August 6, 2003
Interaction and Agency
I just played Dead Reckoning, the interactive fiction (IF) Nick recently translated. Playing this IF reminded me of something I’ve been thinking about for awhile, the relationship between interaction and agency. Before continuing, let me provide a preliminary definition of these terms in the context of new media. By “interaction” I mean the act of physically manipulating an input device (e.g. wiggling a mouse, moving in front of a camera, etc.) and eliciting a response (e.g. an image changes on a screen, motors turn on and off, etc.). Interaction is an abstract concept, saying nothing about the character of the relationship between input and elicited response, just that there is some relationship between them. Agency is a phenomenal category, describing what it feels like as a player/interactor to be empowered to take whatever actions you want and get a sensible response. That is, an experience is productive of a sense of agency if it supports the interactor in forming intentions (based on what’s happening, the interactor can think of something they want to do), taking action with respect to these intentions (there is a way to express the action the interactor wants to take), and interpreting the response in terms of the intention (the system’s response makes sense with respect to the intention). Given these definitions, the question that interests me is whether the sole function of interaction is to produce a sense of agency, or whether interaction can yield other, equally interesting phenomenal experiences.
Some of the previous posts on GTxA have held the position that agency is king, that indeed the design effort in building an interactive experience should be directed towards maximizing agency. For instance, in I Can’t Get No Satisfaction, Andrew describes his dissatisfaction with current interactive story forms precisely because current forms fail to provide a strong sense of agency. If it is indeed true that agency is the only phenomenal category produced by interaction, then it follows that every interactive experience could be laid on a one-dimensional axis, judged by how well or poorly it generates a sense of agency. Playing Dead Reckoning reminded me of other functions of interactivity besides the production of agency. Like most IFs, Dead Reckoning disrupts agency. At any given moment during the experience, only a small subset of all the possible actions (verbs) produce a meaningful response; the rest produce a response letting you know that the action is currently not a useful action to take. In fact, part of the craft of authoring IF is authoring the huge number of “you can’t do that” messages such that they stay fresh, are entertaining, and potentially provide hints towards what actions are currently profitable. This property of having a large number of verbs of which a small subset is active at any moment is disruptive of agency in that many of the intentions a player can form are blocked; the IF world supports the formation of intentions without supporting the resulting actions. Yet I find IF enjoyable to play and aesthetically interesting. So, interaction must be inducing some other phenomenal experience than the feeling of agency. And this experience is rooted in the solving of puzzles, which in Twisty Little Passages Nick links to the literary form of the riddle. While Nick focuses on the formal qualities of puzzles/riddles, here I’m interested in their phenomenal qualities, what solving a puzzle/riddle feels like. In solving a puzzle the player is figuring out the logical structure of the IF (recapitulating the author’s view of the IF), problem-solving (and experiencing a sense of mastery at solving the problem), in literary IF experiencing a transformation of their understanding of the world (Nick describes this as the function of the riddle as a literary form – through its solution to offer a transformed vision of reality), and in story-based IF resolving the tension of what happens next in the story. While this list doesn’t exhaust the phenomenal functions of the puzzle/riddle, it is enough to reveal that these experiences are different from agency; interaction is generating different phenomenal/aesthetic experiences. IF certainly offers some agency (more than hypertext, for instance, which is even more removed from a concern with agency), but its strength is in its exploration of the phenomenal qualities of the puzzle. It is interesting to consider what other phenomenal qualities interaction may generate in other new media forms.
August 7th, 2003 at 3:05 am
Oh, I like this approach, Michael. So you’re suggesting that interaction is a technical, um, fact, that can have several different phenomenological effects in users, such as agency and puzzle-solving.
You define agency as feeling “empowered to take whatever actions you want and get a sensible response” – so you’re not limiting it to being able to affect the outcome of the story or being able to change the fictional world. I have the impression some people use agency that restrictedly. But you’d allow agency to include my feeling of satisfaction that if I click the “comments” link I get a box I can type a comment into and when I click “Post” I get to see my comment on the website, after your post, right?
