n response to the US National Research
Council’s 1997 report that specified a
research agenda for defense and entertainment, the
Modeling, Virtual Environments, and Simulation
(Moves) Institute proposed two games—America’s
Army: Operations and America’s Army: Soldiers—to
recreate the US Army for the beneﬁt of young civilians.
(For more information on the council’s report and how
this idea came about, see the sidebar “America’s Army:
Our goal within America’s Army:
Operations was to demonstrate life
in the infantry. The idea for the game
took shape as a ﬁrst-person mission
experience that starts with training.
For example, a player can’t use a par-
ticular weapon without ﬁrst qualify-
ing on the appropriate range.
We conceived America’s Army:
Soldiers as a realistic look at army
personal and career opportunities
via sophisticated role-playing. The player creates a
character with which to identify, much like the popu-
lar game The Sims, and moves through an interactive
story constructed spontaneously in response to the
character’s goals, resources, and values. (The charac-
ter’s values are based on the Army’s core values: loy-
alty, duty, respect, service, honor, integrity, and
Moves named the overall project the Army Game Pro-
ject (AGP). For the project, Moves envisioned a fully 3D,
accurate gaming environment with technological efforts
more complex than previous attempts. Our goal was
deep immersion in stories that were sensible-yet-
surprising, interactive, and nonrepeating. We also want-
ed to ascertain whether we could compute the aptitude
of users by their proﬁciency in play.
The two games were built in secret for 24 months until
we obtained results, secured approvals from the Army,
and debuted the suite at the Electronics Entertainment
Expo (E3) in May 2002.
Building the R&D teams
The ﬁrst funding for AGP arrived in May 2000. At that
time, a core group at Moves had constructed the large-
scale networked virtual environment, NPSNET,
had brought in John Hiles (who had worked at Maxis
on SimCity Supreme, SimAnt, and SimFarm).
Moves began building a development team by net-
working with friends and consultants. We recruited vet-
eran artists, designers, and programmers from industry
giants such as Maxis, Electronic Arts, Sony, and Kalisto.
Between them, their expertise had yielded some 31 com-
mercial games. On the research side, graduate students
from all branches of the US military and a number of
allied countries worked with Moves faculty on the tough
technical issues underlying the project. Practical sup-
port from the Army included unprecedented access to
posts, equipment, and subject-matter experts.
Twelve months from the May 2000 start, Mike Capps
and Alex Mayberry led the Operations team, while John
Hiles led the Soldiers team. The Operations team con-
sisted of 18 developers (level designers, artists, and game
programmers) and the Soldiers team consisted of 10.
Moves Institute researchers and developers visited
some 19 army posts and videotaped, photographed, and
recorded audio of everything that moved—and didn’t.
The team digitized video of soldiers in training, equip-
ment and weapons, texture details such as chipping
paint, and realistic minutiae such as the appearance of
sand at Fort Benning, Georgia.
The team learned to shoot M-16s and sniper rifles,
hurl hand grenades, and fire mortars. They went on
night parachute jumps with the troops and fed them-
selves to the K-9 corps (wearing padded suits). The post
visits created an evermore motivated and informed
group of developers, eager to attain higher levels both
literally and ﬁguratively.
America’s Army: Operations begins in single-player
mode as a new recruit ready to train. The player
embarks on basic riﬂe marksmanship and combat train-
We discuss America’s Army,
an Internet-based PC game
suite developed by the
Moves Institute under the
auspices of the US Army.
Michael Zyda, John Hiles, Alex Mayberry, Casey
Wardynski, Michael Capps, Brian Osborn, Russell
Shilling, Martin Robaszewski, and Margaret Davis
The Moves Institute
R&D for Defense
28 January/February 2003 Published by the IEEE Computer Society 0272-1716/03/$17.00 © 2003 IEEE
ing (BCT) through a representation of the ranges at Fort
The player’s range score determines his advance-
ment. If he scores poorly anywhere in training, he won’t
proceed to another task until his score improves. If he
scores well, the player may advance to M-24 riﬂe qual-
iﬁcation or US Army sniper school, where he will learn,
among other things, to breathe at the right moment in
the ﬁring sequence, and get the most from an M-24 by
using it on a bipod in a stationary position. (Although
the M-24 can be used while moving, as is common in
other games, the penalty applied to the player’s accu-
racy is severe.)
