INTERNEWS CENTER FOR INNOVATION & LEARNING
innovation.internews.org
CROWDGLOBE
A Report form the Internews Center for Innovation and Learning
July 2012, Washington DC
CrowdGlobe.net
MAPPING THE MAPS
A Meta-Level Analysis of Ushahidi & Crowdmap
THE AUTHORS AND RESEARCH TEAM
Catie Bailard
is Assistant Professor of Media and Public Affairs in the
School of Media and Public Affairs at George Washington University. Before
joining the SMPA faculty in 2009, Catie received her doctorate in political
science from UCLA. She graduated with a 3.947 cumulative GPA with con-
centrations in American Politics, Formal and Quantitative Methods, and
International Relations. Throughout Catie’s academic career, her research
agenda has primarily focused on the intersection of politics and informa-
tion and communication technologies. This fascination with the effect of
media on political behaviors and outcomes began in college as a major in
UCLA’s Communication Studies Department, a top-ranked undergraduate
department, where she graduated
cum laude
. It was this experience that
inspired Catie’s decision to pursue a doctoral degree in political science
at UCLA, which provided her with a broad substantive understanding of
political science and political communication, as well as rigorous train-
ing in methodology. Catie’s primary research focus is the cross-national
analysis of the Internet’s influence on people’s evaluations and expecta-
tions of their governments, particularly focused on individuals’ satisfac-
tion with how democracy functions in their own nations. Other recent
work includes an analysis of the impact of mobile phones on corruption
in Africa (published in
Political Communication
).
Rob Baker
has over ten years of experience as a web and new media
developer, trainer, and manager as Project & Outreach Manager, respon-
sible for documentation, toolkits, and working with clients. Before offi-
cially joining Ushahidi, his contributions to their community earned him
the first ever induction to the Trusted Developer Network for his work as
technical or project lead on dozens of Ushahidi deployments from crisis
response to civic engagement around the world as well as the creation
of the Ushahidi Community website. In addition, he acted as Director of
the Universities for Ushahidi program, a 2011 initiative to train students
from around the world on mobile and mapping technology. Baker is also a
member of the Humanitarian OpenStreetMap Team, writes and develops
mobile-ready textbooks, and teaches online courses on ICT for humanitar-
ians through TechChange. Rob has coordinated international educational
programs and media projects in Africa, Haiti, and the Middle East, as well
as technical development of online projects within non-profit organiza-
tions. He has contributed to the technical development of several open-
source projects and codebases, while he has also produced, edited, and
shot short form video for the web with several feature humanitarian aid
pieces. Baker is a US Delegate of the U.S.-Russian Bilateral Presidential
Commission Subgroup on Mass Media. He is based in Washington, DC and
can also be found at @rrbaker.
Matt Hindman
is Associate Professor in the School of Media and Public
Affairs at George Washington University. His work focuses on political
communication with a concentration on Internet politics. Dr. Hindman’s
book
The Myth of Digital Democracy
, published in 2009 by Princeton
University Press, looks at the Internet’s impact on American politics.
The book won Harvard’s Goldsmith Book Prize as well as the Donald
McGannon Award for communication research. In the past two years, Dr.
Hindman has been given presentations or invited lectures at Harvard,
Yale, Princeton, Columbia, the University of Pennsylvania, Stanford,
and Oxford. The book has been referenced by members of the Federal
Communications Commission in public speeches, and featured on NPR’s
On the Media
. In addition to the book, Dr. Hindman has published on other
topics including online campaigning, “open source” politics, and the online
public sphere. His article “The Real Lessons of Howard Dean” was deemed
the best article of 2006 by the Information Technology and Politics sec-
tion of the American Political Science Association. Hindman has published
several op-eds in the
New York Times
on technology issues. Dr. Hindman
has previously been an assistant professor of political science at Arizona
State University. For the 2010-2011 academic year, he will be a nonresi-
dent Faculty Associate with the Berkman Center for Internet and Society
at Harvard. His next book will be on the political economy of the online
public sphere.
Steven Livingston
is Professor of Media and Public Affairs and Media and
International Affairs with joint appointments in the School of Media and
Public Affairs and the Elliott School of International Affairs. He holds a PhD
in political science from the University of Washington (1990). In addition
to his teaching and scholarship, he has held a variety of administrative
posts at GW, including stints as the director of the School of Media and
Public Affairs and director of the Political Communication Program. He
is also the founding director of the Institute for Public Diplomacy and
Global Community (formerly the Institute for Public Diplomacy). His cur-
rent research considers new forms of technologically enabled collective
action in pursuit of public goods where the state is incapable or unwill-
ing to meet these needs. His most recent book,
When the Press Fails
(University of Chicago Press), looks at media and public opinion dynam-
ics prior to and in the early stages of the Iraq war. His earlier work on
global real-time media coverage of war and conflict is the foundational
scholarly work in this area. His work on collective action, governance,
and information and communication technology has taken him to some
thirty countries in the last four years, including multiple trips to Iraq and
Afghanistan, as well as stays in several African countries, India, Malaysia,
and to South and Central America.
Patrick Meier (PhD)
is an internationally recognized thought leader on
the application of new technologies for crisis early warning and humani-
tarian response. He presently serves as Director of Social Innovation at
the Qatar Foundation’s Computing Research Institute (QCRI) where he
spearheads applied research in advanced computing to develop next-
generation humanitarian technologies. Prior to QCRI, Patrick co-founded
and co-directed the Harvard Humanitarian Initiative’s (HHI) Program on
Crisis Mapping & Early Warning and served as Director of Crisis Mapping
at Ushahidi. He also co-founded the CrisisMappers Network, Standby
Volunteer Task Force and the Digital Humanitarian Network. Patrick holds
a PhD from The Fletcher School, a Pre-Doctoral Fellowship from Stanford
and an MA from Columbia University. He blogs at iRevolution and tweets
at @patrickmeier.
ACKNOWLEDGMENTS
This report was a collaborate effort, drawing on the expertise of a wide
range of contributions from Internews, George Washington University
(GWU) and Ushahidi. Gregory Asmolov, a doctoral student at Media and
Communications Department, London School of Economics, first con-
ceptualized CrowdGlobe as an online research hub for crowdsourcing
deployments and we are grateful to him for inspiring its creation and his
continued contributions. The Internews Center for Innovation & Learning
catalyzed the project; GWU faculty carried out the analysis of the data
provided by Ushahidi. Nikki Usher (GWU), Amanda Noonan (Internews)
and Mark Frohardt (Internews) provided important guidance and feed-
back throughout the dra±ing process of this report. Many thanks to
Brian Herbert (Ushahidi) for sharing the Crowdmap data without which
this report would not exist. Finally but certainly not least, GWU Research
Assistants coded a considerable amount of data that made the empirical
analysis for this report possible.
CREDITS
Graphics: iHub Research
Design: Kirsten Ankers, CitrineSky Design
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
1
CONTENTS
Executive Summary .
...........................................................................................
3
Introduction .
........................................................................................................
5
Ushahidi & Crowdmap .
.....................................................................................................
6
Research Questions And Design .
....................................................................................
8
Quantitative Results & Analysis .
.................................................................................
10
Power Law Distribution .
................................................................................................
11
Survey Research .
............................................................................................................
14
Survey Results & Analysis .
...........................................................................................
15
Challenges .
.......................................................................................................................
17
Successes .
........................................................................................................................
17
Conclusions And Recommendations .
...............................................................
19
Case Studies .
.....................................................................................................
21
Case Study 1: Haiti .
...............................................................................................
21
Case Study 2: Libya .
..............................................................................................
23
Case Study 3: Japan .
.............................................................................................
24
Case Study 4: Sudan .
............................................................................................
25
Case Study 5: Egypt .
.............................................................................................
26
Appendix I – Dictionary For Crowdmap Data .
..................................................
28
Appendix Ii – Survey Research Questions .
......................................................
30
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
3
EXECUTIVE SUMMARY
For this, its first case study, CrowdGlobe has analyzed Ushahidi
and Crowdmap data as well as these platforms’ user base.
The Ushahidi platform, which means, “witness” in Swahili, is a
free and open source tool that integrates information collec±
tion features with a live map. Ushahidi, the company, subse±
quently launched Crowdmap, a hosted version of the Ushahidi
platform, which is easier to use since downloading the so²±
ware and installing it is not necessary. When the CrowdGlobe
research project was launched in October 2011, a total of
12,795 Crowdmaps had been created in over 100 countries.
This presented CrowdGlobe researchers with an ideal first use±
case for the project. The aim of this first report is to develop a
better understanding of how Crowdmap (and Ushahidi) have
been used and to analyze the data they have generated over
recent years.
Our work took advantage of statistical analysis, quantitative
content analysis and exploratory surveys. The quantitative
analysis revealed that 93% of the 12,000+ Crowdmaps ana±
lyzed had fewer than 10 reports while 61% of Crowdmaps
were identical to the default Crowdmap setting, i.e., they had
not been customized or used at all. This “long tail” distribu±
tion of Crowdmaps follows a power law distribution, a common
feature in many online platforms, as well as in a number of
occurring phenomena. Crowdmaps with 21 to 10,000 reports
were selected for further analysis, resulting in a data set of
585 maps. About 30% of these focused on North America while
18% focused on Western Europe and 16% on Africa. On aver±
age, these Crowdmaps had 814 reports but the median number
of reports for this set of deployments was substantially lower,
which is not surprising considering that Crowdmaps follow a
power law distribution.
When the analysis is broken down by region, the relative fre±
quency with which themes emerged in the regional deploy±
ments differed dramatically. For example, taking into account
recent events, it is not surprising that the most common
themes that emerged from the 63 deployments in the Middle
East and Northern Africa pertain to: crime and public safety
issues (43%), human rights abuses (40%), emergency±related
infrastructural issues (30%), and political organization (25%).