See, if that’s not included in “agency”, that kind of adding your own stuff would be an obvious phenomenological pleasure of interactivity. Or rather, a possible pleasure since interaction does not always allow this.
August 7th, 2003 at 1:01 pm
Hmm, I like the idea of teasing apart the different pleasures / phenomenal effects of interactivity. A few thoughts come to mind.
If the user can perform some action and get even the minutest response, strictly speaking I think that could be considered agency. The question becomes what kind and how much. A useful characterization, one that Michael and I often use, is the idea of local and global agency. If the user can press a button in a GUI, and the even if the button simply just animates being pressed and has no other effect (a dead button), we could call that a teeny bit of local agency; you did have some effect on the world, however brief and inconsequential. Its effect is only local; it has no effect on the global shape of events. Also, if the button has some effect in the very short term, but has no longer term effect, that’s also local agency. But if pressing that button changes the overall state of the system in some way that has ramifications later on, then the interaction also has some global agency. How you quantify how much or little local and/or global agency something has is a further question.
Like most IFs, Dead Reckoning disrupts agency. At any given moment during the experience, only a small subset of all the possible actions (verbs) produce a meaningful response; the rest produce a response letting you know that the action is currently not a useful action to take.
In effect that may be true, but I doubt IF authors typically intend that. I assume that ideally an IF would be responsive to all actions at all times? So to identify that as a quality of IF is actually to identify what could be called a limitation of the current form of IF?
It’s interesting, the other effects you describe — problem solving, figuring out the logical structure of the story, that transforms your understanding of the story world — are effects that can be achieved in non-interactive fiction, such as good crime or mystery fiction. (Perhaps that’s why people often refer to books and movies as “interactive” in a certain way.) Obviously the puzzles in a digital story are literally manipulatable, like a rubik’s cube, versus the purely conceptual puzzles in a non-interactive mystery. So the puzzle pleasures in IF can certainly be more “hands-on” than in a book. But the pleasures are similar, no? That’s not a problem per se, but it’s not as fundamentally new as the pleasure of agency.
Are there other pleasures here I’m missing? With IF (and everything else) undoubtedly the whole is greater than the sum of the parts. But personally I enjoy IF most when it offers the pleasures of agency, not just the pleasures of puzzle solving.
(A related question: Can one get any pleasure of agency from a book? Assuming it’s not a choose-your-own-adventure, a form of IF.)
And, in the case of primarily linear IF, if “resolving the tension of what happens next in the story” really just means advancing to the next room / level — isn’t that essentially turning the page, a glorified next button? … Perhaps not. Perhaps the sheer act of participating in the world to advance the story, even linearly, can be pleasurable. But that pleasure is greatly diminished if you think about the fact that there wasn’t much else you could do (little or no agency). Perhaps it’s pleasurable because it gives the illusion of agency. If the illusion works, it works.
Also, not to forget what Janet Murray identifies as 2 other phenomenal effects of interactivity, in addition to agency: immersion, and transformation (taking on a role and becoming someone else). And check out this recent blog post about definitions of interactivity on miscellany-is-the-largest-category.
August 7th, 2003 at 3:23 pm
In effect [having a small subset of verbs produce meaningful responses] may be true, but I doubt IF authors typically intend that. I assume that ideally an IF would be responsive to all actions at all times?
Not necessarily. One of the challenges of current IF is subtly directing the player, to give them a goal. One way of providing this direction is through limiting actions.
In the future it may be possible to create a fully simulated world, or at least one with a much larger range of actions. This future IF may move towards the simulation/toy elements of a Sim City, may require new techniques of guiding players other than limiting a range of actions, or dynamically creating interesting storylines around the player’s actions.
August 7th, 2003 at 3:42 pm
Damn it, you guys got me trying to play Dead Reckoning, and I’m stuck in this AWFUL endless queue, with a woman who ignores me, precious few memories and I’m sure that I need to find a different verb from “wait” to get anywhere…
Has anyone theorised anti-agency yet?
