Virtual Drama as a Learning Medium:

The Caracol Time Travel Archaeology Project

Michael Moshell, Charles E. Hughes, Dean Reed

School of Computer Science

Diane Chase, Arlen Chase

Department of Sociology and Anthropology

University of Central Florida

5 July 1999


A group of 18 undergraduate students from a variety of disciplines spent two semesters designing and constructing an interactive drama-based educational role-playing game using VRML and VRMInet, a Java-based system for shared virtual worlds. The game was designed to teach some of the basic concepts of archaeology and some facts about the life of the ancient Maya, to a target audience of middle school students.

The game is based on an actual Central American archaeology project being carried out in Belize, under the supervision of Diane and Arlen Chase (Chase 99). Five of the students and one faculty member (Moshell) visited the project in March of 1999 to gather data.

This paper reports lessons learned about the design of virtual environments and role-playing for education, with dramatic themes based on real-world learning environments such as the Caracol Archaeology Project.


Figure 1: Caana, the great pyramid of Caracol, in ancient times

1. Background

It is generally agreed that virtual environments (VE) consist of multisensory interactive simulations based on a three-dimensional geometric metaphor, within which the user has free control of the viewpoint. There is less unanimity on whether a virtual environment must be immersive (i. e. monopolize one's senses via a head mounted or CAVE-style display system.(Cruz-Neira 92)), whether the world must be occupied by avatars (visible representations of the participants) and the degree to which text-based simulations (MUDS, MOOS, etc.) and graphical virtual environments are relevant to one another.

Virtual Drama. Several experiments in virtual drama have been reported in the literature. Often these projects involve interactions between human "guests" and autonomous characters (Hayes-Roth 95, Maes 95). In the Oz project (Kelso 93), in addition to interaction with autonomous characters, experiments were also conducted in which live actors instead of computer controlled characters were used. The live actors followed a script, whereas the guests ("users" of the system) were given only a general sense of the objectives and background of their characters. These experiments were intended to develop ideas about how plot structuring devices such as plot graphs could be used to structure the experience of the guests. Ultimately the goal was to automate part of the role-playing.

In contrast, the project described here focuses on human-human interaction in virtual worlds. As in the Oz example, we maintain the distinction between guests and cast members. But we are not working toward automating the cast members' functions. We are instead focused on developing the virtual medium as a friendly and convenient place for experimentation with drama as an educational tool. We are interested in developing the necessary and sufficient affordances for VE to enable learners to take advantage of virtual sets, of mentors-at-a-distance, and of the theatrical magic that computer graphics can make possible.

VE and Education. A number of experiments have been conducted which investigated the use of immersive and non-immersive graphical VE systems for. Some of the most important projects are surveyed in (Moshell 99); a more comprehensive survey can be found in (Youngblut 98).

Only a few of those experiments investigated the educational impact of social interaction within the virtual environment, and these were usually based on two explorers' interaction while pursuing a mutual task (e. g. Johnson 99). Most work on social interaction in cyberspace has taken place in the MUD/MOO universe, where the technology is not as large an obstacle to development and experimentation as it is for 3D interactive systems (Jesup 99)

Two of the authors (Moshell and Hughes) have been developing concepts for the educational use of 2D and 3D virtual environments involving four to eight users in complex stories. The ExploreNet system, originally modeled after the well-known Habitat role-playing environment (Morningstar 91), has been used for a series of experiments with elementary and middle-school students. (Hughes 97,98; Moshell 96.) ExploreNet uses simple cartoon scenes and links multiple computers together. Normally, each computer controls one character ("avatar"). Avatars "speak" when their controlling player types messages. The messages appear above the avatar, and any other player's avatar is in the same scene, that player can read the messages.

For the Caracol Time Travel Project, we decided to move beyond ExploreNet's two dimensional system and to explore the use of three dimensional networked VR for learning. However, we continued to work within the drama-based educational framework we had originally developed in conjunction with ExploreNet. We now explain that framework.

2. The Virtual Academy Model

In the course of these experiments, an educational model called The Virtual Academy was developed. Central to the Virtual Academy model is the use of role categories which include guests, cast members, world builders and tool builders. We are experimenting with the use of multi-age groups in each of these roles, with a range from elementary students through college students to retirees.

Guests are naïve first-time visitors to the virtual world. They have, at most, read some background information and perhaps seen a pre-show. Guests are analogous to visitors to a theme park. Guests are typically students in a school which has two to four Internet-capable computers in a classroom; they are taking a "virtual field trip" which is correlated to subject matter they are currently studying in the classroom.

Cast Members are persons who have learned enough about the simulated world to play specific roles, in costume (as avatars). They are encountered in the world by guests. They need three kinds of skills and knowledge:

World builders are persons who define the world's learning objectives, develop scripts and theatrical scenarios, construct the geometric models and program the behaviors of the objects in the world. World builders also usually serve as the directors, in a theatrical sense, of the actual productions.

Tool builders are technically adept persons who modify the software with which world builders work to construct the world. Because of the immaturity of VE technology, tool builders play an essential role in the production of any piece of interactive drama.

The ultimate mission of the project was to develop an educational virtual environment suitable for use by middle school students. The subject matter and overall presentation were designed for this age group. However, during the portion of the project described in this paper, all the above-named roles were played by undergraduate students at the University of Central Florida. The 18 participants consisted of ten computer science majors, three liberal studies majors, two music majors, two graphic design majors and a psychology major. Several anthropology studens served as consultants. It would have been desirable to include theater majors as well, but none were available at the time.

3. Objectives of the project

The project had as its three main objectives, to improve our understanding in the following areas:

We briefly present our approach to each topic, before describing the actual experience of the class.

Interactive virtual drama for learning.

Concept. The presentation paradigm the Caracol Time Travel project was as follows:

1. A host school (or some other organization) would provide well-trained cast members, and offer to stage performances for other schools with appropriate Internet-accessible computing equipment.

2. A school wishing to participate in Caracol Time Travel (either as host, or guest institution) would need a direct Internet connection (better than a 56k modem), three to five high-performance Pentium computers, and a projection device capable of displaying an image for viewing by the whole class. Each computer would need to be equipped with speakers, a microphone and an appropriate sound card.

3. The guest class would read background material about the Maya civilization before the performance. Two to four members of the class would be selected as guest actors; they would serve as the class' surrogates in the experience. Each guest would sit at a separate computer; the extra computer would be connected to the projector.

4. The guests would experience three episodes, aggregating about an hour of interaction. Their roles would be as archaeology student trainees, working on the Caracol Archaeology Project. They would experience dream-like time travel episodes that provide access to the life stories of people living in the classic Maya period, and participate in a simulated archaeological excavation which discovered artifacts relating to the story.

5. The host school would have undergone substantially more preparation (including a previous experience as a guest school in this or some other virtual dramatic experience.)

Inherent to the Virtual Academy model is a two-pronged question:

1. How can you structure a situation in such a way that students will naturally adopt roles and interact with each other in the general direction the author intended?

2. If this behavior occurs, will the participants learn anything useful from it?

A good deal of research on situated cognition (e.g. CTG 93) has indicated that the cast members - those with primary, predetermined roles to play - will benefit from such an experience. In fact, it has always been a central hypothesis of the Virtual Academy model that the principal learning would occur on the part of the cast members. We need to consider the role of the guests. What is their intended role, and what is the benefit of having them in the scenario?

Guests serve two purposes. They provide a visible audience to whom the cast members are playing. In this capacity and with their ability to move freely about and ask questions of all sorts, they present the cast members with many challenges. They also provide the invisible audience with a means of interacting with the story. We presume that the passive audience will be giving a great deal of advice to the guests, as the story unfolds.

Given the maturity and energy level of middle school students, it is possible that the classroom audience's viewing of the experience while it is in progress, may be too disruptive for the game to work. In this case, an alternative arrangement would be to have the guests interact with the scenario from a laboratory or library workroom, without the class concurrently observing. It will be possible to save and later replay the story for "after action review" by the class.

Key Questions. The current experiment was carried out with college students, a more tractible and focused age level than middle schoolers. This allowed us to focus on a subset of the key questions, which were:

Questions which were deferred until an opportunity arises to work with middle school children included: World-Building

Concept. The basic hypothesis is that students will learn a great deal about the subject matter, if they are required to construct a script and a virtual world which conveys this subject matter to others. This project provided an extended opportunity to explore this idea.

World building must proceed in intimate relationship with script development. It would be desirable to have a mature and stable set of tools for world building, and a team of designer/constructors who were completely familiar with the tools. It would also be desirable for the students to have substantial prior knowledge of the subject matter involved. For instance, for Caracol Time Travel, the design team should have included anthropology majors who had taken archaeology courses to learn both archaeological theory and techniques, and specific information about the Classic Maya.

None of these resources were available, and so the script and the world co-evolved very much under the influence of the world-builders' current mastery of the tools, and within the limits of the abilities of non-specialized students to gather the necessary information. Nevertheless, the project produced useful virtual worlds, and provided insight into the following questions.

Key Questions:

Technical Infrastructure

To build a three dimensional graphical virtual environment suitable for sharing a PC, one begins with tools. Inspired by a 1992 science fiction novel titled Snow Crash (Stephenson 92), a collaborating team developed VRML - the Virtual Reality Modeling Language (VRML 99). VRML has achieved a certain maturity of functionality, but has never developed much economic impact. As such, its performance and the quality of tools available for use with it, leave much to be desired.

Nevertheless, VRML was the most promising tool available to this project in the fall of 1999. A system called VRMINet was developed and tested in the course of this project. Appendix D describes VRMINet in some detail, and points to more complete Wed-based documentation. Users controlled avatars (moving human figures) using a mouse. Generally, each user had a separate computer. The avatars were constructed by several means, including adaptation of other avatars found on the Internet. Only limited motion (a walking cycle) was provided for the avatars, as their primary role was to indicate the location of the action in the story. Their construction is described in more detail below.

Key Questions

4. Building Caracol Time Travel

To tell an interesting story, you have to begin with a story. A core team of eight students with backgrounds predominantly in the humanities worked through the fall of 1998 to develop the central story line. They were then joined in the spring semester of 1999 by ten more technically proficient students to help build and test the virtual world. Storyboards were developed as the script grew. However, as the team had substantial writing talent and very few students with graphical design skills, the storyboards did not play as large a role in the story's development as did the written scripts.

The concept of interactive drama is an intriguing notion but one with which most people have little experience. To introduce the concept, an introductory exercise was held in which a simplified two dimensional "electronic puppetry" system (in fact, ExploreNet) was used. A script was developed by a subcommittee. The actors were encouraged to develop a sense of the overall story and the characters' personalities, and to progressively abandon the literal script as rehearsals proceeded. The most important difference in this warm-up scenario, and the main project, was the absence of guests in the warm-up scenario. That is, everyone in this performance had a scripted role, whereas in the main game the guests are free to react to the cast members' actions as they choose.

