Shopping cart
0 Items | US$ 0.00
II
Acknowledgements
I am deeply grateful to my parents, Gabriele and Dr. Johannes Sieben, who have always
supported me and have given me incredible opportunities throughout my life. I am also
thankful to my two supervisors Dr. Ayúe Göker of the Department of Information Science at
City University London and Professor Frank Webster, Head of Sociology at City University
London, who supported me with invaluable advice on a range of issues. Further I would like
to thank David Corney and Martin Carlos of the Department of Information Science at City
University London, who advised me on technical issues regarding tweet crawling. Lastly I
would like to express my appreciation to Dr. Cornelius Puschmann of the School of Library
and Information Science at Humboldt Universität zu Berlin and Dr. Axel Bruns, Associate
Professor in the ARC Centre of Excellence for Creative Industries and Innovation at Queens-
land University of Technology in Australia, who have graciously shared their tweet datasets
with me.
III
Abstract
This study examines tweet content from key periods of the uprisings in Egypt and Syria of
2011 and 2012, generally known as the "Arab Spring". Some authors and the mainstream
media have suggested that these uprisings were significantly influenced and organised by
Twitter and subsequently referred to them as "Twitter Revolution". Other authors have
strongly opposed this idea and attributed it to self-deception in the light of marvellous inven-
tions of the Western World. They have suggested Twitter was predominantly used as an in-
formation-sharing network. In an effort to contribute data to this debate, this study analyses
tweet content from three different observation periods; two tweet datasets were collected
from other academics and third one was crawled from the Twitter API; this process made use
of the crawling tool cURL and the database software mongoDB.
The combined tweet dataset contained about 1.9 million tweets out of which a sample of
1945 tweets was drawn. This sample was then evaluated in a quantitative content analysis
according to a coding manual. These codes were entered into the statistical analysis soft-
ware SPSS, in which they were also processed.
This study found that in the context of these uprisings, Twitter was indeed used more as
an information-sharing tool and only to a relatively small fraction for organisational purposes.
This result does not negate the possibility of a mobilising effect of that small fraction. A fur-
ther, central result is that almost every second tweet contained a hyperlink and that most of
these lead to visual stimuli.
1
"We had no freedom of assembly in the streets of Cairo, so we assembled in
cyberspace instead."
Egyptian activist
1
1 Introduction
A wave of democracy has swept through Northern Africa and the Near East. While the up-
roar and consequent toppling of their respective dictators was concluded within weeks in
Tunisia and Egypt, Syrians are still fighting, bleeding and dying to this day. The media have
labelled this recent wave of social uproar with the catchy phrase "Arab Spring". In many pub-
lications and certainly in the mainstream media the term Arab Spring often goes hand in
hand with another term: "Twitter Revolution" (Sabadello 2011: 11). This expression first
emerged during the civil unrest in Moldova and the presidential election in Iran of 2009 and
ascribed Twitter a role of growing importance in online and on-the-ground activism; this role
was mainly propagated by mainstream media and bloggers (e.g. Stone and Cohen 2009).
The British author and blogger Andrew Sullivan, who tirelessly twittered about the 2009 pro-
tests in Iran, became one of the frontrunners of the Twitter Revolution advocates; Evgeny
Morozov, renowned for his scepticism towards social media's organisational capabilities,
calls him the godfather of Iran's Twitter Revolution (2009: 1). It was thus probably inevitable
that Twitter's role in the political subversions would become subject of sociological study.
However, it was not only scientists, journalists or bloggers who believed that Twitter was
playing a role in the Iranian uprising. The government of the United States obviously attribut-
ed at least some influence to Twitter when the US State Department asked the platform op-
erators to postpone a scheduled server upgrade so that access to the platform would not be
interrupted (Burns and Eltham 2009: 299; Gaffney 2010: 1).
Once Twitter had been affiliated with revolutionary capabilities for a suppressed populace
for the first time in Moldova and Iran, some were quick to do it again: during the uprisings in
northern Africa and the Near East, specifically during the protests in Tunisia, Egypt, Libya
1
cited in Sabadello 2011, p. 11.