The distribution of themes in the 79 Western European deploy±
ments, on the other hand, paints a very different picture, with
entertainment and leisure appearing in 32% of the deploy±
ments, followed by non±emergency infrastructural issues
(25%), and media reports (23%).
It is also not surprising that
the 16 deployments from the Caribbean region, 12 of which
hailed from Haiti specifically, heavily featured issues related to
the occurrence and a²ermath of a natural disaster (63% and
50%, respectively), emergency±related infrastructural issues
(63%), health and medical±related issues (50%), and crime
and public safety issues (38%).
In addition, surveys were sent to all 12,795 Crowdmap users to
better understand how they used the platform and to assess
their experience. About 80% of respondents are men and
the average age of a user is 40 years old. As for educational
background, an impressive 43% of users have a post±gradu±
ate degree and a total of 84% have at least a college degree.
Approximately 53% of users responded as having no prior
experience in using online mapping technologies, which is not
surprising since the technology is still relatively new. The most
important reason why Crowdmap users used the Crowdmap
platform was to create a map with a specific purpose or event
(40%). About 63% of these users launched a map to cover an
event in the city in which they live.
The purpose of the CrowdGlobe project is to study various crowdsourced±mapping platforms, searching for
data patterns that can tell us more about the functions of these tools and their limits as well as potentials.
The
CrowdGlobe.net
website is an integral part of the CrowdGlobe project, providing researchers with addi±
tional case studies, meta±level datasets and analysis. CrowdGlobe is strictly platform agnostic and seeks to
analyze all crowdsourced mapping technologies.
1
4
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
About 64% of Crowdmap users created a map for non±direct
use, i.e., for training or demonstration purposes or simple curi±
osity. This in large part explains the power law distribution
reported above. That said, about 30% found the technology
counter±intuitive and too time consuming.
In terms of users who did launch a Crowdmap, about 16%
felt that they were not able to generate the required public
awareness vis±a±vis their map to make it as effective as they
had hoped.
In addition, around 19% noted they were able to
make their map almost as effective as they hoped. However,
31% replied that they simply were not able to garner sufficient
interest in their map. This finding also explains the power law
distribution described above. Of those users who responded to
the survey and felt they had been successful in raising aware±
ness of their maps, about 23% noted that they had done so via
engagement with community organizations, civic groups and/
or professional organizations. Around 20% said they had suc±
cessfully built public awareness by engaging with social media,
SMS, etc. Only 6% of users said they used traditional news
media to build awareness of their project.
A separate survey for semi±structured interviews was devel±
oped for users who launched high±profile projects using the
self±hosted Ushahidi platform. A total of 37 high±profile proj±
ects were identified for the survey and seven respondents
completed the survey, i.e., ~19% response rate. All seven
respondents represented formal organizations and had used
the platform in response to a complex humanitarian emer±
gency or “natural” disaster. Two deployments were in develop±
ing countries and the remainder in the “Global South”.
Some of the biggest challenges cited by users of the Ushahidi
platform included “keeping reports up to date, embedding pic±
tures and documents;” “the need to display the data on some±
thing other than a map;” “getting the word out quickly a²er
the launch of the site;” and “getting people to submit reports.”
When asked about failures, respondents’ answers ranged quite
widely, from persistent technical problems to mobilizing vol±
unteer involvement. Many noted that using the platform was
simply too time±consuming.
The findings from the quantitative analysis and surveys pro±
vide the first evidence±based analysis of crowdsourced data
of its kind. In addition, the results supply actionable feed±
back to Ushahidi so²ware developers on what they can do to
improve their platforms and substantially increase the number
of Crowdmaps that gain more traction and possibly greater
impact. It should be noted that since this research over half±
a±year ago Ushahidi Inc. has already been implementing a
number of important changes including a set±up wizard, a wiki
for Ushahidi users, and a review of the Crowdmap. In sum, this
report provides an important baseline study—and indeed the
only one of its kind—which could serve as an important com±
parison if this research is replicated in the coming years.
As we assess the growth and impact of Ushahidi in general and
crowdsourcing in particular we should keep in mind that we
are still at the very start of a transformative process. There is
plenty more to do and learn before we can draw any firm con±
clusions, particularly vis±à±vis impact. Crowdmap, for example,
is barely a year±and±a±half old, which means that users are still
very much in the pioneering and discovery phase. Recall Clay
Shirky’s point that “technology only becomes socially interest±
ing when it becomes technologically boring.” This explains why
the CrowdGlobe Project is intended to launch the means of an
ongoing assessment of where we are now and what we can
expect in the future—hence the interactive CrowdGlobe.net
portal. This is not the final statement about crowdsourcing and
Ushahidi. It is the opening statement of a new field of inquiry
and civic action.
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
5
INTRODUCTION
2
Crowdsourcing has also had a profound impact on the not±for±
profit sector. Wikipedia, of course, remains one of the most
astounding examples of crowdsourcing to date. Even the ven±
erable Encyclopedia Britannica announced that it would cease
publication in the face of crowdsourced information platforms,
especially Wikipedia.
Crowdsourcing is also radically reshaping
humanitarian response—a significant shi² that many humani±
tarian organizations have yet to realize or fully understand,
let alone respond to. While these humanitarian professionals
were, for many decades, most o²en confronted by an infor±
mation vacuum following a crisis, which meant that they were
tasked with providing initial assessments, they are now con±
fronted with a deluge of multi±media, user±generated content
shared on multiple social media channels, o²en in real±time.
Disaster±affected communities are increasingly becoming digi±
tal. Thanks to the incredibly rapid commercialization of mobile
phones worldwide, these communities have become the
source of “Big Data” generated during the immediate a²ermath
of a crisis.
Crowdsourcing is also disrupting the mainstream
media industry as ordinary citizens are increasingly digital and
this has catalyzed the global rise of citizen journalists. During
the Arab Spring, for example, well over two±thirds of the video
footage aired by Al±Jazeera was crowdsourced.
In sum, the majority of digital content shared online and via
mobile phones is now user±generated, rather than produced
by experts tied to formal institutions. This trend is not about
to decelerate any time soon. Quite the contrary, the amount
of user±generated, crowdsourced information will continue to
increase exponentially.
This massive shi² presents both significant challenges and
important opportunities. Yet rigorous, data±driven research
necessary to shed insights on this revolution in information
is lacking. The purpose of Internews’ new CrowdGlobe initia±
tive is to encourage and facilitate empirical research on the
nature and impact of crowdsourced data—particularly geo±
referenced data. CrowdGlobe is a platform agnostic, applied±
research program that seeks to identify trends in both the use
of crowd sourcing technologies and the data patterns gener±
ated by these new technologies. CrowdGlobe aims to produce
in±depth reports on these trends and patterns. The project is
accompanied by the interactive
CrowdGlobe.net
website, which
provides access to underlying datasets, case studies and further
analysis. In this way, the project seeks to catalyze additional
user±generated analysis of crowdsourcing trends and patterns.
In other words, the portal provides researchers with access to
the meta±level crowdsourced data they need to understand
both the opportunities and limitations created by new informa±
tion and communication technologies. Through Crowd Globe.
net, a global community of scholars and activists can now work
together to create best practices in the context of sound ana±
lytics. At a more theoretical level, CrowdGlobe offers an almost
unparalleled opportunity to investigate the dynamics of digital
information in political and policy processes.
For the inaugural launch of the CrowdGlobe project, Internews
elected to use Ushahidi as its first case study. This decision was
Crowdsourcing is changing entire industries across multiple sectors. Coined by Jeff Howe in 2006 to describe
businesses that were openly outsourcing small, incremental tasks to the general public, crowdsourcing has
since become a major business sector itself.
The multi±million dollar company Crowdflower, for example,
crowdsources millions of tasks a year on behalf of top Fortune 500 companies.
At a more theoretical level,
CrowdGlobe offers an almost
unparalleled opportunity to investigate
the dynamics of digital information in
political and policy processes.
6
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
based primarily on ease of access to Ushahidi’s crowdsourced
data—particularly the hosted Crowdmap data. In addition, at
the time this research began in October 2011, some 12,000+
Crowdmaps had been launched in over 100 countries, which
provided a size±able amount of data to analyze. Furthermore,
the Ushahidi platform has been used for multiple purposes
ranging from disaster response and human rights monitoring,
to citizen journalism and election observation. Moreover, the
Ushahidi team had long expressed a strong interest in deepen±
ing their understanding of how users were engaging their plat±
forms and what the data they generated looked like. In par±
ticular, the Ushahidi team was interested in learning from this
baseline study to both improve their technology and develop a
real±time dashboard based on the metric identified in the anal±
ysis. To this end, Ushahidi generously provided all the meta±
data used in this study and Internews partnered with faculty
from George Washington University’s (GWU) School of Media
and Public Affairs (SMPA) to carry out a fully independent and
rigorous analysis of this data.
It is worth repeating that the CrowdGlobe project itself is strictly
platform agnostic and seeks to provide extensive assessments
of all crowdsourced mapping platforms out there. Indeed, this
first case study is simply a starting point to stimulate input
from the wider crowdsourcing community and in the process
identify additional case studies for future reports.
Ushahidi & Crowdmap
In 2008 when several blog±
gers and so²ware devel±
opers responded to Ory
Okolloh’s call for some
means to aggregate and
share the many reports of violence that were coming to her
and other bloggers in Kenya, the world was less than a decade
into the use of several establishing technologies for Ushahidi.
Without them, Ushahidi would have been impossible.
Less than a decade before, for example, the remote sensing
satellite industry had just gotten off the ground–literally–with
the launch of Ikonos in 1999.
It was the world’s first high±reso±
lution remote sensing satellite. It and a fleet of other satellites
that followed helped create the highly detailed geographical
information archives that makes open source digital mapping
possible.
Without georectified spatial data, that is, without
data about a precise spatial reference point on a map, Ushahidi
would not be possible.