Building the Story. We began by reading a number of books about the ancient Maya, and about archaeology. A principal source was Linda Schele's A Forest of Kings. (Schele 90). We visited the Chase's archaeology laboratory, handled artifacts and human remains, and interacted with several anthropology students.

In reading books about how modern Maya life has been studied, we learned about the extensive oral literature of the Maya and how fragments of late Classic and Post-Classic Maya literature were captured after the Conquest, in the form of manuscripts written in Maya languages using the Latin alphabet. The great work from the south of the Maya area is the Popul Vuh, which tells the complex story of two brothers who went into the underworld to battle the gods. In the northern Yucatan (a more relevant region to Caracol) a number of similar but distinct Books of Chilam Balam were preserved. Most pre-Conquest Maya heiroglyphic documents were destroyed by the Spanish; only three undoubtedly genuine books, together with numerous inscriptions on stone monuments and within temples, remain.

An essential element of modern (and presumably ancient) Maya storytelling is that the audience actively participates. One anthropologist, having learned fluent Yucatec Maya, asked his hosts to tell him stories. They reported that they would like to, but he had not yet learned how to listen to them. He gradually came to understand that a complex set of customs governed how the listeners would ask questions, take the part of some of the characters in the stories, and otherwise reinforce the performance of the storyteller. The student designers of Caracol Time Travel resolved to incorporate this technique into the game.

Perhaps the most difficult component of ancient Maya life to understand, is the peoples' relationship to the gods. Modern secular human beings generally believe that their actions are utilitarian and relate to concrete, objective cause-and-effect phenomena. You turn on a light switch when you want light. Few of us have reflected on the abstract nature of, for instance, working forty hours a week for a piece of paper (or an electronic funds transfer) which goes from a financial office one never sees, into a bank which one also never sees, and thus magically empowers a small blue plastic card. This talisman can then be taken to a grocery store or shopping mall, waved ritualistically about, and one's material needs are met.

To the ancient Maya, the work that had to be done to build up the "bank account", consisted largely of sacrifice. Nobles sacrificed their own blood or that of captives, for the good of the community. Heads of household sacrificed to gain their own families' well-being. The Maya needed rain as badly as we need money, for their lives depended on corn. Corn cannot grow in the Maya country without rainfall, for there is little surface water with which to irrigate. Sacrifice paid the bills; smoke from burning paper soaked in blood rose up to the gods, and the rains came.

It was clear, then, that our story needed to center around sacrifice and the Maya's relationships with their gods. To convey this sense of the importance of magic, it was decided to make the story itself magical rather than realistic. All of theater is based on the suspension of disbelief, and even young children have no trouble distinguishing Mickey Mouse's world from the real world.

However, the blood sacrifices of the Maya raise problems if they are to be literally represented in a virtual world. Male sexual organs were a common source of blood; women pierced their tongues. To avoid extraneous modern political issues such as body piercing, we substituted the sacrifice of a precious jade amulet in the story line.

Stories within stories. With the introductory exercise, we began to explore the notion of a story within a story - a "flashback" to a more ancient past. In the exercise, the present was classic Maya, and the past was the mythical time when the gods were setting up the world. In the main game's story, the flashbacks occur by "time travel" between the present world of archaeology, and the living world of the classic Maya.

The story within a story allows people in the present time to exercise the roles of storyteller and active listener, as is done in Yucatec Maya storytelling. This technique is well illustrated in the script for the introductory exercise, in Appendix A.

For the introductory exercise, we used the two-dimensional ExploreNet system (Hughes 95) to develop and role-play a story about the rain-god Chaac. We took a legend from one of our source books, embellished it, and designed two-dimensional graphical characters for the essential roles. Here's the story, in compressed form.

Chaac the rain god has a house on the earth, but usually he spends his time up in the sky. One day, Chaac was cold. He told his two assistants the frogs:

"I'm cold. Go down to my house, and ask my wife for a blanket!"

The frogs didn't speak Maya very well, but they knew they had to try. They went to the house and asked Xumacane (Chaac's wife) for a warming-thing. She took up a burning branch from the fire and gave it to the frogs. They took it to Chaac.

"THIS is not a blanket!" Chaac shouted angrily, and threw down the burning branch with a crashing boom. And that was the first lightening and thunder. "Go down to my house and get me a BLANKET!"

The frogs hurried down and asked Xumacane for some other warming thing. She had some water boiling in a pot, so she poured it into a gourd and gave it to the frogs. They quickly took it to Chaac.

"And THIS is not a blanket either!" Chaac roared, and he poured out the gourd of water. And this was the first rain. "This is your last chance! Go down to my house and get me a blanket to wrap myself with!"

The frogs ran to Xumacane and told her what Chaac had said. She looked around for something to wrap around Chaac, and spied some cotton, growing on a bush. She quickly plucked large hands-full of cotton and gave it to the frogs. They took it to Chaac.

"Ahh... this is soft and warm, so I shall wrap myself in the cotton." And this made the first clouds. And the frogs sang - ap ap ap ap - and ever since then, the frogs sing to announce Chaac's coming with his firebrand-lightening, and gourd-rain, and cotton clouds.

Initial Role-Playing Experiments

We experimented several times with how to role-play this simple story, using ExploreNet. Each of the four characters was initially placed under the control of individual actors, who were given a separate computer. The behaviors of the props (firebrands, gourd, cotton clouds) were set up as actions that could be triggered by a menu selection, when an actor mouse-clicked on the body of the character he was controlling. One student collected sound effects on the Internet and set them up on a separate computer, so that at appropriate times, thunder and the sound of rain could be hand-triggered.

Initially, the ExploreNet communications paradigm (typing one's utterances and sound effects) was used. Two venues (scenes) were developed - the sky scene and the house scene. As the frogs moved back and forth between the scenes, several issues became obvious:

Many of these problems disappear (at least for simple stories like Chaac's rain legend) if you set up the computers and perform the enactment in a space where all the actors can speak out loud, and where the audience can watch any screen they wish to observe. This viewing technique only works for relatively small audiences, however.

Players' reactions. Initially, there was some shyness in acting out a manifestly low-production-value sort of a puppet show. Fairly soon, however, the students began to get into the experience. They developed voice characterizations for the characters they were playing. There were no guest roles in this scenario, and thus effort was concentrated on learning how to move the characters into place as smoothly as possible using mouse gestures, and on the timing of sound and action. In essence, this warmup activity was similar to traditional theater in which guests' roles are passive, and actors do all the work.

After several experimental performances, the script's original author (Robert O'Leary) summarized the evolved story into a script that captured the spirit of dialog-style storytelling. In this version of the story, Ana (a Maya girl) and her two younger brothers Juan and Jose are talking as the story is acted out. O'Leary's script version of the story is attached to this report as Appendix A.

Preparing the Design Teams.

During the fall semester as the story developed, we realized that we wanted to have three major episodes. Three skeletal teams were assembled, each with a leader who had experience and talent in writing. Team 1 was led by Robert O'Leary, a technical writer. Team 2 was led by Mohammed Usman, a sociology major. Team 3 was led by Angela Leavell, a Liberal Studies/Women's Studies major. This leadership structure persisted through the Spring 99 semester as well. The remaining students were distributed across the teams.

In January when the large contingent of computer science students were added, we inventoried their skills at VRML and Java, and made sure that each of the three teams received at least one student who had completed CAP4021 - Building Virtual Worlds, a course built around VRML. Artistic skills were also carefully allocated to each group. However, we did not assign any specific roles or titles (such as Technical Director) within teams. These roles emerged as the teams began to function.

Initially it was our intention to select an overall student leader for the project, but no natural candidate emerged. The leaders of the teams were fully engaged in bringing their episode forward, and so the leadership council consisting of the three team chairs played the essential coordinating role.

Designing the Main Story.

We knew that we wanted to personalize the experience of archaeology along two dimensions: we wanted identifiable characters with plausible motivations, doing things real people do in similar situations. And we wanted to explore both the modern world in which archaeology is taking place, and the ancient world in which the Classic Maya lived.

During September, we assigned short-story writing projects, to develop concepts for individual characters and themes. Gradually, a structure emerged. We knew that the principal historical event which is firmly associated with Caracol, is the conquest of Tikal (a large neighboring city) by Lord Water, a leader of Caracol. We also knew that on top of Caana, the largest pyramid at Caracol, a high-status female burial had been found. Very few female Maya burials receive the ceremonial treatment normally afforded to males. The female's name and precise dates of birth and death are not yet known. We resolved to incorporate a fictitious version of this lady, named Rainbow, into our story.

The overall story line that was adopted, was based on time travel. Guests would play the role of student archaeology trainees, coming to Caracol to work on the project. They would interact with characters representing the children of indiginous people such as the camp's cook. The children would reveal magical pathways that led to the mystic Maya past.

Based on the lessons learned while working with Chaac and his wife, it was decided to use a "Wizard of Oz" principle for characters. That is, each cast member character in the present-time scenario would be matched with a character in the Classic Maya time. They would resemble one another in physical appearance and personality, and would be played by the same actor. The guests, since they travel through time, would not need two characters.

After much adaptation, the story's roles were established. Two guests represent two student trainees named Pat and Jay. In addition, we have the following characters:

Cast members-Modern time, and corresponding Cast Members- Ancient Time

Dr. Dana Williams, project director Rainbow, a noble woman

Dr. Idaho Smith, chief archaeologist Chief Priest

Louise Hardy, student team leader Missaira, a witch

Bud Castle, archaeology student Maya Soldier

Ana, a twelve year old Maya girl Lida, a twelve year old Maya girl

Pedro, Anna's little brother Mouse Jumper, Lida's little brother

<no modern equivalent> Flame Monkey

Not all these characters made it into the final story enactment, due to resource constraints. (There are some inconsistencies in character naming between here and the January scripts shown in Appendix B. Don't worry about it.)

Again inspired by the Chaac story, it was decided to provide for a camera operator and a dedicated computer to render the public view of the performace. This operator would maneuver a viewpoint so as to provide for passive viewers, a consistent view of the story's action. The camera was represented in virtual space by a small camera-like icon on a tripod, that was visible to the cast members but not to the guests.

A small team prepared an introductory movie to set the stage for the guests. This movie was originally intended to be animated, with the incorporation of a mixture of live images of leaves and foreground trees, with animated images of remote temples. However as it worked out, this proved too difficult. A collection of video footage was shot on-site from the hood of a moving automobile touring the camp, and this was edited to serve as a short introductory segment.

