Road to the IGF: Expressive Intelligence Studio's Prom Week

Games like Facade are known for the ways they've quietly tried to push game narrative technology beyond what we've previously known, and Prom Week sees itself as a spiritual successor in the field of sophisticated social simulators. With its own AI engine called Comme il Faut (French for, loosely, "as is proper"), UC Santa Cruz's Expressive Intelligence Studio has led Prom Week to a nomination in the IGF's Technical Excellence category. Gamasutra spoke to Josh McCoy, Mike Treanor and Ben Samuel, three out of four of the game's developers, about the state of social simulation, the process of developing the tech and much more. What background do you have making games? Treanor: Our team comes from all over! As an AI researcher, Josh has previously created crazy-sophisticated RTS bots. Aaron is an interactive fiction author who has made pieces about the recent wars and has the distinction of having authored the largest interactive fiction ever. Ben has made games about lawn mowers, alligators with jet packs, and true love. And I've previously made games about Ayn Rand, marriage legislation and observing animal behavior. What development tools did you use? Samuel: Pretty much every element of the game is the product of Actionscript 3 code (and the FLEX framework), which we primarily developed in the excellent FlashDevelop IDE. Much of our art was created using Illustrator, Flash and Inkscape. Also, every line of dialogue that you see in the game--as well as all of the rules that determine character behavior--was written using a home built design tool. Treanor: The design tool is a real piece of work... but it was invaluable in making Prom Week possible! McCoy: SVN, SCRUM and software engineering idioms helped a lot. How long has your team been working on the game? Samuel: Josh, Mike, and I have been working on the game since its conception in Fall 2009. Aaron joined a year later in Fall 2010. But, I think deep down, we've been building up to this moment all our lives. McCoy: The ideas for the AI system started forming in early 2009 by trying to apply lessons from sociology (with a focus on Goffman's work) to AI systems. Some results from this work was an abstract, experimental game based on the rules of stigma. As creating playable, authorable models of social interaction was my research focus, I had the extreme pleasure of forming the Prom Week development team. How did you come up with the concept for Prom Week, and what made you want to explore social simulation? Samuel: The actual premise of Prom Week evolved from many conversations between ourselves and our advisors, but the initial spark came from a desire to implement a novel AI system into a playable experience. It's our opinion that often times in academia, there is a tendency for people--brilliant people, mind you--to create theoretical AI systems that sound superb on paper, but then are never incorporated into an actual game. It is our humble belief that until you put your AI to the test in an honest to gosh playable experience, you don't really know good of a system it is. And why social simulation? I think a large part of the inspiration came from the fact that there aren't a whole lot of games out there that really let you explore social simulation. If you look at physics games, it's like... we've got that figured out already. We've got physics games out the wazoo. We've mastered that to the point where there are not only immensely popular games with real physics (I'm looking at you, Angry Birds), but that there are also immensely popular games that make use of stylized, almost artistic experiments with physics such as Super Mario Galaxy. There are all of these games that let you PLAY with physics, but short of The Sims, there doesn't seem to be many real opportunities for dynamic playing experiences in the realm of social simulation. On the other end of the spectrum, BioWare's been making some amazing games, but the Mass Effects and Dragon Ages of the world--as incredible though they may be--have social situations that essentially boil down to choosing between one of a handful of pre-scripted options, and the player simply selecting which of a few paths they wish to go down. Admittedly, any given path is guaranteed to be a well written, thrill packed adventure, but the player isn't really playing it, they're just living one of a few stories the developers pre-scripted. We really wanted to see if we could create an experience as dynamic and open ended as The Sims, while still retaining the feeling of a strong sense of narrative found in modern story games. McCoy: As anyone who has read a "choose your own adventure" book can agree, interaction at the story level makes the possible branches of the space of possible stories very large and branchy. The idea to explore social simulation was to take a chunk out of this problem -- if the system can take care of the changing social environment of the story world, what does that free the author's time to do? With the same amount of time, an author or designer could create more or better quality content. We really enjoy the story in games. However, at the same time we want more of it and for it to be more responsive to interaction. We wanted to make a system that could potentially enable the BioWares of the world to do more of what they do best! How did you develop a choice-driven and sensitive social system without the use of dialog trees? Samuel: Through a lot of sleepless nights! We started with the AI system, and modified it heavily as we discovered the needs of creating this brand new social system. As for the technical details, well, we've written some papers on the subject that any interested parties are more than welcome to peruse, but at a very high level, we've written 5,000 plus "social considerations" that influence behavior, where