With georectified data, we are able to
pinpoint locations and geotag them in relation to events that
are significant in the context of a particular Crowdmap deploy±
ment. Put another way, those little red dots would be mean±
ingless and impossible without the relatively new capacity to
use open±source geographical information systems that rest
on remote sensing imagery. That system was eight years old
when Ushahidi was first developed and Google Maps was only
three years old.
Secondly, the remarkable growth of mobile telephony in the
global south (and north) empowered publics (everyone with
access to a handheld device) to be a potential part of a tech±
nologically enabled network that is global in scalability. In
2000, only 2 percent of the population of Africa had access
to a mobile phone; in 2009, 28 percent did. By the year 2015,
Sub±Saharan Africa will have more people with mobile network
access than with access to electricity at home, some 138 mil±
lion people. And by 2020 there will be at least one SIM card for
every person on the continent. This same pattern is found all
over the global south.
In Kenya the post±election violence of 2008 resulted in more
than a thousand deaths and half±a±million displaced individu±
als. The Kenyan government tried to downplay the severity of
the situation. At the same time, because journalists could not
be everywhere and many human rights violations went unre±
ported. Some Kenyan activists therefore decided to crowd±
source and live map the crisis reporting.
They set up a website
with a Google Map of Kenya coupled with a web form and an
SMS number. This allowed anyone with access to the Internet
and/or a mobile phone to send in eyewitness accounts of
human rights abuses. In this way, the “crowd” was able to doc±
ument human rights abuses that would have otherwise gone
undocumented.
The project was called Ushahidi, which in Swahili means, “wit±
ness” or “testimony”. The Ushahidi platform is most simply
described as a multimedia inbox connected to a live map.
The inbox receives data from webforms, as well as emails,
SMS, tweets, pictures, video footage, voicemails, etc. Users of
Ushahidi can use various methodologies to collect this infor±
mation—e.g., crowdsourcing—which they can subsequently
By the year 2015, Sub±Saharan
Africa will have more people with
mobile network access than with
access to electricity at home,
some 138 million people.
See the Libya Case Study
(
page 23
) or view online
at
http://crowdglobe.net/
deployments
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
7
map on a public website. The activists behind the Ushahidi
subsequently launched a non±profit organization of the same
name to provide the underlying technology as free and open
source platform for others to customize and use for their own
purposes. Since then, the platform has gone through several
revisions and integrated additional technologies/media like
smart phone apps and Facebook. In sum, the Ushahidi platform
facilitates live, multimedia and collaborative mapping. Some
20,000 Ushahidi maps have been deployed in over 140 coun±
tries since the original Kenya Crisis Map in 2008.
In the Fall of 2010, the Ushahidi team launched Crowdmap,
basically a hosted version of the Ushahidi platform, which until
then could only be used by downloading and installing the
so²ware on a computer. This o²en proved to be a challeng±
ing process, particularly for non±tech savvy users with little to
no programming skills. Crowdmap, on the other hand, is like
“Google Doc”; no downloading or installing required. The launch
of Crowdmap has considerably lowered the barrier to entry
FIGURE 1:
THE ORIGINAL USHAHIDI PLATFORM LAUNCHED IN 2008
FIGURE 2:
OF THE 20,000+ MAPS DEPLOYED
SINCE 2008, OVER 15,000 ARE USING THE
HOSTED CROWDMAP PLATFORM
8
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
for users interested in crowdsourced mapping. Indeed, of the
20,000+ maps deployed since 2008, over 15,000 are using the
hosted Crowdmap platform.
The Ushahidi and Crowdmap platforms are not the only map±
ping technologies available for collecting and mapping crowd±
sourced data. The free and open±source OpenStreetMap
platform, for example, is primarily known as a crowdsourcing
project for street and road data around the world. However,
the platform has also been used to map crowdsourced data
related to other types of infrastructure such as refugee camps
in Haiti, health facilities in Libya, damaged buildings in Turkey
and disaster preparedness data in Indonesia. Google Maps has
also been used to map crowdsourced data, as has the Google
Map Maker platform, most notably in Southern Sudan. To date,
however, the Ushahidi platform remains the only technology
that is geared towards crowdsourced event mapping.
Research Questions And Design
When the Crowdglobe research project began in October 2011,
12,795 Crowdmaps had been launched by individuals and
organizations around the world.
Together with Internews, the
George Washington University team and the authors of this
report identified metrics that would provide insights on how
users were using Crowdmap. They also identified metrics that
could capture potential patterns generated by the Crowdmap
meta±data. These metrics were broken down into three tiers.
The first tier included quantitative metrics that could be
answered through a series of statistical queries. This included
the topic/theme of the deployment; the number of reports per
deployment; the number and type of categories per deploy±
ment; the number of users per deployment, etc.
The second tier required additional analytical work, drilling
down into the context of the instances, which meant survey±
based research. For example, what kinds of background do
FIGURE 3:
INFO-GRAPHIC ON THE TECHNOLOGY HISTORY LEADING TO THE CREATION OF CROWDMAP
1999
2000
2008
2009
2010
IKONOS LAUNCHED
The world’s first
high-resolution
remote
sensing satellite. It and
a feet oF other satellites that
Followed helped create the
highly detailed geographical
inFormation archives that
makes open source digital
mapping
possible
population with
Access to mobile
phones
PlatForm to aggregate and
share
the Kenya post
election violence reports
population with Access to
mobile phones
ushahidi formed
ushahidi platform
[ crowdsourced event mapping ]
2%
28%
launched
Hosted platForm oF the ushahidi
platForm
15,000+
maps use the hosted crowdmap
platForm
20,000+
maps deployed since 2008
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
9
Crowdmap users have and have they used this kind of technol±
ogy before?
The third tier comprised metrics that could only be produced
through direct engagement with those individuals who created
online maps. What objective did users have when launching
their crowdsourced map? Were they successful? This latter tier
required individual semi±structured interviews.
In sum, the research did not focus on the highly engaged
deployments or the deployments without engagement much
beyond an initial investigation of the Crowdmap platform.
Stated more precisely, we excluded from our analysis deploy±
ments of Crowdmaps with over 10,000 reports and those with
fewer than 21 reports.
The GWU team’s first major
finding a²er analyzing the
Crowdmap data for “Tier 1”
metrics was the discovery
of a “long tail” distribution.
In other words, the vast majority of Crowdmaps have very few
to no reports while only a handful of Crowdmaps have thou±
sands and even tens±of±thousands of reports.
Since Crowdmap is both free and relatively easy to use, this
finding was actually not a surprise. With a few clicks of the
mouse one can “deploy” a Crowdmap.
Crowdmap is, by design,
an easy access platform.
With such a low barrier to entry, a
power law distribution should be expected.
It should also be
kept in mind that more data does not necessarily equate with
better or more successful uses of Crowdmap. The success of a
given Crowdmap deployment is dependent on the purpose of
the map, which is not always well defined, if it is defined at all.
More data does not automatically imply greater impact.
Finally, focus on the long tail of deployments distracts us
slightly from the impressive number of robust deployments
that occurred between August 2010 when Crowdmap went live
and October 2011 when the data were obtained. On average,
there were 914 deployments per month. Future research will
focus on the head of the long tail. So GWU researches con±
ducted a “Tier 1” analysis on the “middle” of the tail, i.e., those
Crowdmaps with reports numbering between 21 and 10,000.
A
total of 585 Crowdmaps (out of 12,795) fit this category, which
was a more manageable number to work with vis±à±vis produc±
ing the “Tier 1” metrics.
See the Haiti Case Study
(page 21) or view online
at
http://crowdglobe.net/
deployments
10
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
For Tier 2 metrics, the team developed a survey, which was
sent to all Crowdmap users. The purpose of this part of our
research was to learn more about Crowdmap deployments from
deployment administrators. What works well about Crowdmap
as a crowdsourcing platform and what doesn’t?
In terms of
Tier 3 metrics, the GWU team engaged users of the download±
able Ushahidi platform. The GWU team wanted to know more
about these more technically savvy users.
To this end, the team developed a separate set of questions for
the purposes of carrying out dedicated semi±structured inter±
views. A total of five high±profile Ushahidi deployments were
selected: Haiti, Libya, Japan, Sudan and Egypt. In addition to the
interviews, background research on these five case studies was
carried out to provide more context to the analysis.
Quantitative Results & Analysis
“Tier 1” metrics were produced by running a series of database
queries on each of the 12,795 Crowdmaps. The findings were
then compiled into a single database, the main results of which
are showed here:
FIGURE 4:
THE LONG TAIL DISTRIBUTION VS THE PARETO DISTRIBUTION
FIGURE 5:
RESULTS OF “TIER 1” ANALYSIS DONE ON THE ROW DATA OF THE CROWDMAP DEPLOYMENTS
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
11
The main results from these initial findings pointed to the
next most logical step in our analysis.
More focus was given
to survey responses from both Crowdmap users and Ushahidi
users. Extant research suggested that the “Tier 1” analysis
would reveal that an 80/20 “Pareto Principle” would govern the
results.
The Pareto Principle is a name for a type of power law distri±
bution: In the Crowdmap case it would predict that 80% of
Crowdmap deployments would have little±to±no data, while the
remaining 20% would have the vast majority of the data.
Yet,
as noted above, while about 61% percent exhibited virtually
no activity beyond installation, 93% of Crowdmap instances
reported fewer than 10 reports. In short, the power law distribu±
tion was far steeper than the Pareto Principle would anticipate.
Power Law Distribution
This result prompted a question about the broader Crowdmap
ecosystem.
Is this sort of distribution common across crowd±
sourced phenomena?
Many natural and social phenomena
cluster around a mean or typical value.
This distribution is cap±
tured by references to averages: average shirt size or average
speed of cars on a freeway.
The past decade and a half has
seen an explosion of scholarly and popular interest in a differ±
ent pattern—power law distributions.