There were plans to incorporate sounds into the project as well, but these largely fell by the wayside as the story developed. Some sounds were recorded on-site in Belize and others were collected from around the Internet, but no effective means was developed for cueing these sounds into the live performance. They had little impact on the overall effectiveness of the dramatic experience, which was driven essentially by the dynamics of the actors' voices and the moving viewpoint in virtual space.

Story Synopsis.

The Caracol Time Travel story incorporates elements of several Maya myths, together with a substantial amount of new dramatic material. The "back story" is presented here, followed by a brief synopsis of the script for the interactive performance, as actually rehearsed and performed. The student-written scripts for these three episodes as it existed prior to implementation began, together with the Chaac episode, are provided in Appendix B. That script is discussed below.

Back Story. In ancient Caracol, an old homeless woman (Missaira) scrabbles around midden heaps, scavenging food. Almost everyone mistrusts her and calls her a chartzcoal (witch), though she is befriended by a young noble girl named Rainbow. She gives Rainbow a sacred jade amulet, to use in a future time of great need. The village boys taunt the old woman. In anger, she magically transforms the annoying boys into chattering monkeys and drives them into the forest.

Rainbow is anguished that the rains have not come. Her people need food and strength for their struggle against Tikal. She prays to the rain-god Chaac, and offers to sacrifice her jade amulet if Chaac will bring rain, even though without the amulet, her soul can never enter the underworld of Xibalba when she dies. She smashes the amulet and places the fragments in a small red ceramic pot. The rains come.

Rainbow grows old as the revered queen of Caracol. When she dies, her body is placed in a tomb high atop Caana.

Figure 2. Rainbow's Funeral. Her spirit appears in the burning brazier to the left.

Virtual Drama: Episode 1.

In modern times, the archaeology project discovers Rainbow's burial, but has no access to her life story. When our drama begins, student trainees Pat and Jay are introduced to the site by the chief archaeologist. After he leaves, the guests interact with the Maya children, who reveal to them that sometimes it is possible to see the past. The children show the guests how to touch a certain symbol on a certain stela (stone monument), and the students find themselves transported back to ancient times. Ana's toucan accompanies them, and leads them to observe Rainbow's grave, as she is being reverently buried.

Figure 3. Ana, Dr. Smith, Pedro and Toucan at Stela before Modern Caana

Figure 4. Arriving at the plaza on top of Caana. Other structures below.

The guests see a young girl, Lida (who resembles Ana.) as she observes a fire burning alongside the burial site. In the flames appears the spirit of Rainbow, who says: Lida, you must help me to recover my amulet. Without it, I cannot enter the underworld.

Time travel ends. The guests find themselves back in modern times. They discuss their experience. Just at the end of the episode, a member of the party notices a small red ceramic shard sticking out of the soil.

Virtual Drama: Episode 2

The student trainees (guests) and advanced students (cast members) Louise and Bud, together with Ana and Pedro, go to a cave that contains Maya artifacts. They look at various artifacts and discuss them.

Figure 5. Examining a map in the cave.

Eventually they come to an ornate box which contains a bone. When someone touches the bone, a ghost-monkey rises in a flickering flame from the box and tells the story of how Rainbow was given the amulet and how she sacrificed it. The monkey explains that their transformation from boys into monkeys was done so as to leave someone behind who could tell the story.

Figure 6. Talking to the Flame Monkey

Virtual Drama: Episode 3

In this episode, the students participate in a simulated excavation. They use a shovel, a trowel, and a sifter to locate four broken parts of the jade amulet. When the parts are brought close together, they start to glow and are reunified. The spirit of Rainbow appears and thanks the young archaeologists for their assistance. The monkeys come in from the forest and are transformed back into Maya children.

Figure 7. Lida at the altar in ancient Caracol (Night)

Evolution of the Script. A close reading of the scripts in Appendix B will reveal that the performance summary above differs in some respects from the intended scripts. As the design teams struggled to make VRML worlds that would support their performances, a succession of simplifications occurred. There was some conflict between the teams developing episodes 2 and 3, over the subject of how the story of the amulet would be revealed. Ultimately the story was disclosed by the flame-monkey in Episode 2, leaving little in the way of surprise to contribute to a dramatic climax for Episode 3.

Also, Episode 3 originally was intended to include a sequence in which the children and trainees gather around a Chaac-altar and make frog sounds, to invoke rain. This is an actual Maya custom that has survived to the present day, and was a component of several rehearsals. Ultimately it was dropped from the final plans for episode three because the world builders were focusing their energy on the excavation.


Design. The three principal issues of world building are design, geometric construction and the development of behaviors for objects. The students designed the Caracol world by storyboarding and scripting the action, while examining maps provided by the archaeologists. A large scale map measuring 16 feet by 18 feet was laid out on the floor of the CREAT lab using electrical tape, and used for 'blocking' - that is, for planning the initial positions of the avatars and of the principal activities. A cardboard model of the principal structures was constructed, to give some sense of the relative heights involved.

Construction. The three episodes used different VRML models, built by three different teams. Episode 1 focuses on Caana. Episode 2 uses a cave, which is entirely a fictional construction. Episode 3 focuses on the A Plaza, which consists of three tall pyramids and a lower but richly complex eastern structure. The team building Episode 3's world included several members who had been to Belize, and produced the most realistic of the three worlds.

The sets for the story's three episodes were constructed using Cosmo Worlds, which is the VRML editor developed by Silicon Graphics, Inc. This tool is satisfactory for some relatively simple purposes but has disadvantages when behaviors are attached to VRML objects. Specifically, the underlying VRML source code is rearranged and rendered much less comprehensible than it was before Cosmo Worlds' intervention.

In accordance with the spirit of theatrical set design, it was decided to keep the virtual world as simple as possible. Photographic texture maps were used to establish the overall sense of place. Since several project members had visited the actual archaeological site, the virtual world's evocation of the Belizan jungle could be verified.

To keep the computers' speed at acceptable levels, the VRML worlds had to be as small as possible. Consequently, the three episodes' worlds were disjoint. It was intended that the episodes would be separated by intermissions, as in a theatrical production, to allow for changing of the scenery.


Behaviors were programmed using JavaScript and VRML's built-in routing and interpolation mechanisms. It was difficult for the programming team to construct satisfactory solutions to some problems such as keeping the avatars' feet on the ground. Each behavior had to be debugged as a stand-alone phenomen, then passed to Dean Reed for inclusion into the networking system. This proved to be a major bottleneck, and an impetus to seek or build a system which could be provided to the world-builders at the outset.

Avatars. Members of several teams contributed to the construction of human figure avatars. Some of the figures were built using Sven Technology's AvatarMaker; others were harvested around the Internet and customized for this project. Photographs were taken in Guatemala at Tikal, of nine year old schoolchildren (with the teacher's permission) and used as texture maps to provide faces for a boy and girl avatar (Juan/Mouse Jumper and Ana/Lida.) The lead archaeologist's avatar was constructed using an image of Michael Moshell, complete with a polygonal model of his hat.

Episode 1's world was based on a high-detail model of Caracol's largest structure, the pyramid called Caana ("Sky house"). This model was initially developed by Tim Murtha, an anthropology graduate student, from AutoCAD data that was imported into ArcView, a geographic information system. Tim's model was elaborated by the CREAT students. Caana is part of a group of four structures called the B Complex. The other three structures were incorporated as flat images into a texture map arranged as a backdrop surrounding the action area. A visitor's center was also constructed, and served as a starting point for the modern episode.

To construct the ancient Caana, the team added roofs (Tim's model represents modern Caana, in which the temples are roofless.) and texture maps of conjectural painting on the roofs' exterior surfaces. The exterior surface of the ancient pyramid was shown as being covered in white plaster. A stela was placed in the plaza before the pyramid, for purposes of the story line. No actual stela has been located at this place in Caracol, but the Maya often moved stela around. Atop Caana, in the actual location of one of the tombs found there (below the left side of the north temple structure), the artists located Rainbow's tomb.

Behaviors for Episode 1. An animated toucan played an important role in Episode 1. The toucan existed in the modern time-frame and also in the past. It perched on top of the stela. Once the guests touched the Caracol glyph on the stela, the camera's viewpoint was bound to an over-the-shoulder view of the toucan. With an 'awwk!' cry, the toucan flew upward, following the steps on the front of Caana, to lead the viewers to the site where Rainbow was being buried. To the left of the tomb's entrance, a brazier was burning. In the flame was shown Rainbow's face. A live actor narrated her speech as described above and in the Episode I script in the appendix.

The toucan's flight path was automated so as to provide for precise control of camera position and direction during the flight. It was found that using a live camera operator for this rapid flying scene was very demanding, and usually resulted in a disorienting experience for the viewers.

Episode 2. This episode was intended to serve as a transition from the introductory experience in which the guests saw Rainbow's funeral, to the actual hands-on digging and discovery in Episode 3. The choice of a cave for this episode was motivated by the successful construction of a complex and interesting cave by one of the modelers. There are no actual caves at Caracol, but there are many tombs. The cave in the virtual world is, however, much bigger than any of Caracol's tombs.

This portion of the story was largely developed before the field trip to Caracol, and represents the largest divergence (in terms of "set design") from physical reality. In the actual performance, the role played by the cave is somewhat museum-like, in that the actors move from artifact to artifact and discuss them.

Behaviors for Episode 2. The principal technical problems to be solved for Episode 2 concerned the problem of zooming in on an object and viewing it from all sides.

The motion control paradigm that is provided by the Cosmo browser is based on using the mouse. The user presses the button, then drags the mouse upward to move forward, downward to move back ("zooming"), or to the left and right to rotate the viewpoint about a vertical axis ("panning") in the desired direction. The four cursor keys perform the same functions, and are often used instead of the mouse. This technique could be referred to as a "unicycle" steering paradigm. No convenient means is provided for traveling sideways or vertically, while looking forward ("dollying", in film terminology.) There are arrows on the "dashboard" display that supposedly provide these translational movements, but their operation is unreliable.

Even if the dollying function were convenient, it would require a coordinated dolly and pan, to go around an artifact and see it from all sides. This proved too challenging for our camera operators. We experimented with pre-programmed camera paths that circumnavigated the artifact, as well as with animated actions that allow the artifact to be rotated about its vertical axis. The latter technique is somewhat preferable, since circumnavigating the object produces a complex background field motion that distracts the viewer from the artifact itself.

Several nicely textured artifacts were developed for this episode. A Maya vase with a detailed painting of the ceremonial ball game was constructed, and made part of the narrative. However, it proved difficult to actually see and comprehend the figures on the cylindrical surface of the vase object in VRML. This was partially due to the cylinder's curvature and partially due to the difficulty of zooming in on an object in VRML, using the Cosmo control system. It would have perhaps been better to include a "hot spot" so that the vase's image was delivered on the screen as a flat, unrolled image (as it was originally found in a reference book, before we wrapped it onto the vase model.)