Weaponry is represented as precisely as possible—for
example, weapons must be loaded and cleared as in real
life, and the load is ﬁnite (Figure 1). The BCT weapons
familiarization includes the M-249 squad automatic
We achieved a high level of verisimilitude—soldiers
who know Fort Benning easily recognize it within the
game. For example, the Operations’ BCT obstacle course
is timed and sequenced as in real life.
Single-player training features the McKenna Military
Operations in Urban Terrain (MOUT) course at Fort Ben-
ning, including use of the ﬂash-bang as the player clears
a dark labyrinthine building of terrorist pop-up targets.
Training progresses to the US Army’s airborne school
250-foot jump tower.
Hearing is believing
Operations is rich aurally as well as visually. Based
on Moves’ research indicating that complex, multilay-
ered sound magniﬁes the sense of immersion in a sim-
ulation, the opulent sound created for Operations by
audio designer Russell Shilling pulls you inexorably into
In ﬁlmmaking, the rule is that if you see a sound, you
should hear a sound. We scrupulously observed this dic-
tum in Operations. Sound effects, weapons foley, and
ambiences were custom recorded or obtained from pro-
fessional libraries. Weapons animations, for example,
are accompanied by detailed and accurate audio repre-
sentations that focus the players’ attention on the
weapons and heighten their emotional impact.
For added realism, footsteps, bullet impacts, particle
effects, grenades, and shell casings are accorded tex-
ture-speciﬁc impact noises. A ﬂying shell casing clinks
differently on concrete, wood, or metal, for instance,
and the distinction is clearly heard in the game. Like-
wise, footsteps on dirt, mud, wood, concrete, grass, and
metal are sounded correctly.
In a typical Operations ﬁreﬁght, bullets whiz and crack
by the player’s ear, slam into the wall behind, and tin-
kle concrete and glass fragments at his feet. The player
hears his own shell casings thunk off the wooden door
behind him and ping the concrete ﬂoor. Meanwhile, to
the clatter of a nearby reload, the enemy creaks across
a steel catwalk overhead. The player hears a ﬂash-bang
grenade scud off the ﬂoor behind him just before being
incapacitated by the roar and ring of tinnitus in his ears.
We simulated the acoustical effects using OpenAL with
EAX 3.0 extensions from Creative Labs.
IEEE Computer Graphics and Applications 29
America’s Army: A Background
In 1997, the US National Research Council (NRC) issued a report
specifying a joint research agenda for defense and entertainment
modeling and simulation.
The report provided a guide to the R&D
necessary to build such systems. Included was an agenda treating
immersive technologies, networked virtual environments,
computer-generated autonomy, standards for interoperability, and
tools for creating simulated environments.
After the report’s publication, the Moves Institute realigned its
research directions with the agenda. The NRC report states that
games and interactive entertainment—not defense research
expenditures—have become the main technology drivers for
networked virtual environments.
According to the report, to keep up with evolving modeling,
virtual environment, and simulation technologies, the US
Department of Defense (DoD) needed to examine networked
entertainment to ascertain the potential for joint investment or
collaboration. In the fall of 1999, the Moves Institute’s Army Game
Project (AGP) emerged as one such potential investment.
The AGP effort originated with a discussion between the
directors of the Moves Institute and of the US Army’s Ofﬁce of
Economic and Manpower Assessment (OEMA).
The Army was concerned about falling recruitment and
perceived the need for new initiatives aimed at computer-literate
recruits for today’s high-tech Army. The discussion turned to the
medium of the PC game.
The Army had previous success using popular entertainment
media—they piggybacked advertisements onto newsreels in movie
theaters in the 1930s and 1940s and employed trailers in theaters
and Super Bowl TV advertisements in recent years. The emerging
question, then, was Could the Army use PC games for strategic
A dwindling number of young Americans have a veteran in the
family with whom to discuss army life. The directors imagined PC
games as a vehicle for communicating what an army career might
entail. They also acknowledged that, since no one had tried such a
thing, the only way to assess the idea was to actually plan and
build a game.
1. M. Zyda and J. Sheehan, eds., Modeling and Simulation: Linking Entertain-
ment & Defense, National Academy Press, 1997.