Power law data have an inverse, exponential relationship
between the magnitude of an observation and its relative fre±
quency. Steep power laws combine concentration at the head
with long, heavy tails. They have no “typical” value, as almost
all of the observations are below the global average.
The size of earthquakes versus their frequency, the size versus
frequency of solar flares, craters on the moon, wars, and even
word use all follow this pattern.
In English, for example, occur±
rences of
a, and,
and
but
are frequent (head), whereas words
such as
oxymoron
, and
polymorphous
are used infrequently
(tail).
The online environment seems particularly prone to the pro±
duction of power laws in a host of different areas, from the ‘link’
topology of the Web to the size distribution of open source
so²ware projects, from blog traffic to the popularity of YouTube
videos. Does the Crowdmap data display the same empirical
regularities?
Analysis of the Crowdmap data suggests that the size of proj±
ects, as measured by reports, does indeed follow a rough
power law.
Formally, we would consider the Crowdmap data
to be power law distributed if the probability that a randomly
selected map had K reports is proportional to K
±alpha
.
The figure below plots the size distribution of Crowdmap proj±
ects on a log scale.
On the Y±axis is the size of a project, as
FIGURE 6:
OUR INITIAL PROCESSING SHOWS THAT THE VAST MAJORITY OF DEPLOYMENTS HAVE LITTLE TO
NO ACTIONABLE
12
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
measured by the (logged) number of reports on the map.
On
the X±axis is the (logged) number of maps that have at least Y
number of reports.
The signature of a power law distribution is that it should form
a straight line on a log±log scale.
To a first±degree approxima±
tion, that is what we see here. In empirical data, it is common
to observe that the very largest observations are smaller than
a pure power law would suggest. This clearly seems true in the
Crowdmap data. A linear fit is overlaid on top of the empiri±
cal data; it covers deployments with at least 20 reports, and
excludes the five largest map projects.
The body of the data is highly log±linear. Using the methods
laid out in Clausset, Shalizi, and Newman (2007), we fit a power
law to the data using maximum likelihood. These methods sug±
gest a relatively steep value for alpha of 1.58. Deciding whether
the Crowdmap data most closely follow a power law or another
roughly log±linear distribution (such as an extreme lognormal
or a power law with an exponential cutoff) is not explored here,
as this technical question makes no difference to the substan±
tive conclusions.
Power law or log±linear distributions are found in so many
domains, in part, because they can be generated by a host of
different underlying mech±
anisms.
Merely observing a
power law in the Crowdmap
data still leaves open the
question of what underly±
ing phenomena actually creates and sustains this pattern. One
possible explanation for the log±linear pattern in the size of
map projects is that they are, in fact, reflective of a power law
distribution in offline data.
It is well documented that several of the phenomena that
Ushahidi attempts to map—such as the magnitude of earth±
quakes or the size of armed conflicts— follow a power law.
A
handful of earthquakes or wars are massive, while most are
small.
If Crowdmap contained data on every single damag±
ing earthquake, for example, it is likely that the distribution of
reports of damage or loss of life across maps would end up
following a power law distribution.
Still, there is suggestive evidence in this data that there may
be a compounding effect of success. The more reports a
Crowdmap project has, the more reports it seems to attract,
leading it to a positive feedback loop. In physics, power law
relationships o²en reflect phase transitions. It is possible that
there is an analogous process by which a map project reaches
critical mass. If confirmed, this may indicate the importance of
strategies to get nascent map projects “over the hump.” This is
a promising area for future research.
What distinguishes Crowdmap users who appear to be mere
“tire±kickers—analogous to those who go to an auto show±
room only to look at, rather than buy, a car—from those who
are fully engaged with the process? Since addressing this
question would benefit Ushahidi with a much greater under±
standing of the strengths and weakness of their crowdsourc±
ing platform, more attention was given to analyzing the 585
Crowdmaps that had between 21 and 10,000 reports.
A rigorous, systematic content analysis of the 585 Crowdmaps
was the carried out by two GWU graduate students.
A GWU
researcher designed the coding instrument and trained the cod±
ers.
The results revealed that the vast majority of these (30%)
focused on North America while 18% focused on Western
Europe and 16% on Africa. On average, these Crowdmaps had
814 reports.
The median number of reports for this set of deployments was
substantially lower, at 94, which is not surprising considering
that the distribution of this set of cases is highly right±skewed.
The content analysis also revealed what the Crowdmaps were
being used for (see Appendix I for a dictionary for Crowdmap
data for the definitions of categories used below). Indeed, the
FIGURE 7:
A FIRST-DEGREE APPROXIMATION OF THE LOGGED NUMBER
OF REPORTS VERSUS THE LOGGED NUMBER OF MAPS
A first-degree approximation of the logged number of reports versus the logged number
of maps shows a power law distribution of “actionable” reports, as determined by the
research staff: reports with more than 20 reports but fewer than 10,000.
See the Japan Case Study
(page 24) or view online
at
http://crowdglobe.net/
deployments
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
13
four most frequent themes that emerged across the data (see
Fig. 2) include: Emergency±related infrastructural issues (which
appeared in 22% of Crowdmap), non±emergency infrastruc±
tural issues (also 22%), crime and public safety issues (21%),
and media reports (21%, which includes both traditional and
new media outlets).
The next most frequent set of themes
included: Civic, non±governmental, and government organiza±
tions (20%), natural disasters (18%), entertainment and lei±
sure (17%), and Health and Medical±Related Issues (17%).
Not surprisingly, however, when the analysis is broken down
by region, the relative frequency with which these themes
emerged in the regional deployments differs dramatically.
For
example, taking into account recent events, it is not surpris±
ing that the most common themes that emerged from the 63
deployments in the Middle East and Northern Africa pertain to:
Crime and public safety issues (43%), human rights abuses
(40%), emergency±related infrastructural issues (30%), and
political organization (25%).
The distribution of themes in the 79 Western European deploy±
ments, on the other hand, paints a very different picture, with
entertainment and leisure appearing in 32% of the deploy±
ments, followed by non±emergency infrastructural issues
(25%), and media reports (23%).
It is also not surprising that the 16 deployments from the
Caribbean region, 12 of which hailed from Haiti specifically,
FIGURE
8:
GEOGRAPHICAL DISTRIBUTION OF THE CROWDMAPS ANALYZED
FIGURE 9:
CROWDMAP MOST FREQUENT THEMES
14
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
heavily featured issues related to the occurrence and a±er²
math of a natural disaster (63% and 50%, respectively), emer²
gency²related infrastructural issues (63%), health and medi²
cal²related issues (50%), and crime and public safety issues
(38%).
As an additional point of comparison, it is interesting to note
that there is limited difference in the frequency of specific
themes when comparing the ends of the distribution in terms
of number of reports.
For example, comparing the themes that
emerged in deployments in the lowest decile in terms of num²
ber of reports (i.e. those in the bottom 10 percentiles, with less
than 32 reports) to deployments in the top decile (i.e. those
in the top 10 percentiles, with more than 1053 reports), the
majority of the themes appeared with commensurate propen²
sity in both sets of deployments.
In other words, the theme
of crime and public safety issues was equally likely to appear
in deployments with a limited number of reports (16%) as it
was to appear in deployments with a large number of reports
(16%). Among the handful that exhibited marginal differences,
only three themes surpassed a differential of 10%²points or
more.
Specifically, these themes included issues pertaining
to animals, fish, and birds (which were 10²percentage points
more likely to be featured in deployments in the top decile),
Environmental issues (which were 14²percentage points more
likely to appear in deployments in the bottom decile), and non²
emergency infrastructural issues (which were 14²percentage
points more likely to appear in deployments in the bottom
decile).
Survey Research
The quantitative analy²
sis has shed important
insights on the patterns
of Crowdmap users. In
order to complement this
research and possibly explain the quantitative findings, this
first Crowdglobe report took a mixed methods approach, com²
bining quantitative analysis with qualitative research. The latter
forms the second part of the report below.
The purpose of the survey research was to complement the
quantitative analysis carried out above. More specifically, the
surveys are meant to place the numerical analysis into context
and provide additional insights to the quantitative trends iden²
tified above. Since the quantitative analysis focused on both
Crowdmap and self²hosted Ushahidi deployments, the research
was composed of two surveys and case study analysis. For
Crowdmap, a dedicated survey was sent out to 12,000+ users.
The questions for this survey are listed online at
crowdglobe.
net/our²report/survey²data.
The Crowdmap survey was shared via Google Forms and 276
users responded to the survey, i.e., a response rate of ~2%.
While this certainly does not constitute a random sample, the
results are highly informative and relevant to the CrowdGlobe
research project and Ushahidi.
FIGURE
10:
THIS GRAPH REPRESENTS THE
PERCENTAGE OF MAJOR THEMES.
See the Sudan Case Study
(page 25) or view online
at
http://crowdglobe.net/
deployments
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
15
A separate survey was developed for users who launched high±
profile projects using the self±hosted Ushahidi platform. A total
of 37 high±profile projects were identified. The second survey
was thus shared with those individuals responsible for these
projects. A copy of this survey is available on
crowdglobe.net/
our±report.
A total of seven respondents completed the survey,
i.e., ~19% response rate. In addition to this survey, five case
studies were selected for more in±depth, secondary research.
The case studies, which included throughout this report are:
Haiti Earthquake, Libya Crisis, Japan Tsunami, Sudan Election
and Egypt Elections.
Survey Results & Analysis
The 276 surveys completed by Crowdmap users provided
interesting insights into the patterns of engagement with the
Ushahidi so²ware. Users’ level of experience in using crowd±
sourcing & mapping technologies prior to trying out Crowdmap
was particularly limited. About 53% of users responded as
having no prior experience, which is not surprising since the
technology is still relatively new. About 27% noted that they
had only used a similar technology once or twice before. About
80% of first±time Crowdmap users were particularly new to this
type of technology. Only 9% of users considered themselves as
having considerable to extensive prior experience.
