Posted by Michael on May 17th, 2006, in Development
The goal of our work that Drama Princess needs to serve is to help give the spectator emotionally satisfying experiences. Such experiences often come from immersion in a virtual reality (a book, a movie, a game). To allow for such an emotional immersion, the spectator needs to willingly suspend his or her disbelief. As stated before, I think there is a lot that can be done to increase this willingness, rather than focussing all one’s attention on increasing the credibility. A comfortable interface and a pleasant environment will create a mood that will allow the willingness of the spectator to grow. The sympathy that the spectator feels for the characters in the game will do the rest.
Allowing this sympathy to happen becomes as important as making sure that the character is believable. And for artists like ourselves, the former seems a lot less daunting (and a lot more gratifying) a task than the latter.
Even hardcore A.I. scientists have realized that the flaws of an autonomous character can generate such sympathy in the spectator that these characters are perceived as much more believable than the ones that are intellectually superior or supposedly more realistic. The key is that people care about the characters.
Another important element in the emotional satisfaction that comes from fiction is empathy with the characters. If the spectator feels what the character feels, or at least is inclined to try to imagine so, the story will appear very believable and even relevant to the spectator’s own life. And when the spectator feels sympathy for the character, he or she will find it a lot easier to feel empathy. So again, sympathy is key.
A last piece of this puzzle has been suggested here before: it is easier to feel sympathy (and thus empathy) for someone who shows sympathy for you. Note that I said “show sympathy” and not “feel sympathy”. All we care about is how things appear to the spectator. So it becomes of utmost importance that our autonomous character displays behaviour that can be interpreted by the spectator as expressing sympathy for him or her. As we know from experience with 8, this can be as simple as recognizing that the player is there by making the character look into the camera from time to time. She’s so cute! 🙂
A few more questions remain. Since we are talking about a situation in which the spectator is represented in the game world by an avatar, we can ask ourselves whether we need to establish sympathy between the autonomous character and the player’s avatar, between the autonomous characters and the player (the camera) and/or between the avatar and the player.
1 Comment »
Posted by Michael on May 17th, 2006, in Development
Illustration from Varieties of Learning by Richard Evans in AI Game Programming Wisdom
Look at that!… Isn’t that beautiful?
This system almost moves me with its clarity and simplicty.
And yet…
Yet I feel, that we don’t need the top part of this diagram at all for our purposes. Does it really matter what our character believes or desires when all we see is him throwing a stone at a house? Does it?
Please somebody protect me from myself. And explain to me why our puppets need to have minds of their own. I can’t find any reason. Why go through all the trouble? Please, someone? Anyone?
7 Comments »
Posted by Michael on May 16th, 2006, in Projects
Chris Crawford’s Storytron is a technology for interactive storytelling. Swat is a tool for writing interactive stories. I’ve been looking at the documentation of this application (Alpha version 0.6). Here’s some ideas I found interesting.
Every event in the story contains all possible types of reactions to this event, called Roles, and an if statement for each of these Roles. So the actors don’t really think for themselves, they just look at the list of Roles and pick the one for which the if statements evaluates to true. This sounds very similar to our own “AI from the outside” ideas and Richard Evans’ “Activity-things”.
For each Role, there are several ways of reacting to an event, called Options. A character chooses an Option based on its Inclination. And this is where the (artificial) intelligence of the system resides. The inclination to do each option is scripted by the writer in mathematical formulas that can relate it to the personality of the character, its history, its relationships, etc. The character will choose the option (i.e. perform the action) with the highest Inclination value. One can imagine that there must be a random number to add some spice to this relatively stiff system.
Swat’s scripting interface (or perhaps even the very use of scripting) requires that a writer breaks down everything into minute elements. Perhaps an extremely clever person can keep the enormous mental picture in his head that encapsulates every element in the scene. But in general, I feel that a lot of the little emotional things that a writer might include spontaneously would get lost in this process. What about, e.g. ambiguity of words, associations, implications of tone? All of these are extremely important for the dramatic impact of a story but how do you express these in formulas that only describe events?