2
and Syria. Subsequently to the media debate, the concept of the organisational and mobilis-
ing capabilities of social media has been put under the scrutinising eye of researchers; some
of whom are more convinced of the importance of Twitter for revolutionary events than oth-
ers. Morozov found that a great number of television networks and newspapers ran stories
advocating how crucial Twitter's role in the Iranian protests was even though the question of
whether or not technology was an essential factor "remains a big unknown" (2009: 10).
Authors such as Stepanova (2011) contend that tools such as Facebook and Twitter had
mobilising effects and functioned as a catalyst to an already tense political and sociological
situation. Stepanova (2011: 3) further claims that "no [...] state, or form of government can
remain immune to the impact of new information and communication technologies on social
and political movements." The political cartoon below accurately visualises this assessment.
Sabadello (2011: 11) on the other side of the argument suggests that the revolutions in the
Arab world would "most likely still have taken place without the Internet" just like other revolu-
tions have in the past.
Picture 1.1 Mubarak vs. Social Media
Source: cartoonaday.com
When one tries to evaluate all these assertions, opinions and discussions on whether or
not Twitter was fundamental, instrumental, a catalyst, helpful or inherently unconnected from
events that transpired, it seems rather striking that very little empirical evidence is presented
3
by most thinkers, bloggers or journalists. To be precise, there is an apparent lack of research
on the content of tweets. What did the people use Twitter for? Did they mainly use it to or-
ganise themselves? Did they use it to spread information? Or was it used for something en-
tirely different? It is thus the explicit aim of this study to shed some light on this issue. I will
endeavour to ascertain what Twitter was actually used for during the uprisings. In order to
accomplish this task, I have collected tweets concerning the Arab Spring and conducted an
empirical content analysis looking at concrete tweet texts and specific metadata. Additionally,
I have analysed two tweet datasets collected by other researchers in early 2011. No matter
the result this study does not claim to ultimately determine whether or not Twitter was in-
strumental in the uprisings but add empirical data to a somewhat hollow discussion.
4
"...let them tweet, and they will tweet their way to freedom"
Evgeny Morozov
2
2 Literature Review and Discussion
As this study is based on an empirical analysis of tweet content and since there is relatively
little literature on this very topic, the literature review section shall primarily focus on studies
that are closely related to this subject. It will further summarise the debate on the influence of
technology (predominantly information and communication technologies such as Facebook
and Twitter) on `revolutionary' events and the so-called Arab Spring itself, namely the upris-
ings in Egypt and Tunisia of 2011 (and the on-going struggle in Syria) but also the protests
spurred after the allegedly forged Iranian elections of 2009.
2.1
Twitter in a Nutshell
Before entering deeper into the debate, it seems necessary to precisely determine what Twit-
ter is, what it can and cannot do. The Twitter platform went online in 2006 and is therefore
one of the younger social media tools when compared to Friendster, Myspace or even Face-
book. By definition, Twitter is a micro-blogging tool written in the Ruby on Rails programming
language and is entirely free of charge to its users (Kavanaugh et al. 2011: 3). It is consid-
ered a micro-blogging tool in reference to the maximum length of messages (so-called
tweets) a user can send, which cannot exceed 140 characters and are often used for "status
updates, news comments, and to `retweet' or repost the messages of other users" (Burns
and Eltham 2009: 298). Every tweet sent by a user is automatically forwarded to their `fol-
lowers', which is a group of people who have decided to subscribe to the tweets of that per-
son. If users decide to forward a received tweet to their own followers, we speak of `retweet-
ing'. Even though Twitter is often referred to as a micro-blogging tool, it is in a very cardinal
way different from a blog. While most blogs are devoted to one single topic or theme, most
Twitter users are not. Hermida (2010) has described Twitter as a form of ambient journalism
2
cf. Morozov 2011: xii. Quote is actually meant sarcastically; Morozov believes the supposed de-
mocratising power stems from wishful thinking on the part of "cyber-utopians".