Operations benefited from the
advice of many talented people in
the entertainment industry, includ-
ing Gary Rydstrom at Skywalker
Sound, who provided helpful
insights into the design of the
weapons’ audio and proper use of
The game’s audio also beneﬁted
greatly from interactions with Dolby
Laboratories and received Dolby Dig-
ital 5.1 Surround Certiﬁcation. It’s
one of the ﬁrst PC-based videogames
to be released with this designation.
Operations team play
In Operations, no one ever plays a
villain ﬁghting the US. Both teams
always see themselves as part of the
US Army and perceive the other
team as the opposition (Figure 2).
Operations offers a variety of com-
bat scenarios, from desert to arctic,
jungle to swamp, to traverse in mis-
sion fashion (Figures 3 and 4). The
goal isn’t to blast everyone in sight,
but to cooperate as a team intent on
a purpose, which might be to iden-
tify a weapons cache, rescue a pris-
oner of war, or perhaps assault an
airfield. We designed scenarios so
that mission goals and objectives
make sense to both teams, requiring
one group to assault and the other
All players abide by rules of war-
fare. If a player violates the Uniform
Code of Military Justice, rules of
engagement, or laws of land war-
fare, reprisal is instant. He will ﬁnd
himself in a cell at Fort Leaven-
worth, accompanied by a mournful
harmonica playing the blues. Con-
tinued violation of the rules may
cause a player to be eliminated from
the game. To rejoin, he must create
a new ID and restart.
The game insists on the mission
orientation of the US Army. Above
all, soldiers must be team players,
following army values and rules.
Built on Epic Games’ latest Unre-
al engine, Operations can support up
to 32 players on a wide area network
and perhaps up to 64 on a local area
network. Nevertheless, as with many
other ﬁrst-person action games, the
maps in Operations have a limit of 26
players to keep teams balanced and
the bandwidth reasonable.
30 January/February 2003
US uniform and
Besides the standard client–
server component, Operations in-
cludes an online database to store
player records. Through a Web page,
the player creates an online account
on the authorization server data-
base. The game contacts the Auth-
Server at various times during play
to update information online. For
example, when a single-player train-
ing mission is completed, the client
uploads the score to the database.
When clients connect to an ofﬁcial
game server, the server contacts the
authorization server for veriﬁcation
of the client’s mission qualiﬁcations
and weapons training to make the
appropriate weapons available.
Online storage lets players download
their characters to any machine run-
ning the game (Figure 5).
Plans are underway to link Sol-
diers and Operations through the
database, so that players can unlock
new abilities within Operations by
completing tasks within Soldiers.
Operations runs on a PC under
Microsoft Windows 98, ME, XP, or
2000, with a preferred minimum of
a Pentium 3 processor running at
700 MHz, 128 Mbytes of RAM, 1.5
Gbytes of disk space, and a 56-Kbyte
modem or digital subscriber line
Because of its complex scenes and
characters, Operations requires a
graphics card with hardware sup-
port for transformation and lighting,
such as Nvidia’s GeForce 2 chip set.
No peripherals except a keyboard
and mouse are required.
While the combat training and mis-
sions on which Operations focuses are
critical, they represent only a small
part of army life and opportunities.
The Soldiers development team
needed to present an unvarnished view of a wide
range of army career fields. The rest of the army that
Soldiers depicts includes basic training (drill instruc-
tors and all), advanced specialty training, on- and off-
duty life, and enjoyment of the facilities available to
army personnel and their families.
We did everything as accurately as possible, not only
in portraying what army life is all about (Figure 6), but
in the appearance of bases, ofﬁces, barracks, and facil-
ities. To complement Operations and depict army loca-
tions with maximal realism, we employed digital
imagery as the display medium.
Home-grown story engine
Unlike Operations, for which Moves was able to buy
the latest commercial game engine, no existing engine
satisﬁed the peculiar requirements of Soldiers. We had
to build and design the entire system from scratch. Sol-
diers consists of four main components: a story engine,
location generator, animation engine, and text-to-voice
Underlying Soldiers is a story engine that constructs
and sequences scenes to create interactive storytelling.