The single biggest problem of Storytron, however, is how it represents the story. It does so in a subset of English, represented in a symbolic way, called Deikto. Here’s a mockup of that representation:
The part on the left says “Knifer has just issued you a threat: Give your bicycle to him or he’ll knife you.” The part on the right is how you respond.
Now, even though Mr. Crawford states that the singular beauty of interactive storytelling is not in its representation – it is in the richness, depth, variety and drama of the interactions it allows, the representation still needs to be understandable, readable at least. Are readers really expected to learn this new language in order to experience these storyworlds? And how will Deikto translate to other languages?
4 Comments »
Posted by Michael on May 16th, 2006, in Books
This book is partially an attack on the traditional games industry (in which the writer was extremely active about a decade ago) and an introduction to the technology that Mr. Crawford has been developing since he turned his back on it. Chris Crawford considers contemporary games to make extremely primitive use of the interactive medium and he argues for an approach that will allow for a lot more variety in the stories that can be told with it. He considers interactivity to be the core of the computer’s innovative potential and the algorythm to be the essential means of expression. He measures interactivity by counting verbs, by counting the amount of things that a player can do. True enough, contemporary games contain very little verbs (run, jump, shoot).
Given the current success of computer games, Mr. Crawford does not believe that any significant storytelling medium can grow out of games. Interactive Storytelling will become an art form on its own that will one day completely overshadow games.
The biggest gift to humanity in this book is a definition of interactivity that puts an end to all confusion and fuzziness that surrounds this term:
A cyclic process between two or more active agents in which each agent alternately listen, thinks and speaks.
Beautiful. Thank you, Chris.
There’s no denying that this book has been very inspiring. Mostly because I disagree with almost all of the solutions that Mr. Crawford offers. 🙂
There’s somewhat of a rant here about how I dislike the concept of a drama manager and in general the obsession with plot found in most people who are interested in stories.
Apart from that, I have an intuitive aversion of the way in which Chris Crawford attempts to express story events in semi-mathematical formulas. It’s as if he’s trying to tell a story to a computer, rather than using a computer to tell a story to a human. It’s the type of thing of which I suspect that it can only relate the essence of an event but non of the little details that make the event meaningful and emotionally interesting. The kind of cutting up into little pieces that causes the soul of an event to slip through the cracks.
But this could be a huge prejudice from my side and maybe his “algebraic approach” does pay off in the end, when his technology will be used by artists clever enough the wield the brush he is creating.
He is well on his way to actually releasing software that will enable artists to create interactive fiction according to his vision. I will be looking into Storytron more closely soon.
3 Comments »
Posted by Michael on May 16th, 2006, in Development
Because of the topic of this research, I’m trying to observe people more purposefully than I normally do. To see what they do, how they behave, how they interact with each other, etc.
It occurs to me that most of the time, people are in between things. You find them moving from one place to another, or doing something with a certain goal in mind. Most of the time, this behaviour is tremendously boring to observe. These people, these real people, are quite different from the characters you meet in movies or novels. The latter are far more interesting. It’s much nicer to go sit in a crowd and read a book than to look at the passers-by!
So this confirms my belief that we should not try to make synthetic humans. Because humans are boring! The only way that you can compensate for the general tediousness of human life is, I guess, to compress time (as is done in The Sims). So that the 90% of their lifetimes that humans spend on being in between things at least doesn’t last too long.
The real solution -in my opinion- is to simply ignore this in-between time and design a system that only deals with the 10% of human life that is interesting to look at. The question that remains is whether this is compatible with the real-time medium. Will a character that does “interesting things” all of the time still seem real? Or do we need a structure that implies all the other things that the character does without showing them (by skipping forward in time e.g.)?