Episode 3. This was the most ambitious episode, in terms of behaviors programmed into the world. A suite of tools each had their own interactions with the environment. A shovel could be selected, and was seen to make digging motions. (The shovel was not "hand operated" because the avatars did not have articulated hands.) An excavation pit was provided, the bottom of which moved downward as excavation continued. At a succession of depths, each of four fragments of the artifact (amulet) was revealed. The artifacts had to be placed into the sifter in order to clean them, at which point the jade green amulet fragments became visible.

When the fragments were placed close together, they moved to form a single amulet and a bright light shone from it.


Each team assigned its members to various roles during the successive rehearsals; nobody "owned" a given role. The last six weeks of the course (twelve class meetings) were devoted to a rotating cycle of rehearsals. In general, we followed a cycle of Episode 1, 2 and 3, though the sequence occasionally changed when some key players could not make it to a particular class.

Four computers were available with high speed graphics cards. During the entire developmental period, our network substrate system was being built and improved. Seldom was it possible to actually have four computers operating at once, and of course we had more than four characters in most episodes. Typically we would not dedicate a machine to camera-viewpoint; and would have to share a machine for two characters. This constraint also meant that a number of functions originally intended for the chief archaeologist or chief priest (such as introducing a story segment and then departing) were taken over by essential characters such as Ana or Bud.

5. Lessons learned

Interactive Drama for Learning

Script Development. Throughout the fall semester, students developed many workable ideas for elements of the script, but no unifying treatment emerged. The leadership team seemed unable to produce a coherent story of sufficient simplicity. Ultimately, Moshell took the scattered pieces and wrote a two page story treatment that established the key character of Rainbow, the idea that she sacrificed a jade amulet to save her people, and that the archaeological project in the modern era could complete the pattern of Rainbow's life by restoring the amulet.

Once this thread was established, the students proved quite capable of embellishing the theme (and in fact of changing essential parts of it).

Leadership and Group Dynamics

Initially we were concerned that the programmers might be headstrong and would only provide lip service to the humanistic students' goals and scripts. This did not occur; in fact the humanists were intimately involved in building the worlds, and programmers tried very hard to meet their specifications. The technical students quickly taught some of the humanists how to construct VRML models, and incorporated some of their products into the scenes. However the bulk of the modeling was done by the technical students.

The team leaders developed skills in understanding their members' capabilities, and in motivating their members to meet deadlines. In some cases this required "working around" some members' limitations and finding meaningful niches for them. One student's forte, for instance, turned out to be the playing of the Fire-Monkey's role in dramatic enactment rehearsals. By training and background this student would have been expected to contribute mainly to the programming.

Based on experience with other group oriented projects, we were concerned that some of the students would not find meaningful roles. We measured group cohesion by soliciting individual private feedback from members and leaders three times during the spring semester, and found very few complaints on this score. There were of course different levels of individual effort observed by both faculty and peers, but no student fell below a 'B' level of effort. As the project proceeded, some students discovered new talents and interests and became "work horses", whereas others initially expected to play major roles were seen to fade into supporting roles. One could sometimes see students thinking through and acting out, for the first time, their own professional self-images and imagining themselves as key players in future real-world projects. Even though several students might have VRML or Java skills in a team, one person was usually trying to integrate the episode's components. This person quickly became a bottleneck, and we would frequently hear team leaders saying "as soon as Jack...." or other references to the fact that the process was stalled because the technical lead was overloaded. Despite specific written instructions which required that each student submit written status reports at certain checkpoints, only six of the eighteen students regularly did so. The usual excuse was "If I take time to write you a report, I'll have less time to work on the world!"

Role Playing

Each team know that a week or ten days in the future, it would be their turn to present an improvisational session. They would schedule their world-building to achieve a functional partially completed world at that point, would meet outside class to assign roles and refine the script, and would be sitting ready to go when class began on the day of their presentation. On a few occasions the teams provided their worlds to Dean for network integration too close to the deadline for him to succeed, and we had to defer those rehearsals. This issue is discussed below under Technology. Guests (who were sometimes chosen from within the design team, and sometimes played by students from other teams) quickly discovered that an important part of their task was to probe the cast members' knowledge and flexibility, by asking novel questions or wandering away from the expected direction of the performance.

The cast members, when asked a question such as "how many people lived on Caana?" would usually make wild guesses, so as to keep the enactment going. Particularly when the cast member was playing the role of an authority figure (e. g. the chief archaeologist) the impetus to sound authoritative was strong, and the answers sometimes ridiculous.

Our recommendation is that in the future, all members of the design and enactment teams be given a standardized set of readings, assigned to study them, and then be given written and graded examinations before they are "licensed" to serve as cast members and sources of information for guests.

Cast members knew whom they were playing, which characters were supposed to be children and which adults. However, the cast members were focused on conveying the facts about the archaeology site and the back story, and so they often spoke as adults even when playing childrens' roles. This was partly a result of the technical limitations which restricted us from having a full cast of characters in each episode.

Certain situations were incorporated into the script to emphasize the youth of the children. For instance, Ana and her little brother Juan would often argue about whether or not to tell these foreigners about the "ghosts" one could see from the past.

The Unanswered Question. From the Objectives, the main unanswered question is "How can the story be structured so as to lead to the desired outcomes?" The experiments reported in this paper were never conducted with truly ignorant guests. At most, guests for a given episode were provided by another episode's team. There was a general will to make the story go as it was designed (after all, the authors were also the actors.) Guests would sometimes (frequently) test the limits of the knowledge and improvisational ability of the cast members.

World Building

Early in the project, several potential team members with backgrounds in computer animation withdrew from the effort. They were correctly concerned that their efforts would be spent in constructing 3d models for realtime use rather than in designing and integrating a demonstration reel of animation. Without that demo reel, their employment prospects were faint.

Students who had participated in CAP4021 - Building Virtual Worlds - were familiar with the Cosmo Worlds 3d modeling system, which is a kind of a relatively low performance tool. Silicon Graphics had ceased supporting the tool and its status was in limbo, but it was adequate for some of our needs. VRML code was also directly manipulated via editing the source files.

As a result of this resource and skill deficit, we realized that we needed to develop a course in modeling for realtime applications. We acquired four licenses of the MultiGen Creator software system, equipped four PCs with Accel Galaxy high performance graphics cards, and hired an adjunct instructor to prototype a course on 3d modeling for realtime applications. While this course had no direct effect on the Caracol project, its creation addresses future needs of a similar type.

Programming the interactions between objects was an arduous undertaking, but was generally successful in achieving some approximation to the desired behavior. Avatars moved rigidly rather than with fluid strides, although by the end of the spring semester, a first crude approximation to a walking cycle was demonstrated. The students in charge of building avatars used photographs of Guatemalan children, Belizan soldiers and Michael Moshell to equip the avatars with human-looking faces. The pyramids, temples and jungle were all represented with actual photographic texture captured in Belize, or with carefully chosen generic textures. VRML's chief drawback as an environment for photo-realistic virtual worlds is the way it applys its lighting model to textured surfaces. This generally produces unfortunate color distortions that vary strongly with viewpoint and view direction, and act to reduce the net benefit of the texture maps. Before March of 1999, the project's only sources of information about Caracol's geography were maps, photos and two videotapes provided by the Chases. From these, the students constructed a cardboard model and a large scale map. The cardboard model proved useful for planning activities that occurred on the plateau, but failed to convey a vertical sense of the site. Visiting the virtual worlds built during this period felt like visiting a cardboard model. One's attention is focused on the structures (pyramids and the ruins on top of them) because there was nothing else in the virtual world except terrain.

When you are at present-day Caracol in Belize, the jungle is the overwhelming presence. Usually you see jungle, with the occasional glimpse of structures (which mostly look like more jungle, just steeper.)

In virtual Caracol, the jungle is primarily represented by a panoramic backdrop which is comprised of photographic images of the jungle, as seen from the archaeology camp. Several layered images provide some motion parallax. The backdrop, together with a carefully constructed blue sky with clouds and the principal structures, is generally satisfactory in conveying the overall sense of the environment when the viewpoint is on the ground plane. However since there are only a few foreground trees, the experience of being atop most of the virtual structures does not resemble the real experience. The exception is Caana - the large pyramid. In this case, one can look downward into the completely cleared courtyard, which closely resembles in the virtual world, the actual scene in Belize.

Design rules would include standard ways of modeling human figures; scale information so that furniture, figures and terrain are of commensurate size; and an up-and-running version control system for archiving work. Fortunately, no important part of the work was lost during the development process. Dean Reed often had to make scaling corrections when integrating models into a scenario. An extensive model of modern Caana existed in AutoCad format. Tim Murtha cleaned up this model using ArcInfo software and exported it to VRML. This model then served as the starting point for the textured modern Caana model, as well as for the geometrically enhanced ancient model. Roofs were placed on the structures, temples were restored based on best available information, and exterior surfaces were colored with white plaster, red/orange plaster (for the apartment and temple complexes) or colored texture maps of Maya art, to represent conjectured images painted on temple roofs. Technology Trying to make something as complex as interactive drama work properly is daunting under any circumstances. To have done so with a substrate that was actively under development was a high-risk venture, but at the time we felt that we had no viable alternative. The Alice virtual world system, relatively stable and with an attractive scripting system, offers no documented and supported distribution capability. Commercial systems such as Alpha World do not support programming of avatar behaviors, except within the scope of an expensive server system.

Our approach of using three teams of programmers who were effectively in competition to produce the best episode, worked reasonably well. They produced interactive features as called for by their scripts, and then provided these scripts to Dean Reed for their integration into the distributed system. On most occasions, Dean was able to get the results to work in time for the next rehearsal, albeit with considerable personal sacrifice and time-stress. Dean's professionalism and hard work made the others' accomplishments possible.

The major breakdown of the system occurred at the climactic performance, when Episode 3's code would not run. Ultimately this problem was traced to a Linux server that had recently been reconfigured. However the Episode 3 design team bore some responsibility because of the rapid rate at which they were adding features, right up to the last minute. They simply overwhelmed Dean's ability to integrate and test features.

This latter guideline may be difficult to implement if memory space is scarce. Buy some more disk space if necessary. Human creative effort is a terrible thing to waste. Students trying to work at home on the development of behaviors found themselves critically handicapped by the slowness of VRML rendering, and often had to work in subset worlds to get the polygon count low enough to achieve anything at all. The computers equipped with Accel Galaxy boards in the laboratory were in high demand. A new feature

Dean Reed introduced an experimental new feature during the project. This consisted of a storyboard sequence down the left margin. Each frame of the storyboard appeared when a corresponding triggering event occurred (or could also be displayed under the control of the cast members.) The concept is that these storyboards would help to cue the cast as to how far through the intended interaction, the group had gotten to this point.

