The Role Of Social Media

Published Date: 02 Nov 2017

Today, mobile technology and social media are integral parts of the ordinary person’s computing experience. For example, Windows 8 directly incorporates social media into the operating system and is designed to work on both desktop computers and mobile devices. Mobile technology and social media are becoming a large part of computing technology. With this in mind, I decided to explore a topic related to social media in my STS research paper. In a more fundamental vein in terms of computing technology, my technical report describes my group project: designing an embedded processor.

My technical project was somewhat open-ended. Ideally, Professor Stan would have liked everyone in my capstone class to design a particular five stage pipeline processor, optimize it, implement a cache, and design an embedded system for the processor to be situated within; then load it to a programmable FPGA board. No other team achieved this goal perfectly, however, and neither did mine. Since processor design was still new to us, we decided to hedge our bets by developing two different kinds of processor simultaneously! At the end of the project, my team had produced a partially working five stage processor, and a single cycle processor that worked in simulation. We had also produced a memory light game that ran on the single cycle – in simulation at least. We ran out of time, but our preliminary testing showed that the single cycle worked partially if not completely on the FPGA board. This project will leave a lasting impression on me. I finally learned how to write hardware description language code effectively.

For my STS research paper, I explored a subject that was very interesting to me: What is the relationship between social media and revolutionary action? In particular, I explored how social media was used in the uprisings in Tunisia. I watched Youtube videos and scoured Tunisian Facebook pages, and I read academic papers, blogs, and books. Although, my paper’s most important contribution is in simply telling the story of social media use surrounding the Tunisian uprisings, I did arrive at one provisional conclusion that I think merits further research. I asked if social media could tease people out on the streets in protest before there is the on-the-ground organization needed to fill the power vacuum created when the existing regime steps out of power.

Thank you to my friend Matthew Lawrence for letting me spend the day at your quiet apartment, working on my STS research paper. Thank you also to my brother Wesley and mom and dad for reading my paper at various stages and offering suggestions for improvement. Thank you to my group mates Torrest Bates and Gautham Kurup, and my two project advisors Ronald Williams and Mircea Stan. Thank you for your help with the Prospectus, Professor Brown. Thank you, especially, Professor Pfaffenberger, for giving me extensions, encouragement, and editorial help, as well as for inspiring me with your enthusiasm for stories surrounding technology.

The Role of Social Media in the Tunisian Uprising

A Research Paper

in STS 4600

Presented to

The Faculty of the

School of Engineering and Applied Science

University of Virginia

In Partial Fulfillment

Of the Requirements for the Degree

Bachelor of Science in Computer Engineering

By

Daniel Charles Toler

May 11, 2012

On my honor as a University student, on this assignment I have neither given nor received unauthorized aid as defined by the Honor Guidelines for the Undergraduate Thesis Project.

Signed________________________________________________________________________

Approved __________________________________________________ Date _______________

Bryan Pfaffenberger, STS Advisor

Abstract

This paper will present a brief history of the use of social media during the recent uprisings in Tunisia, roughly in three stages: before the death of Bouzazi, shortly after his death, and shortly before then President Ben Ali's departure from the country. I will also engage in a discussion about the relationship between social media and revolutionary action in Tunisia, along the way drawing attention to a few parallels between social media and past technologies used in revolutions, and pointing out that social media use did not always favor the activists- the repressive government used it to their advantage also. Finally, I will reach the conclusion that social media was used to destabilize the existing power structure, but failed to provide an on-the-ground organization to replace the existing government.

Introduction

In this paper, I aim to show that social media played a significant role during the Tunisian Riots, and to investigate the role that it played. I have divided the use of social media in the movement into three phases. Social media was used before the beginning of the protests to keep tabs on government spending and to start a grassroots movement opposing the government. It was used during the protests, to spread imagery of protests and atrocities committed by security police. It was also used later in the movement to provide critical commentary on government propaganda, and to combat rumors that threatened the movement. The first part of my paper will describe the history of social media during the Tunisian Riots, broken roughly into these three phases.

In the second part of my paper, I will examine the relationship between social media and revolutionary action in Tunisia. I will ask the questions of why social media was used and how it was different from some other forms of revolutionary communication technology. I will also talk about some of the strengths and weaknesses of social media used for activism. I will not arrive at a definitive conclusion about the features of social media in Tunisia, nor will I try to answer the big question of whether the revolutions would have happened without the aid of social media. But, I will reach the conclusion that social media use did not always favor the activists and may have actually hurt their long-term cause by creating a power vacuum when there was not yet sufficient on-the-ground infrastructure to replace the existing regime.

