old fashioned poking around for clues.

This, at least, was my first impression. Upon considering the matter further, though, I decided that it wasn’t the lack of clue-searching per se that was the problem. Rather (as you might have guessed, given the topic of this post), it was the lack of stream pools.

For me, the game really first perked up during the second trip to the Fairfax residence, when Agents Halerpin and Tarver break in and snoop around, and Tarver finds the box in Jared Fairfax’s desk drawer. Initially, I assumed that I enjoyed this bit because it was the first time that I, as Tarver, actually got to do something that felt like real investigating.

But, later on, I got the same tingly pleasure from exploring Agent Halerpin’s apartment. This wasn’t connected to the missing-person case at all, so it hit me: What I really wanted was not a chance to wander around and find clues, but just a chance to wander around and observe, at all. Being able to open doors on the way down the stairs of Halepin’s apartment, to poke around in side-rooms and get a sense for my partner’s personal history, her politics, and her role as her ailing mother’s caretaker, was a pleasure in and of itself.

It struck me at this moment how different a beast pacing in videogames is from pacing in other forms of audiovisual storytelling. Judged by the standards of, say, television, the first hour of Virginia wasn’t rushed at all. It had contemplative character beats. It had dream sequences. It had seconds-long scenes in which characters were doing nothing more than walking down a hall. Beat for beat, I would venture that less stuff happens in the first hour of Virginia than in the first hour of the Twin Peaks pilot.

But videogame pacing isn’t really about the amount of stuff that happens. It’s more about how much control the player has over the flow of said stuff. And, judged by this standard, the first half of Virginia is furiously paced. There are so many moments where the simple act of walking forward will prompt a scene change, or where the only option given to players is to click on a single thing to prompt the next vignette. As quiet and thoughtfully-written as each of these moments might be, players are still hurtling down a tightly-constrained corridor. I would go so far as to say that the game engenders a sense of claustrophobia.

The second trip to the Fairfax residence is a nice respite from this, and it is followed up by two more respites: The Sojourner’s Truth bar, and Halerpin’s apartment. It’s astonishing how much just the simple act of being allowed to safely walk around without prompting a scene transition creates breathing room, and gives players a feeling of the pace being necessarily slackened for a bit. And it’s unfortunate that such moments didn’t arrive until 60 minutes into a game that will likely take players somewhere in the range of 105–120 minutes to complete.

Virginia‘s creators, Jonathan Burroughs and Terry Kenny, are extremely upfront about getting their inspiration from Virginia from Thirty Flights of Loving, to the point where an acknowledgement of that game is the first thing we see in Virginia‘s closing credits. It seems to me that the weaknesses of Virginia spring from a mis-remembering of Thirty Flights. Sure, the cinematic cuts are flashy, as is its relentless push forward into hallucinatory reverie. But it also has its share of stream pools. In between moments when we’re walking forward to prompt the next cut, there are also moments when we can take a breath, enjoy the view, engage in some activity for a bit without worrying about accidentally skipping forward into a new scene. We can hang out with Anita on the balcony, eating oranges and watch balloons drift through the tangerine evening sky. We can explore the coda’s museum at our own pace. And, crucially, the game actually begins with one of these slower, exploratory moments. The bootlegger’s hideout is full of things to poke around with and examine at our own speed. Thirty Flights reaches the heights that it does not only because of its experiments in visual style, but also because of its expert pacing.

Anyway, again: I’ll likely have more thoughts on this later. [UPDATE: I did! A lot more thoughts.] That’s it for now.

[i]. Jenkins, Henry. “Game Design as Narrative Architecture.” In First Person: New Media as Story, Performance, and Game. Ed. Noah Wardrip-Fruin and Pat Harrigan. Cambridge, MA: MIT Press, 2004.

[ii]. Bernadette Flynn’s articles and anthology chapters on the subject are numerous. As far as monographs go, I was thinking specifically of Nitsche, Michael, Video Game Space: Image, Play, and Structure in 3D Worlds (Cambridge, MA: MIT Press, 2008); Gazzard, Alison, Mazes in Videogames: Meaning, Metaphor and Design (Jefferson, NC: McFarland, 2013); and Totten, Christopher W., An Architectural Approach to Level Design (Boca Raton: CRC Press, 2014).

[iii]. Shklovsky, Victor. Theory of Prose. Translated by Benjamin Sher. Normal, IL: Dalkey Archive Press, 1990.

[iv]. This subsequently leads to the question of how, exactly, I’m defining a “linear” game. My definition of it more or less hews to the description of Half-Life‘s construction offered by Ken Birdwell in his famous postmortem “The Cabal”: “If the players are in the mood for more action, all they need to do is move forward and within a few seconds something will happen.” A linear game is a game in which there’s really only one direction forward, and it is fairly evident which direction that is (i.e., it’s not the direction players came from.) Moving along this direction will prompt the events that lead the game to its conclusion.

[v]. My biggest problems here (spoilers): Once Henry finds Wapati Station, I can believe that his and Delilah’s mutually-reinforcing paranoia leads them to concoct elaborate theories about how Wapati Station been specifically designed to spy on them, rather than just being a geological survey site. (I love the small details of how this paranoia affects the game interface, too, as when the captions of certain items change after we hear Delilah’s outlandish theories about what they’re used for.) This, for me, work as an elaborate red herring.

But unfortunately the whole setup for the “mutually reinforced paranoia makes our protagonists spin out an elaborate conspiracy theory” relies completely on Delilah being initially unaware that there’s a research site at Wapiti meadow. And, in hindsight, I just can’t buy that. The research site is obviously well-funded, probably sponsored by the US government or a major university. As Delilah notes, it obviously took some real work getting the equipment in. And the activities undertaken there are presumably neither nefarious nor  secret. This being the case, I cannot believe that Delilah didn’t even know that there was a fence constructed around the area. It’s a detail that is obviously included to get the player wrapped up in the characters’ panicky conspiracy theories, but it doesn’t fit. It pushes Delilah’s ignorance too far. By the time the truth is revealed that the spying was being done by Ned Goodwin, primarily from his own separate bunker, the whole Wapati red herring feels like cheating.

Nitpicky, maybe, but as I said, it tainted some of the awe I had at the end of the game with some grumbling.

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 )

Facebook photo

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

Connecting to %s