5
breaking news in almost real-time albeit circumventing gatekeepers at traditional news me-
dia. Circumventing gatekeepers is obviously an equally important benefit of Twitter when the
gatekeeper comes in the form of a governmental authority that blocks or controls traditional
media (Papacharissi and de Fatima Oliveira 2012: 4).
A fundamental difference between most social networks such as Facebook and Twitter is
that Twitter does not require reciprocity, a relationship can be (and in most cases is) one-
directional, meaning that following a user does not require that person to follow back (Kwak
et al. 2010: 1). Kwak et al. in their 2010 study on the entire Twittersphere (which was a lot
smaller then with only about 41.7 million user profiles) found that only about 22% of the users
have a reciprocal connection (meaning they follow each other) and conclude that Twitter is
more of an information sharing network than a social network. A feature employed by Twitter
that is quite central to this study is the use of `hashtags', which are generally speaking
words that are preceded by a pound symbol (#). That pound symbol serves as a form of
metadata tag and results in that tweet being found if one searches for the tagged term in the
twitter search bar. These case-insensitive tags can be clicked and lead to a global search for
tweets using that hashtag with the intention of "facilitate[ing] a global discussion on a topic
beyond a user's follower network" (Lotan et al. 2011: 1376). In part due to this function, Twit-
ter has by now evolved to a point where hashtags are often used and integrated in the cov-
erage of live events (O'reilly and Milstein 2009 as cited in Bruns 2009: 298), e.g. by superim-
posing suggested hashtags on the screen. Papacharissi and de Fatima Oliveira (2012: 2)
find that most of the trending topics on Twitter are headlines of breaking news on sports,
cities or brands forwarded from news feeds of more traditional online news media.
The exponential spread of web-enabled smartphones in combination with applications that
make use of the Twitter application programming interface (API) have contributed to making
twittering possible on-the-go; around 80% of people in the UK logging in, do so via their mo-
bile phones compared to an average of 55% globally (Arthur 2012). By its own statistics,
Twitter has revealed that in May of 2012 it had 10 million active users in the UK alone and
around 140 million active users worldwide (ibid) sending 340 million messages every single
6
day. The total number of registered accounts is a lot higher with some reports estimating up
to 500 million accounts as of February 2012.
3
These figures alone certainly seem to make
Twitter quite an interesting subject for sociological study.
2.2
Twitter Revolution or Revolution by the People?
The on-going key debate on whether or not social media can play an important role in revolu-
tionary events and its organisational capabilities in general has been going on for years. This
debate is most certainly not restricted to the recent uprisings in Tunisia, Egypt, Libya or Syr-
ia; the term "Twitter Revolution" has been attributed to the civil unrest in Moldova in 2009 and
the violent demonstrations after the allegedly forged elections of 2009 in Iran. This part of the
literature review aims to do the entire debate justice.
Up until the uprisings in the course of the Arab Spring, the Middle East was the only re-
gion on the planet where democracy was basically non-existent with not a single democracy
among 16 Arab countries (Diamond 2003: 9 and Ajami, 2012: 1). Did Arabs simply not have
a desire for democracy? Survey data seems to suggest otherwise: In Arab countries just as
many Muslims as non-Muslims support democracy (ibid: 9) and Islam appears to have little
influence on political views (ibid: 10). A 2010 survey conducted by the Pew Research Cen-
ter's Global Attitudes Survey found that a clear plurality of 59% of all Egyptians Muslims
found democracy more desirable than all other system of government (Kavanaugh et al.
2011: 5) a fact clearly backed up by millions of people marching in the streets of Benghazi,
Cairo or Damascus calling out for democracy.