The player creates a character with a set of personality
traits (based on the Army’s seven core values) and
IEEE Computer Graphics and Applications 31
daybreak in the
guides it through a career, managing its resources and
deciding its fate (Figure 7).
The player wins his character’s trust through good
decisions that give him increasing access to the charac-
ter’s interior, where he can tweak its goals and values
(Figure 8). These manipulations propel the character
through a career in a number of specialties. Along the
way, the player gets a glimpse of what army life offers
Under the hood
For the past three years, the Moves Institute’s com-
puter-generated autonomy group has been exploring
multiagent system simulation architectures that make
the development of complex, adaptive behavior easier
to achieve and control. Our research has produced four
agent-based simulation design concepts for modeling
multiagent systems and implementing the models in
software simulations: composite agents, goal manage-
ment, tickets, and connectors.
These concepts evolved under the
guidance of John Hiles and through
many hours of discussion between
principal investigators Michael Van-
Putte and Brian Osborn. A novel
simulation methodology emerged,
capable of generating dynamic
plans and interactive stories. Called
connector-based multiagent simu-
lation (CMAS), the architecture
serves as the underlying model for
the story engine.
Soldiers’ story engine is actually a
general-purpose simulation engine
that generates dynamic plans. Sto-
ries evolve as a byproduct of observ-
ing constraints previously defined
on agents within the simulation.
In other words, the stories are
plots generated through discovery
rather than fixed plans defined
beforehand. The engine produces
them through a simulation process
called connecting, whereby agents
are bound together according to a
best-fit axiom. The result of a suc-
cessful connection is the next step in
the plan (or, to put it another way,
the next visualization of a story
scene). By this means, dynamic
plans—or stories—evolve as the
Figure 9 provides a diagram of
how the system works. The scene-
rendering subsystem of the CMAS
takes streams of scenes produced by
the story engine and sends them to
a desktop computer after adding
multimedia (graphics, video, and
speech) to each scene as it generates
(the scene-rendering subsystem is
). The story
engine takes domain-deﬁnition data as its input (con-
sisting of descriptions of the characters, goals, proce-
dures, and objects in the simulation) and outputs a
stream of scenes. The user can update the domain-
deﬁnition data via the graphical interface.
In a CMAS simulation, agents communicate, and
thereby advance the simulation through time, via con-
nections. Agents extend connectors to signal their readi-
ness to participate, and retract them when they aren’t
ready. Part of the story engine’s job is to seek agents with
extended connectors and match them.
Internal behavioral procedures (called tickets) and
the current state of the environment dictate whether an
agent’s connectors are extended or retracted. Tickets
are input to the story engine via the domain-deﬁnition
data set. They specify the actions of individual agents,
which in turn are dictated by the phenomenon being
modeled. For example, if a predator were being mod-
eled as an agent, then its hungry connector would be
extended when the predator was in the hungry state.
32 January/February 2003
screen used to
pick a character
The predator would seek a connec-
tion with agents that are edible—
that is, prey. The story engine
manages the task of matching
predator to prey and updating the
internal states of both. In this way,
it remains independent of the par-
ticular story underway. Then the
input data, not the simulator, drives
Think of a CMAS simulation as a
collection of agents that evolve their
states, and thereby the state of the
entire simulation, by making and
breaking connections. The state of an
agent is updated only when connect-
ed to one or more others. Thus the
story engine is a simulation engine
that manages the evolution of agents
over time by matching them with one
another, combining them in connec-
tions, updating their internal-state variables, and dis-
connecting. A stream of scenes is the byproduct of the
repeated connect-update-disconnect process.
Soldiers’ story engine uses tickets and connectors
extensively to generate interactive, dynamic stories.
Because the intention of Soldiers isn’t to tell a predeter-
mined story, but to cobble a credible plot on the ﬂy, a
typical story consists of goal-driven autonomous char-
acters (such as a protagonist, drill instructor, buddy, and
supporting characters), a narrative structure closely
aligned with the main character, and a collection of
potential scenes. Of course, we also use media, dialogue,
and interactions to populate the scenes. Combined
dynamically at runtime, these elements produce story-
lines that reﬂect the characters’ actions, relationships,
Two primary classes of agents exist in the story
engine: characters and scenes. Through a process of con-
necting, the main character binds to a scene, and the
scene in turn binds to resources such as supporting char-
acters and media elements. Once resource requirements
have been met, the scene executes based on the goals,
values, and resources of the characters.
