« Small But Great Character Moments | Home | Picture Imperfect – Introduction »
Mind Your Verbs!
By Corvus | September 5, 2008
Behavior is a mirror in which every one displays his own image.
–Johann Wolfgang von Goethe
Mike “Rubes” Rubin of Monk’s Brew and I have been going back and forth about how to extract meaningful story-stuff out of player behavior, or rather the verbs a player chooses to express within the game. It started with me saying video game stories would be more meaningful if attention was paid to creating compelling and believable characters instead of simply writing out endless iterations of Campbell’s hero’s journey. This led to Mike reflecting that:
it’s a ridiculous amount of work to try and account for every possible action in every situation of the game, and the effect of those actions on all of the different characters in the game.
While I grant that this is true, I do feel it’s a task that can be made much easier if every available player verb is considered an expression of story from the very beginning of the design.
Think about how much work goes into the animation of character in our video games. If the game uses a third person camera and a sandbox style of play, the amount of animation involved is staggering. Now think about the amount of code used to determine which animation to run when the player interacts with an object. Think about the amount of code needed to govern the impact of those player verbs on the environment. The more objects the player can interact with, the exponentially greater the amount of code making it all work. It’s no wonder that it’s so easy to break the storyteller’s intent in these games. A thousand monkeys with a thousand gamepads will always figure out how to quickly subvert the intent of a hundred programmers.
It isn’t any easier in the textual world of interactive fiction, either. If anything, since the audience will be providing all of their own animation and the audience’s imagination is limitless, it’s even more difficult. Without a rigid and carefully planned approach to the use of player verbs and the objects available, the very thought of building game mechanics to communicate story is quite overwhelming.
Add to this plan the need to go into every dynamic interaction made possible by the available verbs and objects and attach some form of narrative significance and you’re looking at a near impossible task. If you really want to comprehensively implement a system like this, you’re going to start running down the rabbit hole of individual NPC response and a contextual framework to define situational meaning.
Yeah… pursuing a career as a soda jerk down at the local diner is looking pretty good about now, isn’t it?
So let’s scale back. Let’s see if we can approach this from a perspective that sounds a bit more reasonable, less pie-in-the-sky. Applying a system of meaning to a complete vision of a game is too daunting. It’s also not integrated enough to truly succeed. If we’re really going to make sense of it, we need to start smaller–at the very beginning.
In the beginning was the verb. Video games consist of actions the player can take. In other words–verbs. And what are verbs? Verbs are the expression of something I write about a lot. Verbs are the expression of what makes for a compelling storytelling experience. Verbs are the expression of what makes for a compelling play experience.
Verbs are the expression of relationships.
Think about that for a second. Even the seemingly passive verb, to think, that I used as a command in the previous sentence, can be viewed as the expression of an intellectual relationship between you and your life experience. To hate, or to love, might be viewed as the expression of a relationship between you and an external person, object, or action (I hate it when people hit me with bicycles), but it is actually the expression of an emotional relationship between you and your life experience.
Other verbs, such as the ones commonly implemented in games, are much more obvious. Walking describes your physical relationship to the surface that supports your weight. Jumping describes your physical relationship to the world’s gravitational field. To shoot describes, not your relationship to the target as you might first think, but the relationship of the weapon to the character holding it. The projectile issued from the weapon has it’s own verbs that define its relationship to the weapon, the world, and the target.
That might sound like I’m taking a very simple verb, to shoot, and making it overly complex. But, particularly in a game that takes place in a 3D environment, all of these individual parts of the whole process are already being described by discreet mathematics. What is missing from existing verb systems is reciprocity. Relationships are not one way. It is not enough to define the weapon’s relationship with you, but you must also define your relationship to the weapon. And therein lies the issue that I, and many others, are trying to solve.
The true test of character is not how much we know how to do, but how we behave when we don’t know what to do.
–John W. Holt, Jr.
My current design approach is to attach the verbs to the world and other objects, rather than the player character and NPCs. In other words, since jumping is something that only makes sense in the context of walking on a stable surface, the verb to jump is defined as part of that surface, which, depending on the rules of the world, likely inherits the verb from the world itself. In an RPG genre where characters may have a jumping skill, this is represented as a character-specific value that determines the success of participating in the world’s default to jump verb. The same would apply to a gun’s to shoot verb, a frying pan’s to heat verb, a ball’s to throw and to kick verbs, and a sword’s to slash, to thrust and to parry verbs. In this way, verbs are propagated throughout the games systems via the objects themselves, as are the animations needed to portray the world’s denizens implementing those verbs. [1]
This obviously is only part of a total package solution. The most important point here is that the verbs, once attached to the object, force you to consider them in a new light. Because the approach acknowledges the verb as a relationship, it therefore lends itself to more effective integration of game mechanics and story.