August 9th, 2003 at 11:52 am
My definition of agency offers involves the interactor forming intentions, expressing those intentions, and understanding the system response in terms of the intention. And to form an intention, the interactive experience has to offer reasons to form intentions (e.g. “There is a monster coming towards me, I want to run”) offer “verbs” for expressing those intentions (e.g. “If I move the joystick to the right, I’ll move that direction”), and offer a response commensurate with that intention (e.g. “Look, my avatar is moving away from the monster – I am succeeding at running away”). Agency is diminished if the system fails to support one or more of these three relationships.
Jill writes:
But you’d allow agency to include my feeling of satisfaction that if I click the “comments” link I get a box I can type a comment into and when I click “Post” I get to see my comment on the website, after your post, right?
So, in Jill’s example, part of the pleasure of posting is agency because the system supports forming intentions (“What I’ve just read makes me want to respond”), taking action (“Ahh, there’s a response button”), and receiving a commensurate response (“There’s my comment”). In addition to agency, there are other phenomenal pleasures supported by the interaction of posting a comment, such as communication with other people (I wouldn’t really want to subsume the whole social sphere under “agency”) and working through your own thoughts.
Andrew writes:
If the user can press a button in a GUI, and the even if the button simply just animates being pressed and has no other effect (a dead button), we could call that a teeny bit of local agency; you did have some effect on the world, however brief and inconsequential.
Without some larger framing for this button there is no agency, not event a teensy bit. The mere fact of having some effect on the world puts as back in the realm of interaction as a technical fact (I like this expression Jill), not in the phenomenological world of agency. Without a reason to push the button, there’s no agency. If, on the other hand, the button were shaped like a light switch, and there was a representation of a bulb on the screen, and clicking the switch causes it to animate up and down and turn the bulb on and off, now we’d have a teensy bit of local agency. What’s interesting about phenomenological properties such as agency is that they depend on what’s going on in the interactor’s head, on what’s communicated between the technical system and the person, not only on technical facts like counting the number of system actions that are available at each moment.
I’ve written more on my view of agency, relating it to Brenda Laurel’s take on Aristotle’s poetics and Janet Murray’s phenomenological categories here:
A Preliminary Poetics for Interactive Drama and Games. 2001. Mateas, M. Digital Creativity, 12 (3), 2001. 140-152.
Also appeared in Proceedings of SIGGRAPH 2001, Art Gallery, Art and Culture Papers, 51-58. 2001.
August 10th, 2003 at 7:02 am
In a recent review of Dennis Jerz’s “Fine Tuned”, I suggested that a good deal of the humor emerges from the inadvertent results of the player’s actions — in other words, Jerz is playing with the sense of agency and, sometimes, twitting him amusingly on how things don’t work out. (The full review, which contains abundant spoilers, is to be found near the end of the latest SPAG.)
I’m not sure where to fit this phenomenon into the scheme, but it does suggest to me that forming and expressing an intention that doesn’t work out can produce its own pleasures within the realm of IF. As I describe in the review, Jerz sets up a bunch of pratfalls for the player to take, which it is easy to fall into. The player becomes the straight man.
At a less complex level, the refusal messages for inappropriate actions in IF are sometimes quite entertaining. So the intention is understood by the parser, but not carried out; at this point, perhaps the pleasure is a bit like the pleasure of making a joke. The player suggests something stupid or impossible; the parser gets it, and retorts.
I am not thinking of any good examples of this kind of interaction in other genres, but that doesn’t mean they don’t exist.
August 12th, 2003 at 1:52 pm
While I’m sure I failed as often as I succeeded, in trying to make the parser in on the joke, so to speak, I was trying to avoid making the player feel the frustration that Jill felt and that I myself felt when stuck in that endless line with nothing (more) to do. I tried to create a few timers in which NPCs said or did things that suggested more and more precisely what you are supposed to do. My feeling is that I managed to train the player to pay attention to those NPCs, who revealed character and backstory along with hints. Also, for people who were rushing through a game, they could jump right ahead and accomplish certain tasks quickly.