During scene execution, the story engine interfaces
with the location generator, animation engine, and text-
to-voice system to manufacture and present a scene
onscreen. As the scene unrolls, the game updates each
character’s internal state. For instance, in a training
scene a character’s skill may increase and capabilities
grow. When the scene is over, the process begins anew
with the main character binding to another logically
With scenes dynamically constructed at runtime, the
story engine generates a plot adapted to the user’s inter-
ventions. As the story plays out, the player is free to
adjust his character’s goals and values. The result is a
story personalized to the player’s will.
It goes without saying that we had signiﬁcant obsta-
cles to overcome to deliver a compelling product on CD.
For both Operations and Soldiers, we needed to show
real army posts through digital imagery, present a
breadth of career ﬁelds, and reference the Army’s depth
of training and resources.
Many of these challenges centered around squeezing
enough media on a CD to support the hundreds of thou-
sands of scenes the story engine was capable of gener-
ating. For example, each scene is played on a 640 × 480
background or location. At approximately 960 Kbytes
per background (uncompressed), it was impossible to
store enough data on a 650-Kbyte CD to accommodate
the story engine’s range of scenes.
The answer was to use recombinable pieces. We
made every bit of imagery, animation frame, and sound
The location generator constructs backgrounds
appropriate to a given scene and character by manag-
ing a database of chopped-up media pieces with which
to construct locations at runtime. A typical location con-
sists of 6 to 10 media pieces, including a sky, midground,
foreground, and context-speciﬁc props. If we station the
character at Fort Bragg, for example, a scene on the
parade ﬁeld will depict Fort Bragg buildings. The same
scene played out at Fort Jackson would show the build-
The reusability principle also applies to animation of
character actions. We constructed an animation engine
to build photo animations from individual frames of dig-
ital imagery. Army soldiers enacting a range of motions
and gestures were ﬁrst ﬁlmed against a blue screen. We
processed the raw media and a portion used to construct
a database of animation frames.
Just as a cartoon animator flips through a set of
images to produce movement, the animation engine
selects a set of prescribed frames from the database and
plays them sequentially to create animation. Each char-
acter’s database takes about 40 Mbytes of space.
Once we establish and catalog the frame set, howev-
er, the cost of adding animations built from the data-
IEEE Computer Graphics and Applications 33
Scene rendering subsystem
the story engine
in an interactive
base is minute. These 40 Mbytes of individual frames
can produce hundreds of onscreen actions.
Creating a spoken dialogue system to match the story
engine’s staggering diversity presented a daunting task.
Given the photorealistic backgrounds and animations,
synthetic vocalization was out of the question. Instead,
Moves constructed dialogue on a basis similar to the
technique used in constructing locations.
In locations, a template guides the selection of
images comprising the background. In our text-to-
voice system, we used a formal grammar to define
sentence structures, including phrases that vary not
only according to the needs of the script but also to
the speaker’s identity. For example, a morning
greeting by a sergeant and a buddy should differ.
We recorded actors speaking full sentences, then
deconstructed the sentences into elemental fragments
and collected them in a database. The text-to-voice sys-
tem’s sentence generator parses sentence definitions
and constructs sentences from the database.
While our solution isn’t as ﬂexible as a text-to-voice
program, we felt the gain in realism was worth the
With the help of the Army Research Institute, we
looked into whether a game player’s aptitude for an
army career could be computed. The work from ARI
34 January/February 2003
E3 and the Fourth of July Release
Developing the America’s Army suite silently for two years
from ﬁrst funding (May 2000 to product announcement at
the 22 May 2002 E3 conference) was difﬁcult for our large
team of collaborators, but we managed to arrive at E3 with
the story mostly under wraps. Our ﬁrst major press
exposure was on the front page of the Los Angeles Times,
replete with a color shot from the game.
The suite won
several press awards at E3 (http://www.movesinstitute.
The ofﬁcial America’s Army Web site, http://www.
AmericasArmy.com, contained little more than game
images and a location where visitors could register for an
email notiﬁcation of the release date. By 24 May, the Web
site was receiving 180,000 unique visitors per hour, with
18,000 pages served every ﬁve seconds.