To do this successfully, you also have to consider context. Is it enough to declare that the narrative context for the verb to shoot is “violent?” What about the verb to throw? If I hurl a football at the back of your head, that’s one thing. If I hurl it into your awaiting arms, that’s quite another. I feel that context can be easily interpreted on the fly most of the time. A NPC playing a football match against you will clearly react differently to getting a ball kicked into the back of their head than a NPC minding their own business at the coffee shop. Much of this is part of a separate system of defining characters and establishing the all-important social contract between them, but this post has gone on long enough, so that’s a topic for another day…
[1] At this point in time, the demi-procedural animation system needed to implement this successfully in a 3D environment is highly theoretical, but also quite theoretically possible. [return]
Tagged:player verbs, storytelling, video games. | 14 Comments »
September 5th, 2008 at 5:31 pm
I like emphasizing the idea that verbs are relationships, but I’m trying to figure out what that actually buys us. If we adopt conventions like “the verb to jump is defined as part of that surface,” what actually changes?
And if “jump” is defined as part of a surface, how do we differentiate between objects on a surface that can jump and those that cannot(e.g. inanimate objects)?
This sounds like an interesting idea, I just don’t think I’m seeing the big picture.
September 5th, 2008 at 5:52 pm
You’re right that this approach doesn’t net us anything all on its own. Instead, it’s an approach meant to make it easier down the road to script NPCs to have more meaningful reactions to player initiated events.
As far as dynamic actors vs static actors–the verb itself is keyed to allow certain classes to call it. The foliage class, for example, will likely not be included in the list of acceptable classes. The humanoid class would be allowed to call it and would therefore have to be scripted accordingly.
September 5th, 2008 at 6:36 pm
What you’re describing sounds very similar to the AI technique known as Smart Objects. The Sims games are build around such a system. Objects in the world contain all information required to interact with them, ala a Fridge will contain the “Open Door” and “Take Item”, and maybe “Stand staring into Fridge for minutes” animations within its own code. Objects also broadcasts to the world that they are able to satisfy certain needs. So the Fridge will broadcast that it can satisfy the “hunger” need, and any agent (NPC) that is hungry will then head towards the Fridge and the appropriate interactions and animations will be triggered.
There is also a degree of cross-over with the Stimulus system described by Randy and Harvey Smith (No relation), in their GDC presentation on Emergent gameplay (Available here: http://www.roningamedeveloper.com/Materials.html).
In fact I’d recommend looking into Smart Objects similar AI techniques as I believe there’s a lot of potential research that has been done in those fields that would be of great benefit to a participatory storytelling system.
September 5th, 2008 at 6:40 pm
This is pretty much how The Sims AI works, isn’t it? I’m pretty sure there are some articles floating around that talk about the details.
September 5th, 2008 at 6:48 pm
I wonder if I was aware on some level of how the Sims AI worked when I designed this system. Regardless, yes it is very similar (I even have regions and objects that fulfill needs in other portions of my design). The primary difference (as far as I know) comes in with the idea of social frameworks and inter-character emotive responses.
September 5th, 2008 at 7:10 pm
On a related note, Corvus have you read any of Chris Crawford’s material on Interactive Narrative?
September 5th, 2008 at 7:19 pm
I have, and I find us to be diametrically opposed on a number of levels. It’s always an interesting read though.
September 5th, 2008 at 7:36 pm
I read his book on Interactive Narrative and my general intellectual reaction was “Hmm…”, whereas my gut reaction was that he was wrong in some major ways but I’ve yet to reconcile the two and work out where exactly I disagree with him most.
September 5th, 2008 at 9:50 pm
When talk of how helpful gaming verbs and how they can create explicit relationships between objects in a game comes up (as it so often does), I must point to Ninja 5-0 for the GBA (http://www.youtube.com/watch?v=GxLAYkstJzk).
I’m not sure I can recall an action game that gives you a better sense of clarity about how things in the world relate and what the effects of your control input will be. Once you learn enemy behavior and how your character interacts with the landscape, you can generate an incredible flowing momentum. I think you can get a sense of that from the video above.
Comparing my experience and success with this game to something like my experience (emphasis “my”) with the recent Ninja Gaiden games has been pretty illustrative for me about how important it is to create a space to really employ those verbs.
I’m certain the Ninja Gaiden games employed these explicit relationships well, as I’ve seen similar videos of people playing that games just as expertly, but I never felt those relationships were something I had the “space” to grasp and use.