So what role did the Internet play here? Evgeny Morozov holds the opinion that it is just
wishful thinking on the part of people he calls "cyber-utopians" hoping for a new and im-
proved Radio Free Europe that will peacefully bring down dictators at the fingertips of blog-
gers (2011: xii). He adds that these bloggers were mainly Americans who greatly overesti-
mated the role of technology (2009: 10). Further he argues that it is naive to believe that so-
3
cf. http://mashable.com/2012/07/30/twitter-users-500-million/
http://www.mediabistro.com/alltwitter/500-million-registered-users_b18842
7
cial media favour an oppressed population over a totalitarian regime that usually does not
shy away from the most vicious atrocities let alone have any regard for the rule of law (ibid:
xiii). Morozov moreover recognises the fact that dictators have become quite good at using
social media for their own propaganda purposes and against its users (as I will demonstrate
in section 2.4) as cheap surveillance tools (ibid: xiv). He concludes that installing a democrat-
ic government requires more than many cyber-utopians believe and that rioting crowds in
Teheran 2009 are not much different from crowds in Berlin 1989 (ibid: 5); he infers that a
Twitter revolution would theoretically only be possible in a state where rulers are completely
oblivious to social media and have no hand in them, which is obviously a utopia (2009: 12).
Sabadello in his 2009 paper agrees with Morozov that terms such as "Twitter Revolution"
not seldom come from cyber utopians' minds hoping for a better world spurred by technolo-
gy. According to him, these are precarious generalisations and should be avoided. Social
media should rather be considered new tools in an old fight available to both sides.
According to Iosifidis, the `twittersphere' is an "ideal space[...] for initiating public debate
and social change" (2011: 4) and to direct joint action. He further argues that Twitter's capa-
bility of overcoming suppression of free media and general censorship by authoritarian re-
gimes is undeniable (ibid: 7). He refers to Splichal's (2009) account of the Moldovan upris-
ings in which demonstrators allegedly used "social networking website tools like Twitter,
LiveJournal and Facebook" (Iosifidis 2011: 7) to spur and direct the demonstrations as well
as the above mentioned uprisings in Iran. Splichal himself admits that the impact of Twitter in
these events may have been "heavily overstated" (Splichal 2009: 404). No empirical evi-
dence is given for any of these arguments. Contrary to this, the Project for Excellence in
Journalism (PEJ) has gathered quite specific data to back up their claim that "the political
unrest in Iran has demonstrated as never before the power and influence of social media"
(PEJ 2009) by tracking links posted on blogs and Twitter. They found that during the week of
June 15th to 19th 2009 98% of the links posted on Twitter concerned Iran. Those links were
used to spread information or "organize support for those involved in the struggle" (ibid).
While the authors insist that these tweets clearly show cyber activism driven by new media
8
(ibid), they have to admit that protesters may have twittered false information intentionally
and that the potential influence of social media may have been overestimated by the main-
stream media (ibid). It goes without saying that the given numbers are no proof of a direct
causal correlation between what happened online and on the ground, especially since the
location of tweet sources was mostly unknown. According to PEJ (2009) social media also
had quite a stark influence on more traditional news channels since the harsh measures tak-
en against Western journalists lead to an increase of social media topics in mainstream me-
dia with one in every twenty stories dealing with videos or pictures taken by Iranian activists
while the Persian version of the BBC had to rely on user-generated content entirely due to
threats by the Iranian government.
Rich (2011) goes so far as to say that the assumption of the Arab Spring being spurred by
American-built technologies such as Twitter and Facebook was just a truism caused by
Western bigotry about their own fabulous inventions. He further cites CNN anchor Jim Clan-
cy saying that the largest demonstrations in Egypt actually took place on January 29
th
of
2011 a day when the Mubarak government had pulled the plug on the entire Internet (and
mobile coverage) throughout Egypt. So those large demonstrations could not possibly have
been organised or amplified by Twitter or Facebook. Instead the uprising, he maintains, had
very much more to do with the people's poverty, self-respect and inability to nourish their
families than with the simple fact that some of them may have connected through Twitter
(ibid).
Stepanova (2011) contends that the entire upheaval in Tunisia was initiated by a Face-
book group (called "April 6 Youth Movement") controlled by the opposition in which several
ten thousand people responded to a call for action. As mentioned in section one, Stepanova
admits that socioeconomic and political friction had already been so extensive in Egypt that
protests were inevitable, but she also suggests that social media had rallying effects causing
protests to spill over from Tunisia to Egypt. Stepanova does not give concrete evidence for
these assertions. In an effort to corroborate her argument she provides data on overall Inter-
net usage in Egypt being at over 17 million users as of 2011. More recent data by the Inter-
9
national Telecommunication Union, a UN agency, suggests an Internet penetration rate of
35.6%
4
, which equates to about 29.3 million out of Egypt's total population of roughly 82.5
million.