Figure 8. Ana at the excavation. (Storyboard glyphs on left margin.)

This feature was introduced too late for experimental use, but shows promise for managing cast activity in future projects.

User Interface

Cosmo's user interface is adequate for moving forward, but it is very awkward for circling around an object which you wish to observe closely. The viewpoint motion is forward but the object is off to the side.

One obvious means of interacting with objects in virtual worlds is to click on them, and then select from a menu of possible actions. This is the paradigm used in ExploreNet, and is also supported in VRMINet. However there were few behaviors available, and no "stock" behaviors such as pivoting an object about its vertical axis for close inspection.

Over-the-shoulder and out-the-eye viewpoints were provided for avatars, via the "out of body" option in the dashboard control.

A major source of confusion was that there was no convenient means to tell while looking at the screen, which of the visible avatars corresponded to an actor who was speaking. In text based systems like ExploreNet, a character's speech appears centered over their location. In Alpha World, a similar approach is used in three dimensions. However in VRMINet, this space was used for labels that identified the avatars.

Since verbal rather than typed speech was used, it would have been challenging to display the avatars' actual speech over their heads. At a minimum, some sort of motion cue should have been provided.

The use of identifying labels over avatars had a significant disadvantage in a world with as much vertical structure as Caracol. Often the textual label would partially obscure a temple, another avatar or other target for action that was above the labelled object in the current view. Some of this obscurance can be seen in Figure 3 above, where the toucan is standing on, and partially obscuring, a caption of the stela.

Capturing the Final Product

The virtual world, comprised of three episodes and inhabited by avatars, is available today for inspection. However, simply touring the world cannot begin to convey the dynamic sense of a drama in progress. We videotaped the first episode, during the final full-dress performance on April 22. The second and third episodes had technical difficulties that made it impossible to complete the performance, despite several hours of overtime effort. The second episode was ultimately conducted by using a previous version of the model. The third and most ambitious episode had no fall-back model, and so ultimately was not performed.

After the course was completed, Terence Zagers, Joe Block, Ami-Sun Tibbey and Jason Montilla constructed a videotape about the making of Caracol Time Travel.


Experimentation with virtual drama is very challenging. To have much of a chance of success, it is highly desirable to have a reliable and well documented technical substrate, including modeling systems, in place at the outset. The student team should include subject matter experts, skilled 3d modelers and programmers, as well as humanists with a flair for storytelling. Drama students would add ideas and know-how about how to organize a theatrical production.

All of the above elements were missing from the Caracol Time Travel Project, and yet many useful lessons were learned, and a compelling virtual performance was achieved. The final product was perhaps 50% of the way toward being sufficiently mature to undergo field trials in middle schools. It remains to be seen whether the dramatic component of the project will be continued by other students. The VRMINet software will serve as the basis for another round of design and extension of capabilities. The VRML models of Caracol are being adapted for use as part of the Caracol Archaeology Project's Web site, and for possible use in an exhibit at the Orlando Museum of Art.


The authors wish to express their gratitude to Tim Murtha, Matt Tarr and Neal Henry, members of the Caracol Archaeology Project, for their advice and assistance in designing and constructing Caracol Time Travel, and to Jason Montilla, Shaun Murphy, Joe Block and Linda Upham, CREAT's Resource Manager and lab technicians, for making things work when they had to work.


Chase, D. Z. and Chase, A. F. See list of publications and reports at 1999.

Cruz-Neira, C., Sandin, D. J., DeFanti, T. A., Kenyon, R. V., and Hart, J. C. "The CAVE: audio visual experience automatic virtual environment,Communications of the ACM Vol. 35, No. 6 (June 1992), Pages 64-72

CTG (Cognition and Technology Group at Vanderbilt) Designing learning environments that support thinking: The Jasper series as a case study. In Duffy, T. M., Lowyck, J., and Jonassen, D. H. (eds). Designing Environments for Constructivist Learning. Springer, New York. 1993.

Hayes-Roth, B; Brownston, L., & Gent, R.v. Multiagent Collaboration in Directed Improvisation. Knowledge Systems Laboratory, San Francisco CA, 1995. Available at


Hughes, C. E.; Moshell, J. M. and Pullen, M. (1998) Two dimensional Shared Virtual Worlds in Middle and Elementary Schools: Lessons Learned," Proceedings of the Virtual Worlds and Simulation Conference, pp. 139-144, San Diego,CA. January..

Hughes, C. E. and Moshell, J. M. (1997) "Shared Virtual Worlds for Education: The ExploreNet Experiment," ACM Multimedia 5(2), pp. 145-154, March..

Morningstar, Chip and Farmer, F. Randall. (1991). The lessons of Lucasfilm's Habitat. Cyberspace: First Steps. Michael Benedikt, ed. MIT Press, Cambridge, MA.

Jesup, George. 1999

Johnson, A., Moher, T., Ohlsson, S., Gillingham, M. "The Round Earth Project". IEEE Computer Graphics and Applications, to appear. 1999.

Kelso, M., Weyhrauch, P., and Bates, J. Dramatic Presence. In PRESENCE: The Journal of Teleoperators and Virtual Environments, Vol 2, No 1, MIT Press. Winter 1993.

Maes, P. "Artificial Life meets Entertainment: Interacting with Lifelike Autonomous Agents." Vol. 38, No. 11, pp. 108-114, Communications of the ACM, ACM Press, November 1995.

Moshell, J. M. and Hughes, C. E. (1996) The Virtual Academy: A Simulated Environment for Constructionist Learning. International Journal of Human-Computer Interaction 8(1), pp. 95-110.

Moshell, J. M. and Winn, W. (eds) A Special Issue on Virtual Environments and Education. PRESENCE: The Journal of Teleoperators and Virtual Environments. MIT Press. To appear in 1999.

Schele, Linda and Freidel, David A. (1990) A Forest of Kings: The Untold Story of the Ancient Maya. William Morrow & Co., New York.

Stephenson, Neal. Snow Crash (1993) Bantam Press.

VRML (1999). See

Youngblut, C.. Educational uses of virtual reality technology. Technical Report D-2128, Institute for Defense Analysis, Alexandria, VA. 1998.

The students of the CREAT 1998-99 Senior Project:

(The Belize Expedition team members are marked with *)

Team 1: Captain - Robert O'Leary
Frank Gonzales *
Dan Larsen *
Vivian Nguyen
Vincent Yue

Team 2: Captain - Mohammed Usman
Joe Block
Bryan Bloss *
Ahmed Hamman
Pierre Hood
Sangeeta Kulkarni

Team 3: Captain - Angela Leavell *
Christina Dejuan
Jack Ratiwanich
Gilberto Rodriguez
Terry Zagers

Team 4: (Introductory Video)

Captain - Ami-Sun Kibbey
Joe Block
Barry Boulton *

Team 5: (Project Summary Video)

Captain - Terry Zagers
Joe Block
Ami-Sun Kibbey

Appendix A:

Caracol Time Travel, Script 0:

Chaac and the Blanket

Script developed by Robert O'Leary, September 1998

Cast - Outer Story

Ana: a 12-year-old Maya girl who is the narrator of the story. She is smart and proud of her storytelling ability.

Juan: a Maya boy, possibly a cousin of Ana's, who is approximately 10 years old. Like most boys his age, he is skeptical about everything and is not too keen about girls.

Jose: Juan's younger brother, about 8 or 9. Unlike Juan, Jose is enthusiastic about everything, particularly everything related to Ana.

Cast - Inner Story

Chaac: Maya god of fertility, agriculture and practically everything related to life in general. Although he has a fearsome appearance with dreadlocks and fangs, he really cares about the people who worship him. Basically, he is a benevolent god, but you don't want to be on his bad side.

Xumacane: Chaac's wife. Name will probably be changed later, for it is really the name of another character in Maya mythology. A nice older lady, albeit a bit scatterbrained.

Hun Chuwen: Older of Chaac's two messenger frogs. Put up with their master's demands with a sense of dignity and some under the breath muttering.

Vucab Chuwen: Younger of Chaac's two messenger frogs.


Ana, Juan and Jose are searching around the Caana[large pyramid at Caracol], where Jose happens upon a large painted gourd used to scoop water. He picks it up and shows it to Ana. Juan looks on, but tries not to look too interested.

Jose: Look, Ana! Here's a big gourd!

Juan: A gourd? So what? What's the big deal about a gourd?

Ana: It's a nice big gourd, Jose. Jose: Maybe it's Chaac's gourd!

Juan: Hah! That's not likely!

Juan's right. Chaac probably uses a much bigger gourd than that. But it's a very nice gourd, Jose. Jose: You know a lot about Chaac, don't you, Ana? Tell us a story about Chaac.

Juan: A girl tell a story? Girls can't tell stories!

Ana: (getting a tad upset) I'm a girl, and I can too tell stories! In fact, I can tell them better than most boys! Jose: Ana tells the best stories! Better than Tio Pedro!

Ana: (blushing) I don't think I tell them as good as Tio Pedro, though. He's a very good

storyteller. Juan: That's for sure!

Ana: But I can tell them better than you and other boys!

Juan: We'll see about that. So, Ana, can you tell a story about Chaac?

Ana: Of course. I will tell you how Chaac got his blanket of clouds.

Jose: All right!

Juan: Please, tell us.

Ana: Chaac was in the middle of the sky one day, looking for a spot in which he could look over the Earth. The wind was blowing hard...

Jose: How hard?

Ana: Very hard! Hard enough so birds could only fly backwards! And Chaac was getting very, very, very COLD!

Juan: Real cold?

Ana: (shivering) Ex-x-x-treme-ly C-C-COLD! Chae called his two messenger frogs,

Hun Chuwen and Vucab Chuwen: Chaac: (Ana in a deep voice) I am getting too cold here. Please go back to my house on the edge of the world, and ask my wife to get my blanket. Then bring it back here to me.

Jose: And the frogs said "F-ibbit!, Ribbit!"

Ana: They said:

Hun Chuwen: Got to get the ribbit blanket!

Vucab Chuwen: Because Chaac is ribbit cold!

Juan: How cold?

Ana: (in a frog voice) Ribbit cold!

Ana: (normal voice) So the frogs hopped over to Chaac's house on the edge of the world.

They saw his wife Xumacane cleaning. Jose: What was she cleaning? Ana: (making sweeping motions) She was sweeping the floor, swish, swish, with a branch ... and singing to herself when the frogs came. She didn't notice them until Hun Chuwen said "Ribbit!"