any given rule can range from the simple (e.g. "Confident people like to flirt") to the complex (e.g. "If we're friends and you are dating someone that I used to date but that person did something mean to me before we broke up, I'm going to be more likely to give you friendly advice"). These social considerations are all evaluated for truth to determine what social actions the characters want to make with each other, and then, once the player selects which action, we pick from one of dozens of "instantiations" for it based on the social state. In instantiations, we've decoupled specific lines of dialog from specific characters; in general any character can use any instantiation given that the social state is appropriate for that character to say it. So, yes, many sleepless nights. McCoy: Comme il Faut (CiF), the AI system that serves as the core for Prom Week's social reasoning, was aimed at solving two questions raised by Facade. The first was how do you reduce the amount of authoring needed to create a procedural story space. The second question was how to take common patterns of social interaction (like the affinity and therapy games played by Facade's Grace and Trip) and have characters use them as first class behaviors instead of having them implicitly reason about over thousands of lines of code. Basically, how do you decouple the social logic from the characters to avoid having to write a separate way to play the game for every combination of characters that could possibly play it. When you pair the model of social norms described by Ben with these reusable patterns of of social interaction, you get a power, procedural way for the social parts of the story world to be manipulated by the player and characters. Treanor: My hope is that Prom Week demonstrates that tackling the "problems" of game design (e.g. story vs player choice, etc.) isn't something to be afraid of. It seems that too often people will limit their creativity because of lack of imagination and technical ambition. I think the problem is that many designers don't have computer science chops, and many computer scientists don't know how to engage the world. We developed the game and the AI system in parallel and they pushed back and forth at each other like crazy! It's pretty much a given that you can't have rich, character specific dialogue and a dynamic simulation space. It's not like Prom Week is perfect or anything, but it demonstrates that to some extent that isn't true. And that is pretty cool. The ambition of your undertaking seems to suggest to me a major dissatisfaction with the current state of conversation and roleplay in games -- am I picking up on that correctly? Why do you think it's important to evolve and elevate these types of game systems? Samuel: Oh, well, "major dissatisfaction" seems like such a strong word. Personally speaking, I love modern story games. I know this might get some flak, but I was actually a big fan of Final Fantasy XIII, which didn't even HAVE anything as basic as dialog trees! That said, I think that you can only go so far with scripts and triggers; you can use them to create some amazing experiences, but they have their limits. In order to create new experiences, we're going to have to create these new game systems, and you are absolutely correct that the development of Prom Week and CiF was a direct attempt to try to accomplish this. McCoy: This came from love with existing games and frustration with their limitations. I'm hoping existing game story authors will enjoy a paradigm other than dialogue trees to author in ;) Treanor: To me, more than a solution to the not particularly dynamic dialogue in games, Prom Week is a system that is interesting in of itself. Of course, not everyone will see it this way, but what drove me was to do everything I could to make this incredibly complex system accessible to players so that they had entertaining access to how it operates and how it presents what social interaction is. What's next for Prom Week? Samuel: We're definitely hoping to release it on as many platforms as possible in an effort to give as many people as possible the chance to relive this most magical evening. Treanor: Prom Week's AI system, Comme Il Faut, is being used in several other projects including a tool for teaching conflict resolution, providing a safe space for people to experiment with social behaviors and learn the potential consequences, as well as for a cultural training simulator. I personally plan on using Prom Week's technology to make a rich simulation of human/cat relationships. Can you imagine? "Over 5,000 feline social considerations!" Played any of the other IGF finalists? Any you particularly loved? Treanor: Spelunky! Samuel: And, ah, it's nominated for so many things, perhaps it can be content to just win all those other awards, and let us take home Technical Excellence. Maybe? What do you think of the current state of the indie scene? Samuel: It's exciting! I think it's really amazing how much energy is being invested by individuals or small teams into the creation of some truly astounding playable experiences. That said, I feel that the 'barrier for entry' into game creation is still a little higher than I would like (i.e. more or less anyone can pick up a pen and paper and start writing, whereas getting a dev environment set up--let alone learning a programming language--can be a daunting task), but it seems more accessible now than it ever has been, and I'm holding out hope that it will only become more so in the future. The indie scene truly is incredibly inspiring, and we're all honored to be a part of it. Treanor: I love that so many people are making games! It is also awesome to see that indie games are influencing AAA titles. There is a whole lot more sincerity going on in the indie space, and the more of that finds it way into mainstream games the better.