I’ve often thought of creating an in-game hint system where an NPC permits you to skip difficult puzzles, but only after verbally abusing you. My thought was that if the insults were funny enough, then some people would play the game just for the insults.
August 12th, 2003 at 3:46 pm
Jill writes:
Damn it, you guys got me trying to play Dead Reckoning, and I’m stuck in this AWFUL endless queue, with a woman who ignores me, precious few memories and I’m sure that I need to find a different verb from “wait” to get anywhere…
It’s interesting that playing IF is a skill, something that you can become better at over time. I don’t know if Jill plays alot of IF and is just stuck on this puzzle (which happens to me all the time) or is relatively new to playing IF and struggling with more fundamental conventions, but her comment made me think about the non-transparency of IF, the (tacit) conventions one must learn to be able to enjoyably play it. Though I’m not a hard core member of the IF community, I’ve played IF enough that I somehow “get” how to play it. Dead Reckoning took me about 35-40 minutes to play (meaning, to get to “You have won”). This isn’t particularly fast (I wasn’t trying to speed through it or anything) but is rather a somewhat typical play time for short, story-based IF work (like competition pieces). But knowing how to move through such works is certainly not “natural” or innate. I remember listening to Nick trying to convince hypertext poet Stephanie Strickland to try IF at DAC a few years ago. Stephanie’s reaction was basically “What’s the deal with this prompt?”, “What should I type?”, “Why would I want to go north?”, and so forth. Learning these conventions is more than just needing hints – it’s actually a different way to read and write (I’m talking about writing in the context of playing, not authoring). If there were classes on how to “read” IF in the same way that there are classes on how to “read” movies or literature, I wonder what the content of these classes would be? Has anyone made explicit the tacit reading and writing conventions that must be learned to successfully play? I think part of the difficulty of making these conventions explicit is the supposedly natural second-person identification with the narrator (ie. I type in what I want to do in simulated world and the narrator does it and describes what happens), which creates the pretense of no convention. But even classic IF violates second-person identification in all kinds of interesting ways, and plenty of modern IF explicitly manipulates the relationship between player and narrator. I think it’s fine that playing IF is a skillful activity – that’s what makes it an interesting hybrid between literary form and game. But I think that the language of simulation (coming from the simulated world aspect of IF) disguises, and thus makes harder to tease out, the conventions that operate to make IF “readable”.
August 12th, 2003 at 4:36 pm
I think the idea that “playing IF is a skill” is great. However I don’t interpret the kind of frustration Jill is having as a “skill” that needs to be learned. What’s happening is here is an overall design issue / problem that is very common in IF. This is probably the major reason why I don’t play IF very often in its current form. I seem unable to enjoy the meta-puzzle of figuring out valid commands. But even more importantly, I feel an allergic reaction to the idea of a limited, somewhat arbitrary set of commands, that sometimes work, sometimes don’t work. It’s frustrating to me on a holistic, aesthetic gameplay level. (Perhaps not all IF’s suffer from these problems; I admit I’ve not played nearly as wide a range of IF as some of you, but virtually all of the IF’s I’ve tried have these issues.)
IMO, generally speaking it’s not right to expect players to learn conventions that stray too far from “naturalness”. I’m a firm believer in interactive experiences that require no manual, no learning of how the interface works (beyond the most basic digital literacy, such as knowing how to type, move a mouse pointer or joystick). Perhaps it’s the design purist in me, but I think this viewpoint aligns with how the untapped mass-audience of non-gamers feels, or would react.
However I totally realize how difficult it is to avoid this problem. The interactive drama I’m working on with Michael, Facade, by no means solves it. I think Facade’s design alleviates this problem somewhat, or perhaps shifts the problem around a bit. (I apologize for using Facade as an example, since it’s not yet available to play. But it will be soon, we promise.)
So I don’t exactly blame any IF authors for this problem, and think the kinds of solutions Dennis describes are excellent. But, for my money, it’s a major factor in preventing me (and probably many others) from enjoying existing IF.