Jose: Ribbit!

Juan: Ribbit!

Ana: And Hun Chuwen told her:

Hun Chuwen: We've come to get Chaac's ribbit blanket.

Vucab Chuwen: Because Chaac's ribbit cold!

Juan: Ribbit!

Jose: Ribbit!

Ana: Xumacane was a little startled. She looked around the house, but did not see a blanket. So .... she grabbed a....

Jose: A sword!

Ana: No!

Juan: A shirt!

Ana: No! She grabbed a large gourd, filled it with warm water.

Jose and Juan: Oh!

Ana: She gave the gourd filled with water to the frogs. They hopped back into the sky to

give Chaac the gourd.

Jose: Hop! Hop! Juan: Ribbit! Ribbit!

Ana: But Chaac was not happy when the frogs he saw the gourd. He told the frogs:

Chaac: "This won't make me warm!"

And he poured the gourd out over the earth, making rain ... chchchchchch.

Jose: Chchchhchchch!

Ana: And he turned to the frogs and commanded them:

Chaac: Go back to my house and get me a blanket!

So the frogs hopped back to Chaac's house at the edge of the world

Hun Chuwen: Got to get a ribbit blanket!

Vucab Chuwen: Because the gourd's no ribbit good!

Jose: Hop! Hop!

Juan: Ribbit! Ribbit!

Ana: When they arrived at the house, Xumacane was busy shucking maize for cornmeal

to make tortillas. The frogs startled her once again. Hun Chuwen: We've come to get Chaac a ribbit blanket! Vucab Chuwen: Because Chaac's ribbit cold!

Jose and Juan: Ribbit! Ribbit!

Ana: Xumacane once again looked around the house, but could not find a blanket.

However she did find a....

Jose: A pot!

Juan: A hammock!

Ana: No, she found a torch. She stuck it into the fire to light and gave it to the frogs.

Xumacane: Here! A torch will keep him warm!

The frogs took the torch and hopped back to Chaac in the middle of the sky. Jose: Hop! Hop!

Juan: Ribbit! Ribbit!

Ana: Once again, Chae was not happy. He grabbed the torch from the frogs:

Chaac: This is not what I wanted! How is a torch going to keep me warm in this

cold sky!

With that he threw the torch to earth, causing sparks to fly all over .... Cn-rasshh..

The torch became lightning, and the sound of the sparks hitting the earth became thunder.
Jose: Crash! Ana: With that Chaac once again sent the frogs back to his house. They hopped back. Hun Chuwen: Got to get a ribbit blanket!

Vucab Chuwen: The torch was no ribbit good!

Jose and Juan: Ribbit! Ribbit!

Ana: Xumacane was now mixing the corn dough to make tortillas when the frogs came


Hun Chuwen: Chaac wants a ribbit blanket!

Vucab Chuwen: He won't settle for any ribbit thing else!

Jose and Juan: Ribbit! Ribbit!

Ana: Xumacane knew that Chaac was good and mad, but she did not know where any blankets were. So she ran around looking. She went outside, and she saw a...

Jose: Afire!

Ana: No!

Juan: A jaguar!

Ana: No, a cotton bush!

Juan and Jose: A cotton bush?

Ana: Yes, and it was filled with cotton bolls. She picked as many cotton bolls as she could fit in her arms and gave them to the frogs. Xumacane: Here! I don't have time to weave them right now, but if he puts them all together, it will keep them warm. So the frogs hopped back to Chaac, bringing the cotton.

Jose: Hop! Hop!

Juan: (in a frog voice) I don't think Chaac's going to ribbit like this!

Ana: Chaac sighed when he saw the frogs carrying cotton instead of a sewn blanket. But

by this time, he was so cold he would try anything. So he grabbed the cotton. Taking a boll, he stuck it to his shirt ... there ... and there ... and there...

Finally Chaac was covered in cotton bolls and he was warin!

Jose: Yeah!

Juan: Ribbit!

Ana: Chaac liked his cotton blanket so much, he wore it every time he went into the sky. He wears it whenever he brings the rain in his gourd and pours it over the earth .... cchchchchchchch.

Jose: chchchhcchcch.

Ana: And when he throws torches of lightning down ... CRASH!

Juan: CRASH!

Ana: So that is how Chaac got his blanket of clouds. How is that story?

Jose: That was great!

Juan: It's okay. You told it almost as well as a boy would.

Ana: There is no pleasing you, Juan!


Appendix B: Three Scripts for Caracol Time Travel

Episode 1: First Trip to Classic Maya Times

Script developed by Robert O'Leary, January 1999

Setting: Modern day Caracol in front of Caana. Dr. Henry Smith, Dr. Dana Smith, Bud Abbott and Louise Castle, give a quick tour of the Caracol ruins while standing in front of Caana.

Dr. Henry: Over to the left is the ballcourt, where we found an altar stone giving us muc Caracol's history. The Maya played a game which was like a combination of basketball and soccer. Players had to hit the ball through a hoop, but they could not use their hands.

Bud: They played hard, for often members of the losing team were sacrificed. I know if that were the case with me, I would definitely play to win!

Dr. Dana: And over to the left is the main part of the town, and behind us here is the centerpiece of Caracol, Caana. This was their great pyramid. Priests made astronomical observations from its top, and sacrifices were performed up there. But Caana also had the great houses of the priests and the elders. In fact, we recovered a woman's body buried near the top. We don't know much about her, but from the location and artifacts we found, we know she was one very important lady.

Louise: Approximately 1,400 years ago, this was a city of 150,000 people. Caracol was at its height, having just defeated its arch-rival Tikal in a bloody war. Lord Water was the leading noble in the ancient city then. But about 400 years later, there was nobody left. No one is sure what happened here. There are many mysteries to be solved.

Dr. Henry: We have to go take care of some things, so why don't you just wait right here. We'll be back soon.

The four archaeologist exit viewer screen left. From the right, two children appear. One looks to be a girl around 12, and the other is a boy who looks to be around 8 or 9-yearsold. Both have black hair, light brown skin dark eyes and wide nose that many Native Americans have. The girl has a toucan on her shoulder.

Ana: Oh, you must have just arrived! I am Ana, and this is my younger brother Pedro. Ot aunt, Dona Marina, is the camp cook here. We are Maya, and they say our ancestors lived here many years ago.

Pedro: I wonder if the archaeologists are going to find a gravesite today, Ana. You remember when they found that body on the top of Caana?

Ana: I do indeed. They took the body back to the museum in Belmopan. They said she was a great lady from ancient times. She had to be important to be buried where she was. But I heard that they could not find an amulet. I don't know what was so important about the amulet.

Pedro: Maybe we could ask Dona Marina! She says she sees some of the moonlight


Ana: Shh! I don't know if these people know about the moonlight people.

Pedro: They probably do, don't you guys?

Hopefully the guests reply that they haven't. If they have, Ana says, "no that's not the people I am talking about"

Ana: Pedro had to blab! I don't know if you all would understand. You will probably think us all loco! You see, the archaeologists know and love Caracol. But they don't see all that my grandmother Dona Marina sees. She says that late at night, she can see the old Maya people walking around. Mostly, she sees them walking around with their basket and pots. But sometimes, she sees them do strange things, like having fights and sacrifice. Sometimes, I think I see them.

Pedro: Are they ghosts?

Ana: No, they aren't scary like ghosts. It's more like a dream.

Pedro: So she can see Maya people!

Ana: You can see Maya people too! Look at us.we're Maya people!

Pedro: No, the ancient Maya people.

Ana: I think so. It's about this time.close to sundown and the shadows grow..I stand by this stela and start hearing voices.

Murmering in the background along with drumbeats.

Ana: I stand by the stela, put my hand on this symbol, and they start to become visible Ana places her hand on the stela as people fade into the background.

Pedro: What if I put my hand on this symbol?

Pedro does, and another Maya appears faintly in the background. Meantime, there is one glyph on the stela that seems brighter than others. The bird on Ana's shoulder implores the guests to touch that symbol.

When someone does touch the bright symbol suddenly everything changes around them. The stars start shining and more of the ancient Maya appear. Buildings form, the grass disappears and Caana transforms into a shiny white structure, just as it was when it was built. There are also more people, dressed in traditional manner, but Ana and Pedro can not be found. The only one from modern times is Awwk, Ana's pet toucan. The guests themselves becume translucent.

The sky is dark and cloudy, and lightening scares Awwk to start flying. He flys up the steps of Caana as the camera follows. Near the top, a priest is conducting a funeral with his aides. They are putting the body of a woman under a carved witz monster.

High Priest: We consign the soul of our beloved Rainbow Queen to the Bacaab of the East. Guard her well, 0 Bacaab, in her passage into the underworld. We will never forget what her magic has done for us.

The aides close off the tomb as it starts to rain. Awwk starts flying down the pyramid down to the base of Caana where the stela is. The bird lands on top of the stela and look-down on a young girl about Ana's age. She is crying. A young boy (Mouse Jumper) is standing by her. Behind them, a small bonfire is burning despite the rain.

Mouse Jumper: Lida, it's starting to rain. Mother Rainbow is buried now, and there is nothing we can do.

Lida: But I miss herl I can't believe she is gone!

Mouse Jumper: C'mon. Let's go home and get something to eat.

Lida gets up and follows Mouse Jumper, who runs ahead and jumps into a puddle near the stela. She looks forlornly at the ground but looks up to notice a bonfire burning despite the rain. She hears someone calling her name, and the voice is coming from the bonfire.

Voice: Lida! Lida!

Lida starts seeing a image appear in the flame. It is her mother Rainbow, coming out of the flames in her priestess outfit.

Rainbow: Lida, you must help mel Find my amulet!

Lida: Your amulet?

Rainbow: Yes, it is a small jade frog on a string. I had to part with it many years ago. But without it, I can not enter the underworld. It was taken away in a small red pot. Find the pot, and maybe you can find the amulet. When you do, put it near my grave, then I can enter Xibalba.

As Rainbow talks, images of the jade amulet and the accompanying pot appear.

Lida: I'll try, Mother. But where will I find this amulet?

Rainbow: It is here somewhere buried in Caana. Use all the magic I taught you to find iti

Lida: I will use whatever magic possible to bring it to you, Mother. Whatever it takes!

As Lida makes her promise, she and the ancient city fade away, and the modern ruins of Caracol appear once again. Ana reappears. The guests, who were watching all this, lose their transparency.

Ana: Where did you guys go? The archaeologists wondered where you went?

Ana queries the guests on what they saw, making sure they tell of Rainbow, the jade amulet and the pot.