In terms of experience as a contributor to digital maps (as
opposed to hosting), the distribution of responses was consid±
erably less skewed. While 31% of Crowdmap users had not con±
tributed to digital maps in the past, about 22% of respondents
noted that they had considerable to extensive prior experience
in contributing to digital maps. About 30% of users answered
that they had contributed to some digital maps in the past.
As for prior experience in hosting a digital map, about 55% of
users had never hosted one before while only 5% noted that
they had a great deal of experience in hosting such maps.
The most important reason why Crowdmap users used the
Crowdmap platform was to create a map with a specific pur±
pose or event (40%). About 63% of these users launched a
map to cover an event in the city in which they live. About 21%
of users chose Crowdmap for demonstration purposes while
35% used Crowdmap to learn more about the technology and
Ushahidi. In other words, more than half of Crowdmaps were
created for non±direct use. Others wrote in specific answers
such as “Great tool for training others on Ushahidi”; “To use
it for my line of work”; “I work for UNHCR”; “All of the above”;
“College project”; “Wanted to create a social mapping project”;
“Wanted to test Ushahidi to make it better”.
When users began to use Crowdmap, about 45% of them found
that the tool made sense and was easy to use while 36% of
respondents explained that the platform was slightly confus±
ing even though they were still able to figure it out and use
the technology. Around 9% of users complained that they
could not find training material or documentation to help them
use the platform. Approximately 6% of respondents gave up
because they could not make sense of the platform and were
never able to get their map to work, which is a surprisingly low
number.
As the quantitative analysis of the Crowdmap meta±data
revealed, the majority of Crowdmap users who set up an
account do not actually end up creating a map. Users cite sev±
eral reasons for this. About 14% note that they didn’t intend to
create a map in the first place and were simply curious. Around
9% of users said the technology was simply too technically
challenging for them to work with. Approximately 6% of users
revealed that they “gave up” trying to create a map when they
realized it would take too much time. About 3% of users had
security concerns with Crowdmap and therefore elected not to
use the platform. Around 4% of users wanted to create a map
but ultimately felt it was not the right tool for their project.
FIGURE 11:
GRAPH OF THE INITIAL USE OF THE CROWDMAP
PLATFORM
16
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
FIGURE 12:
THE REASONS AMONG SURVEYED CROWDMAP USERS SIGNING UP FOR A CROWDMAP ACCOUNT
BUT ULTIMATELY NOT CREATING AN USHAHIDI INSTANCE.
Lastly, approximately 9% of users did not understand the pro±
cesses for aggregating data from other sources such as SMS,
Twitter, email, etc.
Several users wrote in specific answers to explain why they
never set up any map using the Crowdmap platform. These
included “I was only using the platform as a demo”; “Others may
not be able to easily navigate using the tool”; “Internal buy±
off from organization created barriers—be nice to have quick
summary of benefits/successes”; “the intended users did not
accept it”; “was too confusing”; I wanted to consult maps which
were already done in the site.
..but I didn’t find.
..or couldn’t
find”; “Wanted to create a map but the themes where too limit±
ing”; “because I could not get data from expected sources”; “I
wanted to show others the capability and determine if it would
fit their needs”; “I would like to finish this, but I found it very
challenging to complete the setup. Can you help me?” “The col±
lege project required deployment to azure—not enough time
available to get PhP server running, tie in with sql server and
get apache et al running on azure fabric”; “I only view maps”;
I needed more control for access. As such, I installed Ushahidi
and have been using that”; “Mobile App needed improvement,
need more controls for custom forms.”
Of those users who concluded that Crowdmap was ultimately
not the right tool for them, the most important reason cited
(by 18% of users) was that the platform could not be cus±
tomized to meet their needs. About 7% of users replied that
Crowdmap was too complex while 3% felt the tool was too
simplistic. Others had more specific replies, such as “I did not
understand what Crowdmap’s capabilities were”; “need funding
for project”; “cost of SMS messages”; “It is very hard to engage
people in something new”; “data ownership”; “the context did
not suit—low connectivity”; “I could not spend time learning
the technology”.
In terms of users who did launch a Crowdmap, about 16% felt
that they were not able to generate the required public aware±
ness vis±a±vis their map to make it as effective as they had
hoped. Around 19% noted they were able to make their map
almost as effective as they hoped while 31% replied that they
simply were not able to garner sufficient interest in their map.
Of those users who felt they had been successful in raising
awareness of their maps, about 23% noted that they had done
so via engagement with community organizations, civic groups
and/or professional organizations. Around 20% said they had
successfully built public
awareness by engaging
with social media, SMS,
etc. Only 6% of users said
they used traditional news
The above chart showing the reasons among surveyed Crowdmap users signing up for a Crowdmap account but ultimately not
creating an Ushahidi instance. The majority of respondents cited having no intention to create a map or finding the process to
technically challenging.
See the Egypt Case Study
(page 27) or view online
at
http://crowdglobe.net/
deployments
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
17
media to build awareness of their project.
About 30% of prospective re±users of Crowdmap wrote that in
the future they would consider using the platform if they iden±
tified a need. Around 10% would consider using Crowdmap if
they had more time. About 8% of users would use the platform
if it was easier and/or technical support was offered.
Respondent demographics were also interesting. About 80% are
men and the average age of a user is 40 years old. As for educa±
tional background, an impressive 43% of users have a post±grad±
uate degree and a total of 84% have at least a college degree.
While only 19% of respondents completed the survey for the
self±hosted Ushahidi deployments (i.e., not Crowdmap), the
results were nevertheless insightful. All applications of the
Ushahidi platform were in response to a complex humanitar±
ian emergency or “natural” disaster, with two deployments
being in developing countries and the remainder in the “Global
South”. The UN Office for the Coordination of Humanitarian
Affairs (UN OCHA) accounts for the majority of deployments fol±
lowed by media organizations (Washington Post and Australian
Broadcasting Corporation). Only one deployment was carried
out by a relatively small NGO. Most of the respondents indi±
cated that they had minimal prior experience in using the
Ushahidi platform before deploying their project. But most did
note that they were already moderately experienced in contrib±
uting to digital maps.
Challenges
Some of the biggest challenges cited by users of the Ushahidi
platform included “keeping reports up to date, embedding pic±
tures and documents;” “the need to display the data on some±
thing other than a map;” “getting the word out quickly a²er
the launch of the site;” and “getting people to submit reports.”
Other challenges cited ranged from the difficulties in acquiring
an SMS short code, being too dependent on the Ushahidi team
to fix technical problems and bugs and information overflow.
One recurring difficulty cited was fundraising to set up and
maintain the project.
When asked about failures, respondents’ answers ranged quite
widely, from persistent technical problems such as bugs to
managing and mobilizing volunteer involvement. Many com±
plained that using the platform was simply too time±consum±
ing. Others, like humanitarian organizations, noted the “limited
use of the system by traditional humanitarian entities,” which
explained the “lack of understanding of how the system can be
used,” and the fact that the system was just “too hard for high
level decision makers to get what they needed.”
The media organizations that used the platform tended to
highlight the interface as being problematic: “It was difficult
to drill down and get information for a particular area or time
period (e.g.
. last 24 hours) easily. There were also a number of
features on the default interface that didn’t seem to work, such
as the filter for reports with images and videos. “We wanted to
try to customize the interface a bit more but there were limited
options.” Furthermore, one media group added that their “jour±
nalists tried to verify some of the audience reports as they came
in but found it too difficult and time consuming to do in reality.”
Successes
As for successes, respondents’ responses ranged widely as well.
On the technical front, features such as RSS feeds, dynamic
statistics and embedding mainstream news proved easy to use.
Using the Ushahidi platform allowed one humanitarian orga±
nization to “mobilize key counterparts in the natural disaster
monitoring and response field and learn about their interests
and the potential uses they could give to the platform.
It also
allowed [them] to interact with local municipalities and depart±
mental government in a way we had not done before.”
The fact that existing government data can also be mapped over
time and space enabled one government to better understand
the potential of crisis mapping. Another humanitarian organi±
zation documented how working with the technology and the
Standby Volunteer Task Force (SBTF) has changed the way that
they work internally, in terms of how they organize and process
information. One major humanitarian group explained that their
Ushahidi map “served to help stranded migrants and provide
information on [the organization’s] operations. Another group
explained that while the impact of their map was minimal, the
project’s lessons learned were instrumental in launching their
subsequent map, which provided an “alternative vision of disas±
ter information that allows for various media to be brought in.”
This in turn generated “better awareness for the responding
organizations.”
One media organization revealed that their Ushahidi map
received over 230,000 hits over a 3±week period and a total of
1,500 reports. This “showed the potential of real±time crowd±
sourced mapping tools and showed that this sort of tool could
be used in emergencies±±at least in slow±moving ones such
as floods—without too great an editorial risk.” Viewers were
no longer passive observers but participants in the process.
“It gave our audiences an opportunity to assist in reporting
on an unfolding event as it was happening rather than ringing
the local radio station they could directly submit a report to us
18
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
online. Ideally in the longer term we would like our news and
radio presenters to be using he map as a reference tool as well
as making call outs to the audience to submit reports.” Another
media group explained that citizens were able to leverage the
Ushahidi map to organize their own response efforts, which
“could not have been done with out it [i.e., the map].” Finally,
one NGO highlighted how the technology helped to inspire
greater participation in a peace movement. This finding aligns
with those identified in the short case studies on Haiti, Sudan,
Egypt, Libya and Japan.
All respondents noted that they would use (or already had
used) the Ushahidi platform again in the future. However one
organization expressed some important hurdles: “We had some
of our developers review the code for Ushahidi to consider fur±
ther use and they reported that the code was not well struc±
tured or documented and would be difficult for us to build on
and customize. There would also need to be improvements in
the User Interface to make it more customizable.”
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
19
CONCLUSIONS AND RECOMMENDATIONS
3
What this report has tried to do is set a standard for rigorous,
high quality data analysis of crowdsourced data. It has, to be
frank, offered a few surprises along the way. Yet it is important
to offer a bit of perspective on Ushahidi in particular and crowd±
sourcing in general.