Our booth at E3 had giant display screens high above the
convention ﬂoor. Every two hours an army bugler called in
an armed company of men simulating an air insertion,
including soldiers scrambling down ropes hung from the
ceiling of the Los Angeles Convention Center (Figure A).
We were pleased with the outcome of the conference,
but when it came to posting America’s Army: Operations
onto the Internet, army program managers could only
guess how many servers they’d need on 4 July, the ofﬁcial
launch date. The Army budgeted for 140 servers to seed the
On 4 July at 12:01 am, the ﬁrst 10 levels of America’s
Army: Operations were posted to the Internet, and by noon
the next day, 500,000 downloads of the 211-Mbyte game
had been made. The Army’s 140 servers were swamped,
and we rushed to complete and post the community server
kit the same week.
By 15 July, we were seeing approximately 1,900 servers,
each capable of serving 26
players, for a total potential of
around 49,400 players. By 30
August, Operations was
downloaded 2.5 million times.
Capture the record
Gratifying statistics roll in
daily. Game use as of
16 November 2002 saw
1,007,000 registered accounts,
614,000 graduates of BCT, and
more than 32 million missions
completed (averaging 6 to 10
minutes). Missions per day
average 338,380, with players
typically accomplishing 21
missions after BCT.
Assuming 10 minutes per
mission, we calculate gamers
racked up a combined 263
years of nonstop play in the ﬁrst
58 days alone (one avid player
enacted over 3,600 missions
during this period).
To put it another way, if these
hours were payable at
A Rappellers descend from the ceiling of the America’s Army booth at the E3 Convention,
looks promising and may appear in a later version of the
game. Meanwhile, the conclusion is that the Army won’t
receive aptitude data unless the player willingly for-
wards it to the Army.
If the player requests information about an army
career, we’ll ask if we can forward the player’s scoring
information to a recruiter. If the player approves, the
game forwards the information. If not, it goes
nowhere—cookies aren’t used in this process.
To see how the release of the software turned out,
read the sidebar “E3 and the Fourth of July Release.”
Having a successful online game inside the Moves
Institute is kind of like having your own particle accel-
erator. Lots of proposed applications and interesting
research are coming in the door.
Many related training applications using the Ameri-
ca’s Army code base as a starting point are being con-
sidered. We have funding from one project that’s using
Operations for treaty veriﬁcation preplanning, and an
Air Force group is looking at funding a training level
within the game that will deal with force protection.
Infantry soldiers at Fort Benning are using Operations
in their training regimen to gain familiarity with range
procedures before setting foot on the real range. Also,
the Army’s Objective Force is looking at integrating pro-
totypes of their new weapons systems into Operations
to evaluate their potential utility.
One extraordinary possibility, raised by the Under
Secretary of Defense’s ofﬁce, is massively multiplayer
(MMP) gaming, and the America’s Army project is being
looked at both as a model of how such a development
effort could be carried out within government and as a
possible starting point for an MMP project. Work
involved might include the procurement (or develop-
ment) of a government-owned game engine capable of
full-spectrum combat modeling and large-scale inter-
operability integration, as well as a programming inter-
face for modeling human and organizational behaviors
and stories à la Soldiers. An additional goal would be a
rapid prototyping interface to the MMP that would allow
any mission to be put together nearly overnight. ■
This work would have been impossible without the
support of our sponsor, the Ofﬁce of the Deputy Assis-
tant Secretary of the Army for Manpower and Reserve
We salute the excellent work of program manager Lieu-
tenant Colonel Casey Wardynski, as well as Major Chris
Chambers, of the Ofﬁce of Economic and Manpower
Assessment. The full development team of America’s
Army appears on the Web at http://movesinstitute.org/
1. S. Singhal and M. Zyda, Networked Virtual Environments—
Design and Implementation, ACM Press, 1999.
2. J. Hiles et al., Innovations in Computer Generated Autono-
my, Naval Postgraduate School technical report NPS-MV-
02-002, Monterey, Calif., 2002.
3. N. Elzenga, “The Recruits—Media Architecture System
Description,” Army Game Project (working document),
Moves Institute, 2001.
4. B.A. Osborn, An Agent-Based Architecture for Generating
Interactive Stories, PhD thesis, Dept. of Computer Science,
Naval Postgraduate School, 2002.