Ana: I wonder if she ever found the amulet? I hope she did! It would be terrible to be between here and the underworld all this time!

Pedro: Hey! Look what I foundl

Ana: Pedro! The archaeologists don't want you digging in the ruins!

Pedro: But look at this!

[original text]

Pedro holds up a shard of red porcelain, looking very much like it came from the pot where the amulet is supposed to be contained.

[corrected text]

Pedro points to a shard of red porcelain sticking from the earth, looking very much like it came from the pot where the amulet is supposed to be contained.

Fade out..end of episode.

Caracol Time Travel, Episode 2:

Exploring the Cave and Meeting the Fire Monkey

Developed by Mohammed Usman and Sangeeta Kulkarni

January 1999

CAST MEMBERS: Bud, Anna, Pedro

GUESTS: Jay, Pat

OBJECTS: Cave, Artifacts (4), Map, Bone.

PAT : Oh! There you are. Where have you been?

BUD : We just returned from the excavation site.

JAY : Is that the one Dr. Smith told us about?

BUD : Yeah, this is the latest one and there are many things that

may interest you.

PAT : Things like what?

BUD : You'll see! (In an excited tone)

PAT : Are you telling us that, we can go and visit the actual site.?

JAY : Can we? Please, please, please!

BUD : Sure!

BOTH : Yes!

BUD : I'm afraid that I won't be able to go with you, but Anna and Pedro

have agreed to give you a tour!

ANNA : We'll be more than happy to take you there,

JAY : Thanks!

PAT : Great! Then let's go!

BUD : Guys, please remember not to touch anything.

(If you want more information, just click on the artifact)

JAY : Yes Sir!

(All four of them start walking)

PAT : How far is this site?

PEDRO Actually, it's a cave and it's not very far.

JAY : A real cave? You mean like in ancient times?

ANNA : Yes, very much so. (pointing to the tiny entrance..)

JAY : There it is, I guess!

PEDRO: I think you are right!

(as they reach the cave entrance)

PAT : Hey, look how neat this entrance is. Was there a door on it.,

PEDRO: Could be. Now let's be very careful as we get in, we want to retain everything as it is.

JAY : Yes, we understand.

PAT : What's in here?

PEDRO: Let's explore! (Starts walking)

JAY : Hey, what's this? (Pat and Jay stop as they notice a sculpture, Pedro is still


PEDRO: What's what?(Pedro notices that they have stopped and walks back to them)

PAT : Look at this!

JAY : Is this one of the Maya Gods?

ANNA : Yes, it is indeed! But this is basically an Arm-ornament.

PAT : Really?

ANNA : See, how it is open at either end? There must be another similar one some where.

JAY: Perhaps one for each arm, eh?

PEDRO: This was worn on ceremonial occasions by noble personages.

PAT: Cool!

ANNA See that cylindrical vessel next to it? If you look at the design carefully it's a Ball- game scene!

JAY Oh, yeah; it sure is!

PAT What is it made up of?

ANNA Fired clay and slip.Ceramic vessels of this type were commonly decorated with ritual scenes relating to aspects of the ball-game.

JAY Was the ball-game also played by noble personages?

PEDRO Yes, the figures' rich paraphernalia indicates their high rank within Maya society.

PAT And what could this be?

PEDRO If s a Jade plaque! If you look at it carefully there is a frog on it!

PAT That sounds funny! Lord and Frog?

JAY It's hard to believe that this is more than a thousand years old!

ANNA Wait till you see this! (Anna points to a Box)

BOTH Wow! Beautiful!

PEDRO Isn't it? This box is made of hard wood and has excellent carving!

JAY This looks like the figure of a Maya Lord.

PEDRO You are right! This dates back to seventh century A.D.

PAT : You know a lot of things about these!

ANNA : It's our culture, we grew up listening to such stories of ball games,

Wars and droughts.

JAY : I would love to hear such stories!

PAT : What this could be? Just an open box?

JAY : Let's look at the carvings, it might have some clue...

What's this?

PEDRO :It's called an amulet. Are there two of them?

JAY : One of them looks like to be broken but the other one is intact.

PAT : Maybe some one fixed it!

ANNA : That looks neat though.

PAT : What's in the box?

(all three peep in and find a bone inside, by clicking on the bone a

fire emerges out of the box and the Flame-monkey appears)

FLAME VOICE: These are the bones of my ancestors and I speak for the spirits of the forest. I must tell you of how myself, and many others of the forest came to be... (the young voice narrates.)

No one ever believed me when I said that Misssaira was a chartzcoal (Maya term for witch). I heard the day she was bom; a curse was beset upon her to never feel the warmth of love. From what I know, she has no family, no children or friends. 1 heard others speak of Misssaira creeping into the graveyards late a night to feast on the bones of the dead. And on full moons, she was rumored to have been in discourse with the Dueno of Death! During the days when we all work hard, Misssaira strolls around the village, rummaging through people's garbage, When it is time to eat, she is never around. In fact, no one has ever seen her eat. And when she is around, she's seen talking to the animals in the forest. Like she can understand them or something.

She has a reputation for laughing uncontrollably, like she knows something you don't. But everyone in the village says she's just a sick, old woman. That she is lazy and crazy, so let her be. And I agree she indeed is a crazy woman.

But she did have one friend, who we know as Rainbow. Rainbow was a very special friend to her, for she is the one who neverjudged her. She took Rainbow under her wing and taught her the ways of magic. One item in particular was an amulet, which she gave to her favorite student. f was told that most, if not all of Missaira's powers lay inside that special amulet.

That season, there was a terrible draught, no rain meant no food for the people of Caracol.

Chaac was very angry, and nobody knew what to do. Except for Rainbow. Rainbow knew she had to appease the angry God. So she broke her magic amulet and offered it to the rain god in order to save her people and her country. And it worked! But Rainbow's soul was trapped in this world and her spirit cannot be free unless the magic amulet is returned to her!

Missaira has turned some of us (the ancestors of Caracol) into monkeys, because she wanted this story to be told and Rainbow's spirit to be set free! There are many others like me, all over the forest. Forced to eternity as animals of the forest. Our most important task as monkeys, is to serve as scribes. To draw hieroglyphics onto pottery, structures, plates, walls, vessels, bones and other artifacts which may inform others of our presence. We hope for others to understand so we may someday become human again and free Rainbow's soul.

(The fire and the monkey fade away.)

All three are looking at the box, finally Anna manages to speak

ANNA : That was some story! Now before it gets too dark let's move!

(All three leave the cave, as they walk out sunset animation is to be seen)

Caracol Time Travel, Episode 3:

Rainbow's Sacrifice

Developed by Angela Leavell

January 1999

Archeologist: The discovery of the bone at the cairn leads us to believe that there is something of importance about the site. We have decided to do a dig at this site to reveal other information about the ancient Maya. Go there immediately and your team captain will explain the procedures of excavating a site.

(Upon arrival, the team captain greets the guests and begins to speak)

Team Captain: O.K., Team. Everyone will be given tools for the dig, but first we have to explain about them. Since the site has already been surveyed, our first step is to remove the first layer of dirt very carefully. We don't know how far down our artifacts may be so we have to be cautious not to destroy any of it. Everyone team up with a partner. one will dig and one will sift the dirt. Sifting is an important job because many of the artifacts we can find are small. Move next to your partners now and decide who will do which first. There may be time to switch roles later so just pick one. (once this is done)

Sifters, your job is to take the dirt that your partner gives you and put it *in inside the sifter, which is this box with screen inside. I am sure that everyone knows the concept of "panning for gold", well, this works the same way, just without the water. once the dirt is inside, simply click on the sifter to shake the dirt out and reveal any hidden artifacts. (Demo)

Diggers, your job is to basically remove the dirt from the site and give it to your partner to sift though. This does not mean you can take the shovel and burrow your way to the center of the earth. The artifacts can be very delicate and although the first layer of dirt may be removed by a shovel, the lower layers will need to be dug out with your trowel. The trowel is the hand held tool that looks like a miniature shovel, except more diamond in shape.

If you find a large object, you may need to use the dental pick to remove the dirt around it, or on it, as to not damage the artifact. A brush is sometimes necessary as well when revealing larger items. Let me know if you have any questions.

(When questions answered) O.k. team, get to work!

List of Tools:

Sifter (Screen inside box)

Marshalltown Trowel

dental pick



(When the Red-painted pottery is revealed)

Team Captain: This looks familiar. It appears to be a broken shard from the burial at the top of Caana. Good work, team. Keep at it. If you want to switch roles with your partner, you may do this now. Instruct your partner how to do their job.

(When all items are revealed)

Archeologist: It seems that the red stone artifacts are broken bits of something larger. See if your pieces fit with anyone else's. In this way, we can reassemble the artifact.

Part two

(When the amulet is reassembled, the monkey kids come out of the forest and morph into Maya kids before the audience and cast.)

Monkey kid #1: Finally, the spell is broken! That old mean witch said that if the amulet was put back together that we would be human again.

(Cheering from the other monkey kids)

Monkey kid #2: She also said that we must give the amulet to Rainbow, but we saw the people take her away from her tomb on top of Caana.

Archaeologist: Tomb on top of Caana? Oh, that must be the women's body we

found last month. We took her to The Museum in [Belize City.] -> Belmopan.

Monkey kid #3: You must take us there! We have been waiting for over a thousand years for this moment. Once the amulet is taken to Rainbow, she will be able to enter the underworld, Xixalba, and our duty to her will be complete.

Archaeologist: But [Belize City] Belmopan is a four-hour drive from here.

Monkey kid #2: Rainbow had to break the amulet to save our city once and now it is our turn to save her. (Shouting) Let the gods give you sight!

(Insert transition here)

(Time travel occurs)

(The guests are at the dig site, but the cairn is now a stone altar. There is a distinctive circle drawn around it that encompasses them. A young boy greets them)

Boy: Oh, I'm glad you have made it in time. The ceremony is about to begin. I'm sure Rainbow will please the gods enough to make it rain, especially if you help. We are the frogs of the ritual. when, Rainbow enters the circle, we need to chirp like Chaac's helper frogs. This will help to get their attention to our ceremony. It is very important that you do this, or else this drought will never end and we will all surely die!

(Drumming starts)

Boy: Oh, it's starting, stand near the outer edge of the circle, like me.

(The guests see Rainbow enter the circle. She is dressed in the elaborate priest outfit and holding a gourd. The amulet is hanging around her neck. She gently places the gourd on the altar and removes her necklace.)

Rainbow: Hear me Chaac! Some gods may say that the people of Caracol are not worthy of rain, but I am here today to say otherwise. Our people have shown their strength by defeating not only Tikal, but also Naranjo. But our strength is nothing without the gods! The gods must preserve the people of Caracol for all times! We must not starve to death because you do not give ,us water! our water reserves are low, and we need you, Chaac to make it rain and save my people. I offer you the highest gift, my life amulet, to show our humility and thanks.