As we assess the growth and impact of Ushahidi in general and
crowdsourcing in particular we should keep in mind that we are
still at the very start of a transformative process. This report
might be thought of as offering the equivalent of an Apgar
score for a newborn. (The Apgar score refers to the results of
several simple tests devise to measure the health and viability
of newborn children immediately a²er birth). There is plenty
more to do and learn before we can draw any firm conclu±
sions, particularly vis±à±vis impact. Crowdmap, for example, is
barely a year±and±a±half old, which means that users are still
very much in the pioneering and discovery phase. Recall Clay
Shirky’s point that “technology only becomes socially interest±
ing when it becomes technologically boring.” This explains why
the CrowdGlobe Project is intended to launch the means of an
ongoing assessment of where we are now and what we can
expect in the future. Hence the
CrowdGlobe.net
portal. To be
sure, this is not the final statement about crowdsourcing and
Ushahidi. It is the opening statement of a new field of inquiry
and civic action.
Crowdsourcing is an important new tool of accountability. From
its inception, Ushahidi has allowed motivated populations to
hold governments accountable for misdeeds and mismanage±
ment. From Kenya to Russia, crowdsourced information has
provided insight into situations that otherwise would have
remained out of sight and out of mind (at least for those not
caught up in the chaos). Because journalists and human rights
or aid workers cannot be everywhere to monitor human rights
abuses or the condition of desperate people caught up in des±
perate situations, crowdsourcing could bring a new level of
awareness to circumstances such as these. The data we have
presented show that this clearly does not always happen, and
for a variety of reasons spelled out in our survey data. But the
fact it has happened despite some of the stated challenges and
that it may happen again may give pause to those in power who
expect a free hand just because they have muzzled the press
and intimidated some into silence.
On the research front, this initial study of Crowdmap data
points to the necessity of careful empirical analysis done by
those who are trained in research methods appropriate to the
analysis of network dynamics. In particular and as noted above,
our discovery of the prominence of the long tail in Crowdmap
deployments led to a greater research emphasis on under±
standing the factors that prevent curious “tire kickers” from
becoming active users. Coupled with sophisticated quantitative
methods is the need for in±depth qualitative field analyses of
deployments. We have not been able to do that here, relying
instead on the self±reporting via survey±based research. Future
research should investigate the factors that lead to successful
deployments, and those that do not. This is precisely why the
CrowdGlobe.net website has been launched—to start a con±
versation and a collaborative effort.
This takes us to what is, in a sense, our most important pre±
liminary conclusion. This project’s greatest contribution might
come in the form of establishing best practices for users and
potential users, and perhaps even new procedures at Ushahidi.
Out of the pain of the post±election violence in Kenya in 2008 emerged an ingenious tool for tapping into the
potential power of people who, while physically separated, could become bounded together by electronic
networks. As we’ve seen in this report, there are still encumbrances to the realization of this new potential.
Some view the technology as dauntingly complex, or the social dynamics or public awareness necessities
insurmountable. But other groups have tapped into this new powerful tool to create something entirely new
in collective action dynamics
20
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
that will give them far more visibility. Indeed, as the survey
results clearly showed, strong media and community outreach
is critical to gaining traction. Furthermore, Ushahidi’s user±
interface design team has also carried out a full usability review
of the Crowdmap platform, with substantial changes on the
way. Finally, Ushahidi has partnered with TechChange to offer a
dedicated, hands±on course on how to use Ushahidi/Crowdmap
and create successful deployments. These important improve±
ments, taken together, are bound to generate Ushahidi/
Crowdmaps that gain both more traction and visibility. In sum,
this report provides an important baseline study—and indeed
the only one of it’s kind—which could serve as an important
comparison if this research is replicated in the coming years.
In the meantime, we hope that further analyses will help us
understand the power law dynamics of crowdsourcing, leading,
eventually, to a less pronounced effect, if only at the margins.
For example, despite the surprisingly high levels of education
among many respondents, many still reported that the com±
plexity of the platform impeded their full use of it. This sug±
gests that further refinement and greater ease of use would
be beneficial.
The good news is that Ushahidi has already taken numerous
steps to address these and other challenges identified in this
report. Crowdmap, for example, now includes a dedicated
“wizard” to guide first time users through the customiza±
tion process. The organization has also recruited a full time
Community Manager who has organized more end±user meet±
up’s in 2012 than in all four previous years combined. In addi±
tion, the Community Manager has launched a dedicated wiki to
provide significantly more documentation on how to use the
platform. Crowdmap developers are also developing a public
online library to facilitate the discoverability of Crowdmaps
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
21
4
CASE STUDIES
CASE STUDY 1: Haiti
A devastating earthquake struck Haiti on January 12, 2010,
resulting in hundreds of thousands of lives lost. Within hours,
the Ushahidi’s Patrick Meier and David Kobia launched a live
Ushahidi Map of Haiti. During the first few days, the content
mapped on the Ushahidi platform was sourced from Twitter,
Facebook and other online sources such as mainstream media.
Soon, they couldn’t keep up with the deluge of information
on Haiti. So Patrick Meier reached out to colleagues at Tu±s
University for support and by the end of the first week had
trained more than 100 volunteers on how to live map Haiti.
During this time, a free SMS short code was secured from
Digicel, Haiti’s main telecommunications company. This allowed
anyone in Haiti to text in their most urgent needs and location,
which could then be mapped on the Ushahidi platform.
FIGURE 13:
THE USHAHIDI-HAITI CRISIS MAP SPEARHEADED BY TUFTS UNIVERSITY
22
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
Some ten days afer the Haiti map was launched, the head o±
the US Federal Emergency Management Association (FEMA),
Craig Fugate, noted that the live map provided the most com²
prehensive and up²to²date in±ormation available to the human²
itarian community. What is striking about this statement is that
the map was not launched by FEMA, or the United Nations, or
any pro±essional humanitarian organization, ±or that matter.
The live map was launched by student volunteers ±rom a dorm
room in snowy Boston some 1,500 miles away ±rom Haiti. Over
3,000 reports were mapped and according to the Marine Corps
and US Coast Guard, the Haiti Crisis Map helped them save hun²
dreds o± lives. But this live map would not have been possible,
were it not ±or two other equally remarkable volunteer²led ini²
tiatives in Haiti: OpenStreetMap and Mission4636.
In the wake o± the Haiti earthquake, the Google Map ±or the
Port²au²Prince area was highly incomplete. This made it very
difficult to find street names let alone specific addresses when
mapping in±ormation on the Haiti map. The OpenStreetMap
(OSM) community came to the rescue by crowd²sourcing the
most detailed and comprehensive map o± downtown Port²au²
Prince. They did this by tracing satellite imagery and making
their map openly and ±reely available. Some 600 volunteers
±rom several dozen countries contributed over one million
edits to the Haiti OSM map during this period. Needless to
say, the team at Tufs quickly switched ±rom Google Maps to
OpenStreetMap as a result.
Meanwhile, Mission4636 was in ±ull swing. Thanks to Digicel’s
support, anyone in Haiti could text the number 4636 ±or ±ree
to communicate their most urgent needs and location. In this
way, as in Kenya, the team leveraged the high prevalence o±
mobile phones to crowd²source needs assessments in real²
time ±rom the disaster affected communities in Port²au²Prince.
The incoming text messages, however, were in Haitian Creole.
They needed to be translated into English i± the in±ormation
was going to be used by the humanitarian community. This is
where Mission4636 came in. Within a week, over 1,000 volun²
teers ±rom more than 40 countries were recruited (primarily via
Facebook) to translate incoming SMS’s in near real²time. These
volunteers were predominantly ±rom the Haitian Diaspora com²
munity. Together, they translated some 80,000 text messages,
with an average turn²around time o± 10 minutes per SMS.
Mission4636 volunteers also mapped these messages so they
could be added to the Haiti map being curated by the volunteer
team at Tufs University. Together, OSM²Haiti, Mission4636 and
the Haiti Crisis Map demonstrate the important role that new
crowdsourcing technology can play in supporting humanitarian
response.
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
23
CASE STUDY 2: Libya
On March 1, 2011, the head of the Information Services Section
(ISS) at the UN Office for the Coordination of Humanitarian
Affairs (OCHA) reached out for help. He and his team needed to
get a better picture of the humanitarian crisis unfolding in Libya
in order to plan their response operations. OCHA had no person±
nel on the ground at the time and the ISS team could not rely
on the information produced by the regime in Tripoli. But the
Arab Spring had shown how much information existed in the
social media space, and therefore ISS Head Brendan McDonald
called for a live Crisis Map of Libya to better inform their situ±
ational awareness.
Within hours of this request, a live Crisis Map was launched.
Operational crisis mapping in hostile environments pres±
ents some important challenges, not least of which is secu±
rity. This explains why the two live maps were produced. One
was password protected and exclusively for the humanitarian
community while the other was public but on a 24±hour time
delay and with heavily redacted information.
Just days a²er the launch, the Executive Director of the World
Food Program (WFP), Josette Sheeran, noted that the live
map provided an excellent resource to plan her agency’s relief
operations along the Egyptian and Tunisian borders. Like all
the other case studies featured above, the Libya Crisis Map
was made possible thanks to a vast volunteer network. Unlike
some of the earlier crisis mapping efforts, this network was
organized and prepared. The Standby Volunteer Task Force for
Live Mapping (SBTF) is a global volunteer network of some 700
volunteers based in close to 70 countries who are trained in live
crisis mapping operations. Many of these volunteers are vet±
erans from the Kenya, Haiti and Russia Crisis Maps. Together,
they mapped over 2,000 reports on the Libya Crisis Map. So
when the UN’s Brendan McDonald reached out for help, he was
actually requesting the activation of the SBTF.