5
This is indeed quite a high amount for an Arab country (Arab average is 29.1%) and
even more so for an African country (continental average is only 12.4% as of 2011), but
compared to developed countries with 70.2% Internet penetration, this is still comparatively
low.
6
The Arab Social Media Report, authored by the Dubai School of Government (2011:16-
17), found that Twitter has a penetration rate of 0.15% (which would equate to roughly 1.2
million people) of the population in March of 2011 but only 131,204 active users in total. In
her argument Stepanova joins Splichal in saying that the role of Twitter in comparison to oth-
er technologies may have been "somewhat overemphasized" (2011: 4). She adds that no
direct correlation between Internet or social media access and likelihood of uprisings can be
proven as Arab states such as Bahrain with a very high Internet penetration rate of 88% and
states with very low penetration such as Yemen all saw mass upheaval (2011: 3). Naturally
all these numbers do not negate the possibility of a mobilising effect through Twitter, but they
surely seem to make it relatively unlikely that a significant proportion of the population could
have been reached through Twitter. It appears more likely that most people heard about the
events through more established channels like TV, Radio or even word-of-mouth (Sabadello
2011: 12).
Gladwell (2010) argues that strong social ties among the revolutionaries are the basis of
any social uproar or political coup d'état. Ties between friends on Facebook or even more so
between followers on Twitter, however, are weak ties as most of them may never have met
or do not keep regular contact. Those ties according to Gladwell are not just weak but
also loose and therefore lack functionality as there is no single, controlling authority. These
lose social media networks are incapable of setting strategies and reaching unanimity.
4
cf. International Telecommunications Union: http://www.itu.int/ITU-D/ict/statistics/
5
According to the World Bank: http://data.worldbank.org/indicator/SP.POP.TOTL?cid=GPD_1
6
cf. .International Telecommunications Union: http://www.itu.int/ITU-D/ict/statistics/
10
Gladwell concludes that networks need to be hierarchical in order to be effective in organis-
ing movements a quality that Twitter clearly lacks.
2.3 Quantitative
Studies
There are several authors who have taken a more quantitative approach to this issue which
makes them more instrumental to this study. Among those, a study done by Gilad Lotan of
SocialFlow, a social media marketing company, and researchers from Microsoft and the Web
Ecology Project, is closely related to this study. Gilad et al. have analysed information flows
during several short periods during the uprisings in Tunisia and Egypt in 2011 by analysing
tweets. Their study, however, primarily focuses on analysing how information flows between
different types of actors while this study primarily puts emphasis on tweet content. In order to
accomplish this, Gilad et al. set out to find near-duplicate tweets sent by different users and
then selected the top ten per cent of the most sent tweets. Out of those tweets they compiled
a list of actors who started those information flows (tweets) and were retweeted at least 15
times. One of the study's more significant findings was that most news on Twitter in those
periods of social upheaval were co-produced by bloggers and activists and to a slightly less-
er degree by journalists (Lotan et al. 2011: 1400). Further findings were that tweets of indi-
vidual activists and journalists were retweeted far less than tweets sent by mainstream media
accounts (ibid: 1398).
Another relevant study was done by Papacharissi and de Fatima Oliviera in 2012. Their
study examines the usage of Twitter as a news-reporting tool. Since they restricted their in-
vestigation to the uprising in Egypt, they only followed the #egypt hashtag over the course of
a month from January to February 2011. For their analysis they collected tweets from the no
longer existent archive service Twatterkeeper and used computer-mediated software to iden-
tify patterns in terms of amount and content of tweets. Papachariss and de Fatima Oliviera
found that tweet content was dominated by news, opinion and emotion and were often hard
to tell apart. Further it became apparent that instantaneity in terms of coverage of events as
they happened dominated over the entire dataset with the subsequent subjectivity that fol-
11
lowed. They also concluded that actors, by tweeting and retweeting, turned single events
swiftly into stories.