No Comments »
Posted by Michael on May 15th, 2006, in Development
The interaction between autonomous characters in a realtime fiction can be defined very loosely. Since we don’t have a fixed story to tell and all meaning should come from the spectator’s own imagination, a great deal of the autonomus behaviour can be governed by randomness.
One of the characters in our virtual space, however, will be the avatar of the spectator. Contrary to the Non Playing Characters (NPCs), the behaviour of this character will be controlled by a human. This means that we cannot use the concept of “activity-things” (Richard Evans) for these interactions. It seems that we will not be able to cheat as much when the spectator knows what is going on in the mind of one of the participants.
For instance, when one NPC seems to have forgotten what he experienced with another NPC, the spectator will probably make up a story why this is.
For example, in one scene we see the Wolf behaving very aggressively towards the Deafmute Girl en she runs away. In the next scene we see both of them play together. The spectator can imagine that at some point these characters got to know each other and decided that they like each other. When the Wolf is aggressive towards the avatar (Red), however, the spectator (who controls the avatar) will find it highly suspicious if suddenly the Wolf wants to play with her.
Perhaps this is just a problem of consistency. A problem that is fairly easy to solve.
But I can imagine that there may be other occasions when the relationships that the NPCs have with the avatar will require a bit more detail than the relationships that they have with each other. So it’s probably a good idea to keep an eye open for this and make sure that there is sufficient room in our design to add such detail.
1 Comment »
Posted by Michael on May 14th, 2006, in Development
What the tamagotchi shows us is that an artificial system does not have to be very complicated before its owners start to attribute autonomous, life-like qualities to it. It suggests that whatever other properties the intelligent robots of the future must have, they will be designed so that their owners want to take care of them.
Luc Steels, Digital Angels, 1999
Interesting how even “real” A.I. experts recognize the importance of “fake” A.I. After all, there is a very thin line between characters that are perceived to be autonomous and characters that truly are autonomous. If only in the philosophical sense that questions the very existence of that which is not perceived.
Equally interesting is the point made by Richard Evans as well (and waiting in the wings here to be elaborated upon): that it is very important to make the autonomous character lovable: if the spectator cares about the autonomous character, he or she is going to be much more inclined to perceive it as intelligent. So we were right when we designed our Little Deafmute Girl to blink cutely and look into the camera once in a while.
Here’s another inspiring quote from the conclusion of the same text:
[the robot of the future] will be far from the superintelligent, powerful, never failing robot, simply because an autonomous intelligence operating in the real world through sensors will always have to cope with limitations in its information resources. This weakness should be reflected in the robot’s design and behavior so that owners are inclined to take care of their robot and feel responsible for its behavior.
Luc Steels, Digital Angels, 1999
1 Comment »
Posted by Michael on May 13th, 2006, in Development
“It is our claim that the best way to include [activities] is for there to be actual ‘activity-things’, non-physical entities with their own state and internal logic to them, existing in the game world. Based on their state, they influence the behaviour of the agents within them, and based on what happens, they change their state.”
By Richard Evans and Thomas Barnet Lamb, Social Activities: Implementing Wittgenstein
While Richard Evans’ work will probably always allow his characters to have more individuality than we need them to have for our purposes, the idea expressed above, connects very nicely to our ideas of Usage instructions, AI from the outside and Modeling relationships. We have developed a preference for designing the things between the characters rather than what happens inside of them. The viewer can only perceive the outside anyway, the things that are expressed. Social activities are a prime example of how individual autonomous agents can be directed as a group. They don’t need individual minds to make decisions. Mr. Evans would not go this far because he wants his simulations to really work. While we only care about the illusion that takes place in the player’s mind.
The use of Finite State Machines to define these group activities, as suggested by Richard Evans, is fascinating and definitely something to keep in mind when the time comes to design Drama Princess.