FIGURE 14:
THE LIBYA CRISIS MAP LAUNCHED FOR THE UNITED NATIONS
24
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
In McDonald’s own words, the SBTF provided invaluable sup±
port to the UN’s operations in Libya. In an thank±you email
addressed to the volunteer network, the head of ISS wrote
the following: “Your efforts at tackling a difficult problem have
definitely reduced the information overload; sorting through
the multitude of signals on the crisis is no easy task. The Task
Force has given us an output that is manageable and digestible,
which in turn contributes to better situational awareness and
decision±making. I look forward to further developing our col±
laboration in this emergency and beyond.”
Around the same time that the Libya Crisis Map was launched,
the International Organization for Migration (IOM) also launched
a live crisis map to help inform their decision±making.[3] The
purpose of this live map was to monitor the migration crisis
resulting from the violent conflict in Libya, which resulted in
the need to evacuate thousands of stranded migrants along
the borders of Egypt, Tunisia and later Niger.
CASE STUDY 3: Japan
During the Libya Crisis Map deployment, another crisis on the
other end of the planet struck. The devastating earthquake
that shook northern Japan on March 11, 2011 resulted in
a massive tsunami that took tens of thousands of lives and
destroyed critical infrastructure such as mobile phone commu±
nications. Inspired by what they had seen in Haiti, volunteers
from the Japanese OpenStreetMap (OSM) community launched
their own live crisis map and shared the word via multiple social
media channels
The OSM volunteers who were operating out of Tokyo closely
monitored the Japanese Twittersphere, mapping an average of
3,000 tweets per week during the first month of operation. In
this way, the Japan Crisis Map provided the most comprehen±
sive and up±to±date information available on the impact and
resulting needs. In total, the Sinsai.info map was accessed
by over half±a±million users most of whom were based in the
FIGURE 15:
THE JAPAN CRISIS MAP LAUNCHED JUST HOURS AFTER THE 2011 TSUNAMI.
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
25
disaster affected area. Obtaining relevant data in a timely man±
ner from the Japanese government was difficult, which made
the Sinsai.info deployment even more important. Again, how±
ever, ordinary volunteers spearheaded this initiative, and most
of them had never done anything like this before. The author
and other volunteers from the SBTF also provided the Japanese
team with both strategic and technical support.
CASE STUDY 4: Sudan
Sudan Vote Monitor was the first time that an Ushahidi plat±
form was deployed in a country under authoritarian rule. The
pilot project was led by the Sudan Institute for Research and
Policy (SIRP) and Asmaa Society for Development, in collabo±
ration with other Sudanese civil society organizations. The
purpose of deployment was to utilize the Ushahidi platform to
support the independent monitoring and reporting of Sudan’s
first multi±party elections in 24 years.
The initiative complimented the paper±based independent
monitoring efforts of formal election monitoring groups and
offered Sudanese NGOs and the public at large an independent,
online platform for election observation for the first time in
Sudan’s history.
The Ushahidi platform was considered particularly useful in
Sudan, Africa’s largest country, where long distances and inad±
equate infrastructure posed a significant challenge to civil soci±
ety election monitors.
The spread of mobile communications throughout the coun±
try in recent years offered a unique and feasible opportunity to
utilize SMS to overcome this challenge. Participating civil soci±
ety groups deployed over 2,000 independent local observers
throughout the 15 northern states. According to SIRP, “these
observers continuously reported back what they witnessed at
various polling stations across these states, using standard
paper reporting forms. When texting, they used code, e.g., 1
= election fraud, 2 = voter intimidation, etc. This was done to
FIGURE
16:
SCREENSHOT OF THE SUDAN VOTE MONITOR PLATFORM AFTER THE ELECTIONS
26
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
provide more cover to the citizen monitors.” It is unclear how
many text messages were received, however.
The site went live on April 10, 2010 with web and SMS report±
ing in both English and Arabic to coincide with the start of
the elections held April 11±15, 2010. Response was relatively
strong both inside and outside the country given that this was
the first project of its kind in the Sudan. According to SIRP, “a
total of 564 reports were received from the web (or trans±
lated from paper±based forms) from 419 locations, covering
26 election±monitoring categories. The web±based platform
attracted wide interest from citizens, a variety of international
organizations active in Sudan, as well as the local National
Telecommunication Commission.” However, the Sudan Vote
Monitor website was blocked by the Sudanese government
for two days before it was unblocked following US govern±
ment pressure. In addition, fewer than 300 reports are actually
mapped on the Ushahidi platform and certainly not from more
than a dozen or so different locations.
In general, and compared to the other Ushahidi case studies
figured in this report, the Sudan Vote Monitor project is not
generally considered a success. The project was not well orga±
nized and only came together at the last minute. Indeed, the
SMS short codes that were used for the project were only made
available the night before the elections. So the organizers had
had no time to get the word out about the SMS let alone carry
out trainings or any simulations. This perhaps explains the
fact that only 300 reports were submitted when there were a
reported 2,000 election observers.
CASE STUDY 5: Egypt
In the Fall of 2010, the Egyptian and Cairo±based Development
and Institutionalization Support Center (DISC), used the
Ushahidi platform to launch U±Shahid; the goal of which was
to monitor the parliamentary elections in November and
December 2010. This independent initiative became particu±
larly important when the Mubarak regime announced that it
would not permit any official international election monitoring
groups into the country.
The project was rather simple on paper: use the Ushahidi plat±
form to monitor the elections by allowing people to send SMS,
Tweets, Facebook comments, voice mail, e±mail and reports
via web±form to the live map. DISC decided to draw on both
crowdsourced reporting and “blogger±sourced” information.
This meant getting the word out to the wider public while navi±
gating the restrictions imposed by Egyptian national security,
and also training a large network of trusted bloggers across
the country. Despite government restrictions, training for these
bloggers took place in 5 major cities: Cairo, Alexandria, Assyut,
Mansoura and Port Said.
On the technology side, DISC translated their Ushahidi platform
entirely into Arabic since the U±Shahid project was not meant
for an international audience but rather an Egyptian one: “an
Egyptian project for Egyptians” noted one blogger. Egyptian
so²ware developers integrated Twitter, Flickr and YouTube
with Ushahidi. Since Facebook was and continues to be an
important platform for Egyptian youths, the group also created
a Facebook feature that enable comments on a Facebook wall
to be easily mapped on the Ushahidi platform.
During the elections, DISC mapped 2,700 reports, which
included 211 supporting pictures and 323 videos. The team
of Egyptian bloggers was also able to verify more than 90%
of the content that ended up on the map by using basic jour±
nalist techniques such as triangulation and follow±up. Most of
the mapped reports, however, came from the pre±established
network of trusted bloggers, which did not require immedi±
ate verification. In total, the web±based map received close to
60,000 hits, the vast majority of which came from within Egypt.
Interestingly, the next highest number originated from Saudia
Arabia with just under 5,000 hits. The group was also pro±
active in disseminating this information, printing press releases
and combining both new and traditional media for maximum
impact. Their efforts were featured on Egyptian television, on
BBC Arabic and dozens of articles in ten different languages.
Indeed, both local and global media used the data generated
by U±Shahid as part of their election coverage.
Naturally, the project also got the attention of the Egyptian
government. Surprisingly, however, this attention began even
before the project formally launched. The Egyptian state con±
tacted DISC’s director Kamal Nabil when the program design
was still being developed. The government official told Nabil
that his name was recurring “too o²en” in phone conversa±
tions between activists. The Egyptian Ministry of Interior sub±
sequently shadowed the project in different ways: by tapping
the cell phones of bloggers who comprised the core team; by
requesting copies of the agendas for all meetings related to
U±Shahid; and by requiring that a list of all individuals trained
on the use of the platform be submitted to them. E±mail
addresses, Facebook pages and Twitter accounts of the core
team were reportedly all under surveillance since the start of
the project, and the Ministry of Interior openly asked Nabil what
his reaction would be if they were to shut down the U±Shahid
project before the elections.
DISC was not immune to this government strategy: several
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
27
new Facebook groups were launched to engage in personal
attacks against the core team by accusing them of being affili±
ated with the United States, under the pretext that they had
participated in a Freedom House±organized conference in DC
earlier that year. Some of those Facebook groups called on
young Egyptians to “watch out” for projects that could endan±
ger the national integrity and the political independence of the
country. Activists reacted to these attacks by conducting a vir±
tual battle. Once a government±supported group was identi±
fied, dozens of activists would write on the group’s wall and
basically occupying the entire wall with counter opinions. One
of these Facebook groups was completely overrun a²er the
group’s name was changed from “Youth for Funds” (sarcastic),
to “State Security for Intimidation.”
DISC was well aware that technology alone would not change
the political situation in Egypt. They also knew that Egypt’s
National Security could shut down the project and block access
to the website whenever they wanted. Furthermore, everyone
involved in the project knew full well that their involvement
in U±Shahid could get them arrested. As recent events have
clearly shown, countries like Egypt and the Sudan are particu±
larly agile in surveilling digital activists during election periods.
But this did not discourage the Egyptian activists. The ability to
do something different, to have an alternative was enough to
be the difference. At the end of an U±Shahid training workshop
in Cairo, one participant spoke with the lead trainer and simply
said: “You know? We may all end up in jail, but before this I
thought there was no hope to change anything. Now I can even
dare to think it is worth a try.”
FIGURE 17:
SCREENSHOT OF THE U-SHAHID PLATFORM AFTER THE ELECTIONS
28
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
Civic and Neighborhood Organizations
•± Civic±Organization
•± Support±Groups
Corruption
(Non-election related)
•± Bribery
•± Corruption±other
Crime and Public Safety (Non-War/Non-Protest related)
•± Attack
•± Gunshots
•± Looting
•± Murder/Deaths
•± Presence±of±Military
•± Presence±of±Police
•± Rape
•± Riots
•± Sexual±Assault
•± Sexual±Harassment
•± TheF
•± Vandalism
•± Arson
Education and Schools
•± Educational±Programs±and±Training
•± Schools
Elections
•± Campaigning
•± Election±Results
•± Elections±Other
•± ²raud±(Clear±instances±of±rigging±and±ballot±tampering)
•± Problems±at±Polls±(e.g.±Missing±Ballots±Voters,±Names±
Missing, Problems with Machines, Long Lines, etc.)
