Halo 3: The Theory and Practice
of a Research-Design Partnership
Microsoft Game Studios | firstname.lastname@example.org
Microsoft Game Studios | Randy.Pagulayan@microsoft.com
In the September-October issue of interactions, cre-
ative director for the Windows Core Innovation Team,
August de los Reyes, and I described an approach
to designing emotionally engaging products. The
approach is based on the James-Lange theory—a
pioneering theory of emotion that places physi-
cal activity as the source of emotions, rather than a
product of emotion. In this approach emotions are a
“readout” based on our activity and the context in
which it occurs. This has clear implications for what
user researchers focus on during the design of prod-
ucts, and speaks toward the relationship between
researchers and designers as they work together to
create compelling products.
In the article mentioned above, we pointed out
the parallels between this approach and the frame-
work of mechanics, dynamics, and aesthetics that is
used extensively in game design developed by Marc
Leblanc ( http://algorithmancy.8kindsoffun.com/).
In that approach designers control mechanics; the
behavior of players is considered dynamics and the
conclusions that players reach about the game are
aesthetics. Halo 3 is an example of the application of
this framework and the James-Lange theory.
Halo 3 is the third game in the Halo series; it is
a first-person shooter developed by Bungie Studios
for the Microsoft Xbox 360. The game, released in
late September 2007, holds the record for the highest
grossing opening day in entertainment history, mak-
ing $170 million in its first 24 hours. This achieve-
ment is even more striking when we consider that
most videogames lose money.
During the design of Halo 3, we were able to collect
and analyze large amounts of behavioral data and
monitor conclusions users reached about the game.
This combination of behavior and conclusions was
critical. Games designers are reaching for an aesthet-
ic experience—an emotional conclusion about the
game. But that aesthetic experience is based on how
users play the game. By synchronizing both behavior-
al and aesthetic measures we were able to provide the
design team key evidence of when their intent was
not realized. Without both of these measures design-
ers would not have been able to make fully informed
decisions regarding design changes.
Three key characteristics of the data reporting
1. Data reported very quickly. As many of you
realize, timeliness of data is a key to effectiveness.
Toward the end of any development project, hard
decisions must be made to get the product out the
door. No one can “wait for the data to come in.” So
for us, every second counts. There is little time to
spend doing a thorough analysis on thousands of
data points. In Halo 3, we needed to be able to col-
lect hundreds of hours of player time over a week-
end and turn around our recommendations within a
day or two.
2. Data reporting that speaks to designers.
Although we are the experts in our field and in data
analysis, we should stop assuming our partners can’t
handle looking at numbers. One of the key facets to
our approach was presenting the player experience
using numbers and charts directly to designers and
letting them do some exploration themselves. The
behavioral data we collected was analyzed and plot-
ted in terms that made sense to the designers, such
as the location of player deaths during a mission.
The reporting system supported easy, one-click drill-
down to deeper levels. This required investment up
front in terms of the research questions they wanted
answered, and our being able to build views of data
that were simple to understand and easy to identify
3. Data reporting that links quantitative and qualita-
tive data. Straightforward linking between quanti-
tative data (number of deaths) and qualitative data
reporting (video) was another critical factor. Starting