2 Comments »
Posted by Michael on May 12th, 2006, in Books
This book reminds me a bit of those manuals with instructions on how to make love to women. To a man, it all sounds like such a hassle 😉 : she has to be in the right mood, and you have to have foreplay and you have to find some tiny nerve center, not to mention a secret zone hidden inside and then you have to make sure that you don’t come first. A hassle! No wonder men have sex with other men all the time instead! 😉
Seriously, I imagine that most male programmers who read Mrs Ray’s recommendations sigh and groan to the tought of having to take all the things into account that females apparently need in games (female -but no “hypersexualised”- avatars, emotional involvement, non-confrontational competition, short play times, etc). Much like they would if they were to read a sex manual.
Yet many of them have sex with women regularly. And I bet many of these women don’t dislike it most of the time. So surely, there must be different ways to stimulate a man into pleasuring a woman better.
Because that’s what games have in common with sex, right? Pleasure.
Anyway.
Most things that Mrs Ray mentions sound simply like recommendations for good design to me. And not terribly gender-specific. I know many men who enjoy the elements in games that are considered typically female in the book. Good design is well-balanced per definition. And that includes gender-balanced.
On the other hand, it is undeniable that current computer games are heavily biased towards excessively male play-features. I find this odd. I don’t think that I know any design discipline that is so gender-biased. Most chairs fit both male and female buttocks! And if there is a gender-bias, like in fashion, it usually doesn’t mean that there is no equivalent for the other gender (although one could argue that the male equivalent of fashion is often of much lower artistic quality than its female counterpart).
So Sheri Graner Ray does have a point. The extreme one-sidedness of game design is very problematic. I personally find design with an extreme gender bias simply bad design. And I (even though I am a man) would love to see many more of those supposedly feminine features in games. So I applaud the effort the book makes to try and convince game designers to include women in their target audience (if only for profit). I’m just not sure if it has the right tone. It does put the problem on the agenda. As an economic issue more so than an artistic one. So perhaps it is making a difference already.
1 Comment »
Posted by Michael on May 7th, 2006, in Development
I might be supersticious. 🙂
There is a certain form of beauty in simplicity, a certain form of elegance.
Artificial Intelligence is often complex. Even when applied to games. I think this comes from the fact that it tries to replicate human beings. And when analyzed, human beings turn out to be complex.
But when not analyzed, they are not! When you are not trying to replicate it, and you are just living, life is pretty simple. If humans would require A.I. to regulate our lives, we’d all be exhausted all the time! 😉
For technical reasons, simplicity is also desirable. The more complex a program is, the more unwanted effects it can generate (including bugs). Also (pertinent in the case of the author) the simpler a system’s design, the less programming skills are required to build it.
Programmers often seem to think that quantity is the solution to all problems. Just add more pixels, more polygons, more CPU cycles, more behaviours, etc and ultimately the problem will be solved. But I really think that in the case of A.I., you can never add enough. Your copy will always be inferior to the original.
The solution lies , I think, in finding a different problem to solve. The problem should not be to try and imitate life. It should be to create inspiring drama. So instead of a scientific problem, it becomes an artistic one.
The artist tells stories that are superior to life while the A.I. scientist always underperforms. The artist succeeds in doing this largely through stylisation: leave out uninteresting aspects and exaggerate the interesting ones. Simplify things to make them more clear. And -especially- appeal to the spectator’s uncanny ability (and desire) to fill in the gaps. When the artists omits certain things, the spectator can add them. He almost always does this in appropriate ways because both painter and spectator are humans and often they have a shared cultural background. And even if he fills in the gaps in ways that the artist wouldn’t have, it is still acceptable because most of the time, the spectator will fill in the gaps to make the art piece become more enjoyable for him.
There’s no need to paint the things that the spectator can imagine. In fact, and this is where stylisation really pays off, in my opinion, a lot of the joy of experiencing art lies in this activity of filling in the gaps. Perhaps because requiring the spectator to do some work to understand the scene immediately leads to empathy.
So, superstition or not, for the Drama Princess project, I will reject any system that seems too complex. If there is a solution, it needs to be simple.
1 Comment »