September 6th, 2008 at 7:03 am
Top post Corvus. I like the simplicity (although don’t have any idea how it would work out in practice, in the writing of the engine) of attaching context to the world, rather than the player.
On a different note, I was reminded of Mozilla’s Ubiquity when reading this. Interaction through verbs.
Kudos!
September 6th, 2008 at 10:20 am
Although single words have many meanings, the situation the avatar is in tells the player roughly what will happen should the player choose the option. Imagine if all verbs were to be clear cut in a sentence;
‘Pull the trigger while aiming the gun at the person’ replacing ‘Shoot’.
The player will tire easily due to all that unnecessary reading.
*scratches head*
Or am I completely missing the point here?
September 6th, 2008 at 10:29 am
@David Lee: You might be. The verbs I’m referring to are actually game mechanics and do not necessarily dictate player experience. In other words, if this were to be implemented in a textual IF system, the player might shortcut it all by typing
SHOOT TERRY WITH SHOTGUN.But the system, when deciding what happens as a result of this player action, treats the process as a collection of micro-verbs, if you will.
September 6th, 2008 at 9:15 pm
Wow, so much attention this past week. Very cool.
It’s interesting that this blog also made me think of Chris Crawford’s writings, specifically <i>Interactive Storytelling</i>, and I think there are some parallels here. He devotes an entire chapter of his book to "verb thinking" instead of "noun thinking", and his first rule of software design: ask "what does the user do?" I think it’s a good principle in general, and almost identical to your statement that "video games consist of actions the player can take," although I agree that the similarties mostly end there.
I think I get the jist of this, for the most part, but I’ll admit it’s still a little fuzzy for me.
Attaching verbs to the world and objects, rather than the player, does seem to make sense, at least in certain situations. I could imagine something like the <i>throw</i> verb, where throwing one object may be interpreted differently based on the object being thrown. For instance, if the player wanted to throw a football, it would entail a different type of animation than if he wanted to throw a chair, and that could be dictated or determined by the object. That’s a fairly superficial way of looking at it, I think, but I’m finding it difficult to delve much deeper than that.
One thing you said in the comments I found interesting:
<i>"As far as dynamic actors vs static actors-the verb itself is keyed to allow certain classes to call it."</i>
This, in a way, gets back a bit to what I was trying to communicate in the quotation of mine above. You’re talking about only allowing certain classes of actor (or object) to call a particular verb. This sounds a bit like what I described in my blog as limiting certain actions. Let me give what I hope is a decent example.
In <i>Vespers</i>, one of the puzzles (without hopefully giving away too much) involves moving around a certain piece of furniture. There are a few hints in the game that lead the player in this direction, but the reality is that only this one object can be moved around in this way. There are plenty of other pieces of furniture in the game, but any attempt to move those is rebuffed. In fact, if the player did happen to <i>try</i> (unsuccessfully) to move furniture around prior to encountering this puzzle, I imagine the puzzle’s solution would then be quite counter-intuitive.
Now, one could certainly argue that this is less than optimal puzzle design, and in a text game, it probably wouldn’t be too difficult to implement a way of "moving" any piece of furniture in a way that doesn’t mess with the overall game. But in a 3D world, that becomes a bit more difficult. Do we allow the player to move any old piece of furniture? How much? In what direction? What happens when the player starts totally rearranging rooms? Sure, it would make for an interesting simulation for those players interested, and most people would probably abandon the exercise once they realize that it accomplishes little. But that means implementing the ability to move every piece of furniture that exists in the game, and that amounts to a lot of work, just for one relatively meaningless action.
So I’m not sure if this is completely related, and I’m not sure if I’m really getting all of this, but it’s a great discussion and I’m glad you’re engaging in it.
September 8th, 2008 at 6:10 am
You’re only really limiting things in that trees and washing machines aren’t allowed to throw chairs… unless you decide to create a class of tree that is allowed to access the throw verb. Chairs, trees, and washing machines, however, would be throwable, if the the actor accessing the verb met the requirements for the throw (that’s where we get into the RPG elements of having skill requirements that effect how successful the actor is at the verb).
Your chair situation is Vespers is exactly the sort of thing that drives me crazy. Why can I move this one chair when all the other chairs in the world are immobile? It breaks the reality of the game and reminds me I’m just facing a series of puzzles with fancy dressing.
But if all the chair were the same and players were allowed to put them wherever they wanted, and this one chair just happened to be covering up a clue… that not really more work if the chair’s behavior were all built into their very existence as chairs. You wouldn’t have to specifically implement the move verb for each chair, only assign the model to the class and the move verb would be automatically implemented.