Michael Zyda is the director of the
Moves Institute at the Naval Post-
graduate School (NPS), Monterey,
California. His research interests
include computer graphics, large-
scale, networked 3D virtual environ-
ments, agent-based simulation,
modeling human and organizational behavior, interac-
tive computer-generated stories, computer-generated char-
acters, video production, entertainment/defense
collaboration, and modeling and simulation. He’s the prin-
IEEE Computer Graphics and Applications 35
minimum wage ($6.75 an hour), the bill would
hit $15,590,367 for 58 days. And if we project
the 4.6 years of play per day to 1,679 years of
play per annuum, we’re looking at $99,279,270
of intensive effort donated gratis by America’s
Though the Army has used http://www.
goarmy.com as a recruitment site for years,
trafﬁc is way up—about 28 percent of the hits
now originate from the game.
Mass production and distribution of the
America’s Army: Operations CD began
1 September. In midautumn, army recruiters
received CDs directly from the factory, in initial
quantities of 50 units apiece (for a total of
300,000 to all stations).
Approximately 100,000 CDs are slated for
army events and 100,000 for fulﬁllment by mail.
A million disks will be inserted in popular
gaming magazines. In addition, Operations will
be featured at this year’s college immersion tour,
available for a test drive via eight interactive
kiosks in the tour’s 53-foot trailer. To
complement the exhibit’s parachute simulators,
Operations will be conﬁgured to highlight its
single-player Airborne School missions.
1. A. Pham, “Army’s New Message to Young Recruits:
Uncle ‘Sim’ Wants You,” Los Angeles Times, 22 May
2002, p. 1.
cipal investigator of the America’s Army PC game. He
received his BA in bioengineering from the University of
California, San Diego, La Jolla, his MS in computer sci-
ence from the University of Massachusetts, Amherst, and
his DSc in computer science from Washington University,
John Hiles is the technical director
for computer-generated autonomy at
the Moves Institute, a research pro-
fessor at NPS, and program manag-
er for the America’s Army suite. His
main work is in using agent-based
simulation and entertainment tech-
nology to model human and organizational behavior. He
has a BA in creative writing from the University of Cali-
fornia, Santa Barbara.
Alex Mayberry is creative director
for the Moves Institute and the exec-
utive producer of America’s Army:
Casey Wardynski is the origina-
tor and director of the America’s
Army game suite, and director of the
Army Ofﬁce of Economic and Man-
power Analysis at the US Military
Academy, where he is an associate
professor of economics. He has an MS
in public policy from Harvard and a PhD in policy analy-
sis from Rand Graduate School.
Michael Capps is a producer with
Epic Games in Raleigh, North Car-
olina. He is a former research assis-
tant professor at NPS, as well as
former producer of the America’s
Army: Operations game. He received
a BS degree in mathematical sciences
and an MS in computer science from the University of
North Carolina, Chapel Hill, an MS in electrical engineer-
ing and computer science from MIT, and a PhD in com-
puter science from the Naval Postgraduate School.
Brian Osborn is an active duty
naval aviator and researcher in the
Moves Institute. He’s the principle
implementor of the Moves Story
Engine. His concentrations include
agent-based modeling and simula-
tion, computer-generated autono-
mous behavior, and interactive narrative. He has an MS
in operations research and a PhD in computer science from
Russell Shilling is the technical
director for immersive technologies
at the Moves Institute, an associate
professor at NPS, and the lead audio
scientist and designer for the Ameri-
ca’s Army: Operations game. He
received his PhD in experimental psy-
chology (neuroscience) from the University of North Car-
Martin Robaszewski is a pro-
grammer on the America’s Army:
Operations video game. His main
areas on the project are user inter-
face, networking, and mission-script-
ing. He’s primarily self-taught and
has worked on numerous platforms
using many different programming languages.
Margaret Davis is the writer and
Webmaster for the Moves Institute.
She has an MA in British literature
from California Polytechnic State
University, San Luis Obispo.
Readers may contact Michael Zyda, the director of the
Moves Institute, at firstname.lastname@example.org.
For further information on this or any other computing
topic, please visit our Digital Library at http://computer.
36 January/February 2003