Content uploaded by Albrecht Schmidt
Author content
All content in this area was uploaded by Albrecht Schmidt
Content may be subject to copyright.
WorldCupinion: Experiences with an Android App for Real-
Time Opinion Sharing during World Cup Soccer Games
Michael Rohs, Sven Kratz, Robert Schleicher
Deutsche Telekom Laboratories, TU Berlin
Alireza Sahami, Albrecht Schmidt
University of Duisburg-Essen
ABSTRACT
Mobile devices are increasingly used in social networking
applications. So far, there is little work on real-time emotion
and opinion sharing in large loosely-coupled user
communities. We present an Android app for giving real-
time feedback during soccer games and to create ad hoc fan
groups. We discuss our experiences with deploying this app
over four weeks during 2010 soccer world cup. We highlight
challenges and opportunities we faced and give
recommendations for future work in this area.
Author Keywords Mobile applications, mobile social
networking, opinion sharing, real-time feedback, sports.
ACM Classification Keywords H5.2 [Information
Interfaces and Presentation]: User Interfaces.
General Terms Design, Experimentation, Human Factors
INTRODUCTION
Mobile devices are increasingly used for mobile social
networking. One explanation for this development is that
mobile devices are almost always with their users, have
continuous wireless connectivity, and feature increasingly
capable user interfaces. They can thus serve as ubiquitous
input devices and sensors for user reactions, emotional
responses, and opinions around large public events [1].
The goal of the work presented here is to investigate mobile
social software for large user communities. The project
focuses on usage patterns and usability questions related to
interfaces that can scale to an extremely large number of
simultaneous users. We picked the soccer world cup 2010
as a use case for this research, because it is an event with
extremely high public attention in many parts of the world
and many people have a high emotional involvement to (at
least some of) the matches. The matches are also
synchronized in time with many simultaneous viewers and
thus many potential users. We focus on exchanging
spontaneous emotional feedback between users who are
part of a virtual fan block.
The particular test application, World Cupinion, is an
Android application that lets soccer fans express their
emotions about events and moments in soccer matches
during the matches. Through this application users can
support their favorite teams and share their opinions with
other fans. The design is focused on scalability to a large
number of users, on simplicity, as we expected that the
users’ focus of attention is on match itself, and on short
bursts of usage, with the application serving as an ambient
display most of the time and short interactions occurring
when interesting events happen.
Acquiring, managing, and distributing user data in large
user communities in real-time is a challenge that systems
and user interfaces need to address. In the long run, we
envision applications of this type being used by hundreds or
thousands of users simultaneously, loosely linked by
temporary shared interests and preferences, thus forming
ad-hoc virtual communities.
This work addresses the following aspects and research
questions:
• How to design mobile social networking applications
supporting the connectedness of peers in a large loosely
coupled community?
• How can experiences effectively be shared in real-time
across a large number of devices?
• How can we design for awareness of community
opinion? How can information related to shared
experiences be visualized?
• How can user attention be managed in cross-media
applications that combine television or public events with
mobile device usage?
In the following sections we discuss the concept of real-
time emotion sharing, give an overview of the design and
system architecture of our test application, discuss the
distribution and publication channels for the application,
and report on the experiences we made with the public
prototype. We conclude with recommendations for research
in the large and with giving ideas for future work.
Real-Time Emotion Sharing
Taking a closer look at what information the
audience/watchers of sport broadcasting actually wish to
share with their friends or fan group, it turns out to be
mostly the preliminary evaluation mixed with the personal
emotional impact of specific events during the game, much
less a “cold” rational assessment of the ongoing maneuvers
on the field. This is no surprise, as emotions are known to
have a strong social component [4] and probably even
developed to provide a fast and immediate way to
communicate the momentary state of an organism to the
environment [3].