(Viewpoint changes to a short animation of Rainbow's hands breaking the amulet and a light being released from within.)

(Rainbow picks up the gourd and faces away from the altar, dipping water out and moving around the circle repeating this behavior)

(Drumming intensifies)

Rainbow: (Chanting to the drums) Make it raini Make it rain! Make it rain!

(The crash of lightening is seen and heard. Rain begins to pour from the sky. Everyone rejoices and the guests are sent back to their proper time.)

Appendix C:

Reality and Fantasy in Caracol Time Travel

In the development of the story line for the Caracol Time Travel interactive drama, the following elements were based on themes and facts from the literature:

- the legend of Chaac, pouring water down from the sky using a gourd

- the actual geography of Caracol

- the existence of a noble female burial on top of Caana, the largest pyramid at Caracol

- the common use of jade amulets

- the importance of sacrifice in winning the favor of the gods

- the importance of rain to the agricultural economy of the Maya

- the legend of monkeys being naughty boys transformed by a witch

- the legend that the inscriptions on temples and monuments were made by the monkeys

- the artifacts found in the cave, and their uses

- the altar and ceremony for Chaac, with children imitating the croaking of frogs

- the altar is realistically modeled after the circular three-pedestal altars of Caracol

There is some evidence that images were painted on the roofs of the temples, but the images actually used in the virtual world are taken from various artifacts shown in photographs in reference books. There is no reason to believe that the temple roofs actually bore these or similar images.

Fanciful elements which were added to provide a more substantive story included

- the rest of the Chaac introductory story, with firebrands and cotton for a robe

- the idea of a "life amulet" and its significance for the afterlife

- the existence of a large natural cave at Caracol

- all the personal names in the story

- the specific stela standing in front of Caana, where the toucan perched.

- a paper map in the cave. No paper maps from the Maya have been found.

The plazas upon which the pyramids and other structures stood, was covered in white plaster during the Classic Maya period. However it is colored brown in the virtual world, because a white plaza makes the white pyramids difficult to see and understand, in a low resolution PC-based virtual world.

Appendix D:

The VRMINet Virtual World Networking System

Developed by Dean Reed and Charles Hughes

June 1999


VRMInet: Networked Virtual Reality. VRMINet is a system in which multiple users may interact in a VRML world. The name "VRMINet" is based on VRML + RMI (Remote Method Invocation), and also reminds one of the occasional use of "vrmin" as a humorous and somewhat pejorative term for geeks who have access to no "real", professionally supported graphics systems. In our case the choice of VRML was not driven by lack of alternatives, but by a combination of technical and political factors.

VRML viewed as a graphical representation system is a relatively mature and powerful scene graph based architecture. It was developed by people who had extensive experience with Silicon Graphics Performer and Inventor. VRML's shortcomings as a basis for substantial projects are based on two problem areas: its relatively low performance, and the lack of a large body of ongoing development in substrate tools such as rendering engines, browsers, optimizers and utility libraries. Low performance manifests itself in numerous ways such as simplistic lighting models, unreliable texture handling and rapid performance degradation as the world's polygon count increases. Performance degradation occurs because there are no effective level of detail or load module control mechanisms. Performance also suffers because the Cosmo browser reinterprets the ASCII text of the scene graph, rather than pre-compiling into an efficient intermediate form (as was done by the now-defunct WorldView browser from Platinum/Intervista. The "big boys" in PC-based graphics remain Direct3D and OpenGL, with hardware-specific systems such as 3Dfx's Glide environment also receiving attention.

Microsoft and Silicon Graphics are developing Fahrenheit as a next-generation multi-platform system intended to combine the performance advantages of Direct3D with the cleaner architectural ideas from Inventor and VRML. However, it uncertain if or when Fahrenheit will move from alpha to beta testing. Java3d is receiving a good deal of attention, though it still suffers from some of the same kinds of immaturity as VRML.

ExploreNet: Distributed VR for Education. ExploreNet is a different kind of system than VRMINet, in that it is intended to be a tool for use by children and educators. As such, ExploreNet incorporates a scripting language and a carefully designed user interface. ExploreNet's scripting language needs a complete redesign, and its user interface will need many improvements before it is adequate for 3d use. However ExploreNet embodies lessons learned in seven years of experience that should carried forward.

The currently available VR system most similar to ExploreNet in terms of its target audience (education) and level of development is Randy Pausch's Alice system. Alice has a much more mature scripting language and user interface than ExploreNet, but has no publicly available mechanism for building or operating shared virtual worlds. Alice is very well programmed and achieves high graphics performance due to its careful use of Direct3D.

In the best of all possible worlds, ExploreNet 5 would have Alice's reliability and ease of use, VRMINet's sophisticated approach to networking, and many features of ExploreNet 4's user interface. Indeed, we wish we had had such a system for building Caracol Time Travel. But one uses what one has.

VRMINet: The Technology

Part 1: The User Interface

Worlds are hand-built for use with VRMINet. The manuals located at (Reed 99) explain the process.

When a user logs in, the following Java applet screen opens. It can be minimized at any time and then redisplayed by clicking on its icon in the task bar.

The Cosmo browser is running, and the user is initially controlling the position of the avatar which was selected - and seeing the world through its eyes. If the "Out of body" button is clicked, this stations' viewpoint is detached from its avatar and can be pulled back to get an over-the-shoulder view; or indeed maneuvered in such a way as to frame up any desired view of the scene, with or without including one's own (static) avatar in the scene. The same button toggles the viewpoint back into the avatar.

The "Online users" window provides the user with a list of the other humans in the world. The available sounds menus contain, respectively, sound effects (or utterances such as "hey!"), and short canned speeches (if desired.) The "available gestures" menu selects behaviors your avatar knows how to carry out. The "free bodies" window lets you take into your possession any un-owned objects in the scene.

Free objects (which are not avatars) in the scene also have an ExploreNet-like control mechanism. You can click on the object and see a menu appear within the scene, listing the optional behaviors you can specify for this object.

Part 2: Behind the Scenes

VRMINet Version 0.4 is implemented using Java 1.1 specs and relies on RMI (Remote Method Invocation) for message passing over a TCP/IP connection. This system ties Java and VRML together, via external-authoring interface (EAI) classes provided with Cosmo 2.x, in such a way that traditional object oriented behaviors may be associated with an object written in VRML and then distributed over the network.

Rendering Layer: Cosmo Player-VRML.

In the fall of 1998, empirical tests were run using Java 3D and VRML as a rendering device for the VE. It was shown at that time that VRML generally outperformed Java 3D and had the additional benefit of having world generation tools and world browsers readily available.

The choice of the Cosmo player as a rendering device was made largely due to the current market share of the product. Cosmo player has in the past been one of the most widely used and supported VRML browsers. Another critical factor in using the Cosmo player is the existence of an external-authoring interface in which programs developed in Java can communicate dynamically at runtime with VRML objects. This EAI made it possible to implement a distributed virtual world system (DVWS) using the existing rendering layer and build Java based support objects that model the VRML objects.

Transport Layer: RMI. Many distributed virtual world systems currently exist but are optimized for scalability and network performance only. These systems generally implement TCP/IP or UDP network communication with socket bases and a rigid protocol. The result of these systems is a highly scalable but inflexible system that requires protocol extension on client and server components any time new features are added.

The Caracol project required extensibility and flexibility versus scalability and high network speed. In Fall 98 we tested three distinct transport layers for the impending system: socket base, Remote Method Invocation (RMI), and JINI/Tuple-Space. RMI was chosen as the transport layer due to its acceptable response time and flexibility. RMI distributes objects among connected computers without the need for protocol extensions or explicit serialization.

Event System: Java Bean, delegation. Given that a flexible communication layer and a rendering layer exist, then conventions for object interaction must be established. Communication on the local (client) systems must be very extensible and easy to use. An event-based system is desirable in any simulation environment and is the method that we chose for VRMINet. Furthermore a one to one event system such as the VRML event system, which is known as containment, is rigid and can be hard to extend. One of the primary tasks was to implement a delegation based event system in which objects could dynamically register interest in each other and respond to events as needed. This delegation model resembles the Java Bean architecture in which a Model-View-Component interaction occurs. Furthermore, multi-threading the individual event component allows for maximum flexibility for the event receiver.

Discoveries. Much concern over the network performance of VRMINet was raised during the initial trials with the CREAT project. Player avatars seemed to suffer from network delay when a large load was placed on the system. Observation later showed that delay in the VRMINet system is largely related to rendering speed on the machine hosting the world.

Some minor performance boost was gained by allocating more CPU time to the rendering device via Thread, and running the VRML browser under optimal display speed conditions. Further performance was gained by decreasing the granularity of message passing for movements. On fast machines, the VRML browser produces a large number of movement events for trivial moves. It was observed that for a move of 1 meter, some fast machines could produce over 50 messages that in turn must be sent across the network and then executed by the receiving object and then rendered on the video display card. By reducing granularity, we were able to decrease the total number of messages passed on the network and thus reducing the number of rendering calls to the browser of the receivers.

Major reductions in performance were noticed in worlds that were built using extremely complex structures, or overabundant simpler structures. The VRML browser uses a great deal of resources when complex models are to be displayed, which leaves very little space for the JVM to execute. Continuous garbage collection can cause undesirable performance from the JVM and hence, the VRMINet system. Best results were obtained from worlds that were optimized by polygon count, model count, and texture

complexity. Optimizing on these criterions resulted in performance boosts for the overall system.

Overall, we have shown that in an environment with fast network connections, RMI provides an acceptable object distribution layer. Early empirical results show that a socket layer outperforms RMI by speeds of nearly 100% for straight message passing on a simple protocol. But since VRMINet is a distributed object system, arguments for a socket layer should be ignored due to the need for an extensible message/object passing system, which is not provided by a simple socket layer. A system could be built from sockets but this would be, in essence, just a re-implementation of RMI, which is built on sockets.

Integrating worlds into the VRMINet system. Taking a stand-alone world and integrating it into a distributed system is a complex task. One of the primary problems was the decision on what had to be distributed and what could be done locally. For example, in episode 1, a toucan flies the guests around the Caracol area in response to a touch on a magic stela. The authors of episode1 implemented the whole scenario in VRML without thinking of the consequences of moving the episode into a shared environment. Ultimately, much of the VRML work had to be undone and re-implemented using Java so that all of the players would experience the toucan ride at the same time.

For more details, see the complete system documentation at (Reed 99).

REFERENCES for this Appendix:

Reed (1999). www.