It comes down to a fundamental UI principle: if you give players an interface, ideally you need to support the full range of inputs possible with that interface. Games typically have tighter designs than IF because they limit the UI to simple physical moves (run, jump, shoot); IF laudably opens up the interface to language input, but at the expense of not being able to handle most of the things you can say. (The in-between is menu-based dialog, which works from a UI-design standpoint, but whose arbitrariness creates feelings of frustration just as painful. Also, hypertext fiction, a cousin of IF, explicitly shows you the possible choices, in the currently available links. It has similar limitations, but at least avoids the meta-puzzle of searching for what works to move forward.)
What is the real solution? Parsers and AI that can interpret a wider range of expression from the player, and more generativity from the story system to handle all of the new responses required for this wider range of expression.
Back to work…
August 13th, 2003 at 11:28 am
The command-line interface is definitely a constraint. Nick is probably the one to speak knowledgably about the peculiar virtues of constrained writing… why else would anyone write a sestina, other than to flex his or her creative skills… and who would really appreciate those creative skills, other than people who are trained to read the sestina? When Emily created her NPC Galatea, she constrained the form of IF by, for instance, restricting the PC to one room and completely doing away with the inventory. There’s only so much that a Lone Coder can accomplish, particularly when we’re talking about a labor of love. But the existence of what we can call the standard skill set — the IF player’s knowledge that, for instance, you can’t just type “give me the hammer,” you have to type, “Thor, give me the hammer” — makes individual innovations stand out.
Regarding IF in education… I start by asking my students to play “Pick Up the PHone Booth and Die,” which is a game that you can win in one move… while the parser offers a large number of realistic and/or humorous responses, of course students run into the problem of the stupid parser, but the game is short enough that they aren’t frusrated for long. Then I move on to the Hitchhiker’s Guide to the Galaxy, which has lots of very funny refusal messages (that is, entertaining variations on “you can’t do that,” often containing at least a subtle hint). And all the death messages contain a hint at what you should have done to avoid the death.
After about 15-20 minutes of HHGG, I move on to Adam Cadre’s 9:05, which relies upon the player having been trained to follow IF conventions, but turns an important convention on its head. (I won’t give the spoiler!) Anyway, in the last few minutes of the class, I tell the students what to do in order to uncover the hidden meaning of the text, and usually they go “Ewwwww!” in a way that I’m sure would gratify Cadre if he were to witness it. Then, when I come back to IF later, thy are ready to analyze the text more thoroughly. In past classes I’ve asked them to look at Granade’s Losing Your Grip, Short’s Metamorphoses, and Fine-Tuned (though I don’t tell them who wrote it… my goal is to show them how to accept criticism graciously, and I wouldn’t want them to feel uncomfortable telling me where they think a game sucks).
Back in the day, when IF was fresh and new, the idea that you could actually type words to the computer and you could type back (compare to the cryptic letter and number codes for early 70s games like Wumpus and mainframe Star Trek) was considered amazing. Will Crowther invented the genre so that his young children (ages 5 and 8, or thereabouts) could play it with him. The entertainment value of simply typing in something in order to see what comes next is really no longer capable of holding someone’s interest (certainly not when there’s a blog to feed and comments to reply to!)
Stephen Granade has written a thoughtful piece on whether IF needs AI… He concludes “no”, but one of the reasons is that we don’t yet have AI good enough.
August 13th, 2003 at 2:32 pm
Even watching a movie or looking through a newspaper (even for a literate person) is a skill, as McLuhan reminds us. But there’s another issue: As Andrew points out, even those with the skills appropriate to the form can be frustrated by poor design. Still, I’m not so sure that the “constrained” command line [*] requires IF authors to handle every possible text.