The sudden onset and strong expressive component of
emotions make them an ideal candidate for mobile
communication as it allows the user to somehow extend
his/her reach beyond the usual radius of face-to-face
communication. On the other hand, these properties also
impose a number of requirements for any application:
feedback should be quick and if possible “analogous,” i.e.
non-verbal to avoid the necessity of lengthy formulation to
describe a simple and transient affective rush. Emoticons
appear to be an appropriate way to communicate these
states [2]. In addition, the provided rating scheme should
contain domain-specific labels (e.g. “yellow card”) as well
as domain-independent features (e.g. “like-dislike”) [5].
APP DESIGN
As mentioned above, the design criteria included simplicity,
since the user’s focus of attention is on the match itself, and
short-term usage, since situations arise quickly an
interaction might just involve stating one’s opinion about
the current event. Moreover one aspect was to visualize the
aggregated opinion of a potentially large number of users
and to use the screen as an ambient display. The latter
feature gives the user the opportunity to observe how the
fan aggregated opinion evolves even though he or she is not
actively interacting and probably react to it by rating again.
The app is structured in three screens (Figures 1 and 2). The
first screen (Figure 1, left) shows the list of upcoming
matches with their starting times and dates in the user’s
local timezone. The timezone played an important role here,
since we intended to deploy the app in the Android Market
for worldwide distribution. The game selection could have
been automized, except for parallel games during the first
phase of the tournament, but we decided to keep the list in
order to allow users to plan their viewing times in advance
of the games.
After selecting a game the user would enter the “arena” for
that game (Figure 1, right). That screen allows the user to
give feedback during the game and to see the aggregated
opinions of the fans of the own and the other team. Initially
the rating buttons are disabled and the user has to select the
team he or she wants to support in order to activate the
interface. This design decision means that users have to be
fan of a particular team in order to provide input. The input
buttons cover most of the display area to be easy and quick
to press. Below each button there is a horizontal bar that
indicates the average opinion regarding that input category.
For example, if the bar below the “thumbs up” button is
half filled that means that 50% of the fans have pressed that
button during the last 30s.
Figure 1. WorldCupinion screens: Initial screen (left) showing
the match list and main screen (right) used during a game.
A feature that was added one week after publishing the
game is to see the aggregated opinions of the fans of the
other team as well: The statistics of the own fans are shown
in green, the statistics of the other team in blue, the blue bar
is located behind the green one, and the green screen is not
fully opaque to see the blue one behind it.
Another change to the main screen (Figure 1, right) was the
replacement of an “offside” icon with a “yippee!” icon
(Figure 1, right, center icon). We made this change after we
realized that the “offside” icon was rarely used. On the
other hand the app lacked a way to express strongly positive
emotions, e.g. when the own team scored a goal. Adding
the “yippee!” icon provided a way to express that kind of
emotional feedback [6].
Figure 2. Geographic distribution of fan opinions.
The third screen, “world opinion” (Figure 2), shows the
geographical distribution of fan opinions of both teams on
the map. The underlying idea is that this visualization
shows geographical clusters of users having opposing
opinions. The map view is based on the standard Google
Holland
Spain
23 fans
34 fans
Maps APIs with icon overlays for the feedback that was
given at a particular location. Using Google Maps APIs
allows interactive panning and zooming of the map.
However, we restricted the maximum zoom level for
privacy reasons.
SYSTEM ARCHITECTURE
World Cupinion is implemented as a client-server
architecture. The World Cupinion mobile application sends
two basic request types to the server, update requests and
input requests. Update requests are used to poll the state of
the mobile application’s user interface, and input requests
are used to send user opinions to the server, as soon as an
opinion button has been pressed. The map view sends a
further request type, to which the server generates a
response containing the user inputs of the last 5 minutes.
The server logs all inputs to a SQLite database and
maintains statistics of the user opinions received in the last
30s. These 30s statistics are sent to the mobile clients in
response to update requests.
We initially used UDP datagrams for communication, as
our protocol does not require an active connection. UDP
also imposes a lower load on the server, which is beneficial
if there are many simultaneous server requests. However, it
soon appeared that certain network firewalls and also
mobile network providers may block UDP packets that
have non-standard destination ports. To remedy this, our
mobile application has a fallback mechanism that
automatically switches to HTTP requests if UDP
communication is unsuccessful. User input events are
always sent via HTTP to ensure that they do not get lost.