Most other analyses rarely show any empirical data on this topic and therefore often seem
speculative rather than based on scientific, replicable data.
2.4
Government Responses to Social Media and Online-Activism
It is often overlooked by journalists and authors alike that social media are not tools solely
available to the rioting populace but also to vicious and bloodthirsty tyrants and some of
them are and have been very adept at using and/or controlling them. The Great Firewall of
China is one prominent example; this wall is a cyber-wall that keeps all undesirable (undesir-
able by the Communist Party that is) digital traffic out of China by blocking certain foreign
websites, filtering domestic ones and monitoring Internet access of individual citizens. I will
here depict how governments in the Arab Spring have influenced Internet connectivity, used
social media and blocked websites.
A precedent was set in 2009 when the Iranian Government under Mahmoud Ahmadinejad
cut connectivity to Twitter for about an hour. But soon they had to realise that people quickly
found ways to circumvent that blockade, e.g. by sending content to users outside the country
via e-mail and asked them to tweet for them. Foreign proxy servers were also heavily used.
The measures taken in Egypt legitimised by the government through the emergency law
that had been in place for over 30 years are unprecedented in their scope in Internet histo-
ry (Sabadello 2011: 13). Dunn (2011) reports that before the protests began, Egypt limited
their efforts to monitoring the communication of individual opposition actors and did not do
much about general Internet infrastructure in terms of controlling or filtering. That all changed
in the months from October 2010 to February 2011. The Egyptian authorities began by selec-
tively shutting down access to websites spreading information on the allegedly forged par-
liamentary election of November 2010 and intermittently blocking mobile phones from lead-
ing political activists. They soon moved to fully blocking websites such as Twitter and Face-
book on January 25th 2011 (Dunn 2011: 18). On January 27th however, they went a step
12
further. Not only text messaging services but also Internet access were blocked entirely
throughout Egypt. The latter was accomplished by ordering all Internet Service Providers
(except a minor one that mainly served the Egyptian stock market) to cease operation (ibid:
19 and Kavanaugh et al. 2011: 5). This seemingly giant task was accomplished in an
astoundingly short time period of only fifteen minutes. Kavanaugh et al. report that 97% of all
outgoing traffic from Egypt was cut off (2011: 9). This was only possible since most of
Egypt's Internet was routed through just four ISPs who obeyed government orders without
challenge (Sabadello 2011: 16). This Internet blockade lasted until the February 2nd and text
messaging was not restored until four days later. In addition to that, voice services were of-
ten disrupted in urban areas on the so-called "Day of Rage" (Kavanaugh et al. 2011: 9).
Dunn reasonably speculates that this unprecedented attack on Internet freedom and services
on which the Egyptian people had come to rely on every day for their social interactions or
business operations, may have been counter-productive as it probably sparked more outrage
at the government by people that were still unaffected, especially apolitical people (ibid).
Shortly before but also after texting services were restored, the Ministry of Interior as well as
the military even forced mobile providers to send a series of text messages to all subscribers
asking them to stop all demonstrations and return home.
Ultimately, Egypt's media strategy seemed ineffective as most activists always found
ways to evade Twitter blockades put up by the government (for instance by calling support-
ers outside the country via landlines and having them tweet for them, or by using Google's
voice-to-Twitter service (Stepanova 2011) to the same effect) and even shutting down the
entire infrastructure was not enough to keep Mubarak on his throne. Also quite ironically, the
largest demonstrations occurred on a day where most of Egypt's Internet connections (and
thus all connections to Twitter and Facebook) and mobile phone coverage were unavailable
(Rich 2011). Stone and Cohen (2009) refer to Harvard Law School professor and expert on
the Internet, Jonathan Zittrain, saying that in purely technical terms, censoring Twitter is an
extraordinarily complicated challenge due to its many input sources (e.g. desktop pcs, mobile
phones, applications, etc.) and even more channels for posts to be published, thus shutting