Andrew writes, “if you give players an interface, ideally you need to support the full range of inputs possible with that interface.” But graphical adventure games don’t allow you to meaningfully click on any region of the screen; hypertexts (with a few exceptions, Lust being one) don’t let you click on every word; Doom and Quake don’t let you use every key of the keyboard to allow you to control your avatar. Nor is every possible Command-, Control-, Option-, or Alt- sequence a valid option in Word. (Or even in emacs, despite years of trying.) With very few exceptions the workings of computer interfaces are constrained in part by convention, so that some inputs that are allowed are meaningless.
We should never expect IF to have to handle the command “donate organs to science” just because people can type what they want, any more than we should expect Facade to understand us when we type in Hindi. The frame of interaction in IF requires that we type commands to a player character as if instructing an actor on a stage to do physical tasks in the immediate surroundings.
Now, this may all actually be pointing out a flaw in Olvido Mortal – in this work, the interactor is supposed to do “more” than is conventionally understood and handled, in a few cases. This also might suggest that IF in general may not have found the right conventions for what can be typed and understood and what can’t be. (Infocom certainly seemed to find a level that worked commercially and led to lots of fun.) Or perhaps the sort of inputs that can be handled will just naturally be ever-expanding, or otherwise different for different works.
Note that these problems aren’t specific to textual input in IF, and they don’t have anything to do with whether or not AI of any sort plays a role in IF. All software has defined capabilities.
Even if different conventions suit different works, we’ll still need to figure out how to indicate what these limits are, and how to expose the affordances of the interface so that issuing commands seems natural – even if the interactor isn’t sure what commands are necessary to unlock a particular work’s riddle.
Anyway, I actually think Michael’s original comment about how interactive computer works can be engaging (not just immersing, but compelling thought and interaction from the user) without particularly offering much agency is the really interesting point here … I just don’t have much to say about that at the moment, except that it draws a very important distinction that we should all be exploring in the future, and in the context of other works and forms.
[*] Every interface is constrained; there’s nothing special about the command line in this regard.
August 16th, 2003 at 8:53 am
“(A related question: Can one get any pleasure of agency from a book? Assuming it’s not a choose-your-own-adventure, a form of IF.)”
This is a question I’d thought about for a little bit. If this were possible, it seems like you’d need a book (written in 2nd-person) that was able to set up situations and manipulate the reader’s psychology so well, that there’d be practically perfect alignment between what the reader would intend to do in the given situations, and the actions the book says the reader takes. But would this really give the reader the pleasures of agency? Maybe, but the idea of a book like this is probably pure fantasy.
August 16th, 2003 at 12:03 pm
Irene Kacandes wrote about “forced performatives” which are sentences that do this. It seems impossible to sustain for much more than a sentence, but does occur in novels in short, short bursts. I wrote an essay a couple of years back that discusses this kind second person address, relating it to interactive works: Do you think you’re part of this? (PDF).
August 16th, 2003 at 10:07 pm
Jill, do forced performatives, by themselves, allow a reading experience to provide agency? Your essay doesn’t seem to claim that, instead pointing out Murray’s distinction between participation and agency.
Interestingly, it may be possible to answer Walter’s question “Can one get any pleasure of agency from a book?” affirmatively, even if a non-cybertextual book cannot provide any agency to the reader. How is this? It could provide the illusion of agency, or evoke the sense of agency, so that the reader can pretend to have influence over what text is presented next. Then, the reader would actually experience some of the “pleasure of agency,” even though there was no real agency provided. I don’t know that I’ve ever seen this done, however. I could allow myself to pretend that Calvino’s If on a winter’s night a traveller was addressing me, but I can’t remember any times when I allowed myself to think that I could influence what signs were actually printed in book that I was holding …
The distinction between participation and agency probably is the most productive way to think about what a fixed text can provide.
September 11th, 2004 at 5:25 pm
Rob Zubek has begun a related discussion on his blog, to which I added a comment, that also links to a recent related discussion on buzzcut.
October 25th, 2007 at 10:34 pm
Mario & Luigi Superstar Saga: Boss battles and cybernetic rituals.
When we fight a boss battle, we give up the freedom to choose and change the game’s world in favor of performing a specific set of actions under the game’s direction. What’s up with that?