Supporting HTTP requests has the further advantage of
enabling the implementation of platform-independent web
interfaces. Although we did not originally plan to use a web
interface, we implemented one for evaluation purposes. It
turned out that this was useful, as our statistics show that
about 13% of opinions originated from the web interface.
A further important issue of the application is energy
consumption. Over the 90 minutes of a game (plus the 15
minute break and an optional 30 minute extension), the
application continuously communicates with the server via
the mobile phone network or via WiFi. There is a tradeoff
between the update rate of the interface and energy
consumption. In pilot tests we found that one update every
3s is sufficient. A significant contribution to energy
consumption comes from continuously using the device as
an ambient display for the opinion state. Even if the user is
not interacting with the device the community opinion is
updated and shown. This is technically implemented with a
“wake lock” that prevents the display from switching off
completely. Usually, a NexusOne mobile phone that is fully
charged at the start of the game has a battery level of about
50% at the end of a game.
DISTRIBUTION AND PUBLICITY
Distribution via the Android Market
World Cupinion can be downloaded for free from the
Android Market. An advantage of using the Android
Market as a distribution platform over Apple’s iTunes
AppStore, for instance, is that published applications appear
almost instantly for download, and are not subject to a
lengthy reviewing process with the risk of rejection of the
application.
The ability to rapidly push new releases of the application
to the Android Market allowed us to publish weekly
updates containing bug fixes or new features during the
actual soccer world cup.
Public Relations
It of course does not suffice to simply release a new
application into the wild. Potential users need to be
informed of the application’s existence in order for them to
download it.
We used a number of channels to make the application
known to potential users. In addition to press releases made
by the Deutsche Telekom Laboratories and the TU Berlin,
we tried to promote the application in internal events
(summer party, weekly lab meeting) and external events
(lab open house) of the Deutsche Telekom Laboratories. At
these events we distributed flyers advertising the app,
containing a QR code linking to the app on the Android
Market. We also created a website (www.worldcupinion.
com), and actively used social media (Twitter and
Facebook) and forum entries to reach as large an audience
as possible. Finally, we sent it to a number of mailing lists
in our lectures, posted a message about it on an Android
developer forum.
Considering that we only released our application one week
before the soccer world cup, our publicity measures were
relatively successful. At the end of the World Cup, we had
registered a total of 1645 downloads and 448 “active”
installations (=29% of all downloads). The number of
active installations denotes the number of users that still
have the app installed on their devices.
We were surprised to what extent users apparently
download and install an application without actually using
it for a longer time. It appears that the abundance of
available mobile phone applications let mobile applications
become a disposable article like promotional gifts, a
tendency much less pronounced for traditional software.
Updates
The Android Market allows to easily publish updates. We
took advantage of this feature several times. If a long-term
study is conducted it allows to carry out several design
iterations while keeping the user base of the app. Besides
bug fixes, the changes and updates related to (1) removing
the “offside” icon and replacing it by the “yippee” icon, (2)
adding group-generated sounds, (3) showing the opinion of
the opposing fans, (4) sending notifications when a game
starts (and restarts after the break), and (5) the addition of
an in-application questionnaire displayed after the
conclusion of the World Cup. These changes were not
obvious from the start and reflected insights gained from
application usage and user comments. The update
mechanism provides a convenient way to do these changes.
On the other hand, one has to be careful not to confuse
users when features change. If some users do not update the
application there might be inconsistencies between
deployed application versions, e.g. some users might still
have the “offside” icon in the place where users of the
updated app have the “yippee” icon. We expected that most
users will update their app, and for us having the flexibility
to try out different versions was more important than
version consistency. Unfortunately, the server protocol did
not include the version number, so we could not track the
percentage of users connecting to the server with outdated
application versions. This is something we will clearly
consider in the future when deploying similar apps.