•± Riots/Protests±by±Voters
•± Voter±Intimidation
APPENDIX I:
DICTIONARY FOR
CROWDMAP DATA
Emergency-Related Infrastructure Problems
•± Donation/²undraisers
•± Electricity±Outage
•± Emergency±Messages±and±Info
•± ²ood±Shortage/Needed
•± Medical±Care±needed
•± Relief/Aid±Needed
•± Shelter±Provided
•± Roads±Damaged
•± Shelter±Needed
•± Water±System±Problems
•± Relief/Aid±Provided
•± Medical±Care±Provided
•± ²ood±Provided
•± Water±Provided
Entertainment and Socializing
•± Entertainment
•± Social±Gatherings
•± Where±to±Eat
•± Shopping
•± Hotels
Environmental Issues
•± Animal/²ish±Deaths±±
•± Over-Grazing
•± Pollution
•± Environmental±Issues±Other
Human Rights Abuses
•± Citizen±Attacked/Beat±by±Police/Military
•± Citizen±Killed
•± Disappearance/Kidnapping
•± Excessive±and±Inhumane±Punishment
•± ²orced±Displacement
•± Suppression±of±²ree±Speech
•± Suppression±of±²reedom±to±Organize±and±Gather
•± Threats
•± Torture±(in±custody)
•± Unlawful±Arrest/Detention
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
29
Media
•± Blog±reports
•± News/Media±Reports
•± Twitter±Reports
Medical
•± Disease/Sickness
•± Health±and±Well-being±Information/Programs
•± Health±Emergency/Crisis
•± Health±other
•± Medical±Supplies/Aid/Care±Provided
•± Vaccinations
•± Women’s±Health±and±Reproduction±±(Includes±preg
±
nancy and childbirth)
Natural Disasters-Afermath
•± Death
•± Displacement
•± Houses/Property±Damage
•± Injuries
•± Missing±Person
•± Trapped±in±Home
•± Unstable±Structure
•± Animal±Problems
Natural Disaster--Type
•± Earthquake
•± Fire
•± Flood
•± Hurricane
•± Landslide
•± Snow
•± Tornado
•± Tsunami
Political Organization
•± Boycott
•± Protests
•± Strikes
•± Military±Defections
Prices
•± Prices±of±Food±and±Commodities
Public Goods Provision and InFrastructure (NON-Emergency)
•± Electricity
•± Food
•± Property/Houses
•± Roads
•± Sewer±System±and±Sewage
•± Trash/Garbage
•± Water
Vacant Property
•± Squatters
•± Vacant±Homes
•± Vacant±Lots
War
•± Civil±War
•± Con²ict±with±Foreign±Country
•± Intergroup±Violence±(Sectarian,±ethnic,±religious,±etc.±
conflict WITHIN country)
Cities as Categories
•± Cities±(Code±for±Site±Title!)
30
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
EXPERIENCE
1. Before you used a Crowdmap, how would you describe your level of experience using this (or similar) crowdsourcing technology?
•± No±experience
•± Some±experience±(1-2±times)
•± Moderate±experience±(3-5±times)
•± Considerable±experience±(6-10±times)
•± Great±deal±of±experience±(more±than±10±times)
2. How would you describe your level of experience as a contributor to digital maps?
•± No±experience
•± Some±experience±(1-2±times)
•± Moderate±experience±(3-5±times)
•± Considerable±experience±(6-10±times)
•± Great±deal±of±experience±(more±than±10±times)
3. How would you describe your level of experience in hosting digital maps?
•± No±experience
•± Some±experience±(1-2±times)
•± Moderate±experience±(3-5±times)
•± Considerable±experience±(6-10±times)
•± Great±deal±of±experience±(more±than±10±times)
4. What would you say was the most important reason why you engaged with Crowdmap:
1. I wanted to understand more about Crowdmap.
2. I wanted to create a map with a specific purpose and/or event.
3. I wanted to learn more about Ushahidi.
4. I wanted to demonstrate Crowdmap and/or participatory mapping to others.
5. Other ± please specify: _______________________
5. If you setup a Crowdmap but did not actually use or create a map, please tell us why.
1. I was curious and did not intend to use a map.
2. I wanted to create a map but found it to be too technically challenging.
3. I wanted to create a map but realized it would take too much time.
4. I had security concerns about the collected data.
5. I wanted to create a map but Crowdmap was not the right tool.
6. I did not understand the processes for aggregating data from other sources (SMS, Twitter, email, etc.)
7. Other ± please specify: _______________________
APPENDIX II:
SURVEY RESEARCH
QUESTIONS
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
31
6. If you concluded that Crowdmap was not the right tool, what was the most important reason you reached this conclusion?
•± It±could±not±be±customize±to±my±needs.
•± It±was±too±complex±for±my±needs
•± It±was±too±simplistic±for±my±needs
•± Other±-±please±explain±brieFy±below:
__________________________________________________________________________________________
__________________________________________________________________________________________
7. If you did build a Crowdmap platform, which of these statements best characterizes your experience?
•± I±was±able±to±generate±the±required±public±awareness±of±my±map±to±make±it±as±e²ective±as±I±had±hoped.
•± I±was±able±to±generate±the±required±public±awareness±of±my±map±to±make±it±almost±as±e²ective±as±I±had±hoped.
•± I±was±not±able±to±generate±the±required±public±awareness±of±my±map±to±make±it±as±e²ective±as±I±had±hoped.
8. To the degree I was successful in generating public awareness of my Crowdsource map, I did this mostly by
•± Engagement±with±traditional±news±media±(newspapers,±radio,±television).
•± Engagement±with±community±organizations,±civic±groups,±and±professional±organizations.
•± Engagement±by±social±media,±SMS,±texting,±etc.
9. If answered in the affirmative (options ‘a’ and ‘b’) in question six, I also generated public awareness of the Crowdsource map by:
•± Engagement±with±traditional±news±media±(newspapers,±radio,±television).
•± Engagement±with±community±organizations,±civic±groups,±and±professional±organizations.
•± Engagement±by±social±media,±SMS,±texting,±etc.
FUTURE USE
Please indicate all considerations that apply OR Please indicate the most important consideration
In the future, I would consider using Crowdmap if:
•± It±was±easier±to±use.
•± I±identi³ed±a±need.
•± I±had±more±time.
•± I±felt±more±con³dent±about±the±security±of±contributors.
•± It±had±better±instructions.
•± If±technical±support±was±o²ered.
•± I±would±not±try±to±use±Crowdmap±again.
When getting started with Crowdmap, I found it:
•± Made±sense±and±was±easy±to±use.
•± Was±slightly±confusing,±but±I±³gured±it±out.
•± It±just±didn’t±make±sense±and±I±never±got±it±to±work.
•± Hard±to±³nd±the±training,±documentation.
32
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
DEMOGRAPHICS
Will you please tell us a bit about yourself?
Gender
•± Male
•± Female
Birth Year _______________
Occupation ________________________
Education
•± Completed±secondary±education
•± Some±college
•± College±degree
•± Some±post-graduate±education
•± Post-graduate±degree
In what city, country or region were you living when you used Crowdmap?
____________________
Was this also the location of the event or process for which you used or considered using Crowdmap?
•± Yes
•± No
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
33
34
±
MAPPING THE MAPS: A META-LEVEL ANALYSIS OF USHAHIDI & CROWDMAP
ABOUT THE INTERNEWS CENTER FOR INNOVATION & LEARNING
The Internews Center for Innovation & Learning supports, captures, and shares innovative approaches to communication
through a creative program of research and development worldwide. Founded in 2011, the Center seeks to strike a balance
between local expertise and needs and global learning in order to develop a comprehensive approach to understanding and
catalyzing information exchange.
In Internews’ 30-year history of promoting independent media in more than 75 countries around the world, the last five
years have arguably seen the most changes in the global media and journalism environment. Across all Internews programs,
adoption of cutting-edge technology is integral to advancing the work of the journalists, bloggers, citizen reporters, schol-
ars and others who provide a vital interpretive role for their communities. The Internews Center for Innovation & Learning
deepens and enhances our capacity to link existing expertise to research that helps define, understand and monitor the
critical elements of changing information ecosystems and to pilot projects that apply and test the data, platforms and
digital tools to meet information needs of specific communities. This is far from a solo endeavor. A network of partners,
ranging from technologists to academics to activists is critical to creating and sustaining a dynamic and iterative collabora-
tive space for innovation.
Internews is an international non-profit organization whose mission is
to empower local media worldwide to give people the news and infor-
mation they need, the ability to connect and the means to make their
voices heard.
Internews provides communities the resources to produce local news
and information with integrity and independence. With global exper-
tise and reach, Internews trains both media professionals and citizen
journalists, introduces innovative media solutions, increases coverage
of vital issues and helps establish policies needed for open access to
information.
Internews programs create platforms for dialogue and enable informed
debate, which bring about social and economic progress.
Internews’ commitment to research and evaluation creates effective
and sustainable programs, even in the most challenging environments.
Formed in 1982, Internews is a 501(c)(3) organization headquartered
in California. Internews has worked in more than 75 countries, and cur-
rently has offices in Africa, Asia, Europe, the Middle East, Latin America
and North America.
Internews Washington, DC Office
1640 Rhode Island Ave. NW Suite 700
Washington, DC 20036 USA
+ 1 202 833 5740
Internews Administrative Headquarters
PO Box 4448
Arcata, CA 95518 USA
+1 707 826 2030
www.internews.org
E-mail: info@internews.org
Twitter: @internews
facebook.com/internews