EXPERIENCES
The first people using the app during an actual game were
friends, colleagues and students who were made aware of it
by our public relations activity. Here the focus was in part
more some kind of 'beta testing' and providing feedback to
the developers than using it for rating per se. However, the
Vuvuzela functionality was quickly used to emphasize
important moments or echo the sound coming from the TV,
which then led to a more comprehensive usage of the app to
comment the ongoing game. We were well aware that there
were several other “vuvuzela apps” available, but still we
thought that this feature might serve as some kind of door
opener to attract the user's attention initially, and our
observations within our peer group confirmed that
assumption. Probably also other research projects might
benefit from this approach. However, there is one caveat
when adapting popular "recreational apps" or functionalities
for research applications: we found that users expect
running software and will give low ratings if an app is not
polished or crashes during usage. There were occasional
crashes of the map view in our app, which only affected the
map view and not the other parts of the application. Users
were very critical about this, as is documented by one user
comment mentioning this issue. Moreover, a few handset
types had problems. Even though platform fragmentation is
much less of a problem for Android than for other systems,
this issue appeared. Of course, providing an industry-
strength app as a research prototype is not feasible for most
research labs as it requires more development resources
than are typically available.
CONCLUSION
Evaluating essential usability aspects of mobile social
networking apps relies on a critical mass of users. We tried
to gain a large number of users for a mobile app for sharing
opinions in large user communities in real-time, by picking
a popular topic in which people have emotional
involvement and simultaneously follow a shared event. The
2010 soccer world cup provided an ideal setting for this
research, because it is quite popular in many parts of the
world and extends over four weeks, which allowed us to do
several design iterations. We tried to gain as much user
attention as possible by publishing the app on the Android
Market and by using a number of measures to make it
known to potential users. The market update facility
allowed us to very simply try out design modifications and
to improve the prototype based on usage statistics and user
feedback.
We have not yet finished to fully evaluate the database logs
and the in-application questionnaire yet. To give the reader
an impression of usage statistics, on average 28.6 (+/- 19.1)
users were active during all games. The average
participation lasted 681 (+/- 1316.2) seconds during which
17.6 (+/-33.6) actions, i.e. button presses were performed.
The large standard deviations already indicate that an
overall analysis might be of limited informative value and
characteristics of the particular game have to be taken into
account here, which we are currently working on.
As future work we intend a better integration with social
media platforms, such as Facebook and Twitter. We also
intend to do an analysis that could give us further insight
into the question to what extent user input correlates with
the actual soccer match. We hope that this analysis can also
tell us about typical patterns of behavior. Finally, we
envision other types of content, such as talk shows or
political debates, in which user responses to broadcast
content might be helpful.
REFERENCES
1. Diakopoulos, N. A. and Shamma, D. A. Characterizing
Debate Performance via Aggregated Twitter Sentiment.
In Proc. of CHI 2010. pp. 1195-1198, 2010.
2. Derks, D., Bos, A. E., & von Grumbkow, J. (2008).
Emoticons in computer-mediated communication: social
motives and social context. Cyberpsychology &
Behavior, 11(1), 99-101.
3. Ekman, P. (1999). Basic Emotions. In T. Dalgleish &
M. Power (Eds.), Handbook of Cognition and Emotion.
(pp. 45-60). Sussex, UK: John Wiley & Sons.
4. Ochsner, K. N., & Schacter, D. L. (2000). A Social
Cognitive Neuroscience Approach to Emotion and
Memory. In J. C. Borod (Ed.), The Neuropsychology of
Emotion (pp. 163-193). Oxford: Oxford University
Press.
5. Pang, B., & Lee, L. (2008). Opinion Mining and
Sentiment Analysis. Foundations and Trends in
Information Retrieval, 2(1-2), 1-135.
6. Schleicher, R. and Trösterer, S. The ‘Joy-of-Use’-
Button: Recording Pleasant Moments While Using a
PC. In Proc. of INTERACT 2009. Lecture Notes In
Computer Science, vol. 5727. Springer-Verlag, pp. 630-
633, 2009.