This topic is interesting to me, because I have been observing the increasing ubiquity of Facebook in our daily lives. More and more products, in the real world, are being tagged with the phrase "like me on Facebook." For example, I was reading a Taco Bell cup the other day that encouraged me to "join the conversation" surrounding the new Doritos taco. The cup was decorated with text blurbs, Facebook comments people had made about the new taco on the company's Facebook page. Increasingly, the lines are being blurred between the real world and the Facebook world. Indeed, Mark Zuckerberg has cheerfully predicted that eventually people will be wearing devices that automatically update their online status without any action on their part being required (Kirkpatrick, 2010 p. 314).

The corporate philosophy of Facebook involves making the world more "open" and "connected" (Zuckerberg, 2009). Privacy scandals have caused a backlash against the company; still, Facebook executives seem to continue to hope that people's aversion to sharing personal data will eventually go away. They seem to envision a world where there will be no privacy concerns, because everyone will be sharing everything, a world where increased transparency among the world's members will cause larger groups of people to act like a village. A village is close-knit enough to have a gift economy where selfishness is seen and discouraged and generosity is rewarded. According to the book, The Facebook Effect, this is a strong ideological belief of the organization's leader (Kirkpatrick, 2010, p. 287).

I found that many of the things that were being said about Facebook were also being said about other forms of social media. Social media is still fairly new, and surely some of the claims about its positive influences must be over-hyped. Still, the idea that a technology like social media could be rapidly and significantly changing the cultural landscape and even power structure of our globe was intriguing. It is a startling thought, though quite possibly a false one. But it seems to have currency with many people; for example, the most viral video to date- Kony 2012 -popularizes the idea (Invisible Children Inc., 2012). And the idea is unsettling to me. If the cultural landscape of the world is changing, I would at least like to know about it.

The so-called Arab Spring riots have caused Western media to question whether or not widespread use of social media is increasing the likelihood of democracy taking root in repressed countries. This is an example of a culture-changing feature of social media that some people claim for it: inherently democratizing. Tunisia was the first country to rise up, and this once calm, stable country sparked a whole chain of uprisings throughout the Middle East. I chose to study Tunisia, because it was the first, and because social media clearly played a role.

History of the Revolution in Tunisia

The generally-accepted beginning of the Tunisian Revolutions has little to do with social media. The birthplace of the movement is Sidi Bouzid, a small town in Tunisia. And today, there is a simple grave there, visited occasionally by pilgrims, which contains the body of Mohammed Bouazizi, a simple vegetable peddler. Bouazazi set himself on fire on December 16, 2010 as an act of desperation, sparking a Tunisian uprising that eventually led to the ouster of then-President Ben Ali and to a string of uprisings in repressive countries throughout the Middle East. He is admired as a martyr in Tunisia, (Ryan, 2011b) although according to his family his only aspiration was to own a pickup truck, so that he could more easily deliver his wares. Allegedly, on the day he set himself on fire, a police-woman confiscated his wares and required a ransom of 10 dinar (one day's wages). When he offered to pay the fine, the woman humiliated him by slapping his face and insulting his dead father. She also refused to give him back his wares. Denied audience at the municipal government building after attempting to appeal his mistreatment, Bouazizi returned with a container of flammable liquid, which he splashed all over himself and then lit. Soon after Bouazazi's self-immolation, locals sprang to the street to protest at the government building. Many frustrated jobless college graduates and other jobless people lived along a strip of houses where Bouazizi lived (Abouzeid, 2011). Still, given the widespread discontent over perceived government corruption and insensitivity to ordinary people’s needs, the riots probably could have begun in any town. Videos of the protests in Sidi Bouzid show people attempting to scale the gates of the government building. The videos, released through social media sites like Youtube and Facebook, later proved influential in starting protests around the country, according to Al Jazeera, whose media team found the videos while trolling the web (Ryan, 2011c).

However, the role of social media in the revolutions began before Bouazizi's self-immolation (and the string of imitation martyrs who self-immolated throughout the Middle East after his example). It began with the subversive online activities and propaganda of the protest group, the Takriz. The name Takriz is a street-slang profanity, and its members are known for being raucous and "dirty" (Pollard, 2011). At the 1999 Tunisian soccer cup, two Takriz members observed the eruption of fan violence that ended up killing some and injuring many. The violent soccer fans involved were known as Ultras, and Takriz began actively recruiting them for protest in Tunisia. According to James M. Dousey, "The mosque and the soccer pitch have been the only release valves for anger and frustration among the young under autocratic Middle Eastern rule" (Dorsey, 2011). The Takriz recognized, however, that the violent energy that the Ultras released at soccer games could be channeled into political protests. Similarly, they appealed to street youth, using the language of the street, slang and profanity. In the 2000's, having strongly alienated themselves from the government and even the middle class, who saw them as a group of thugs, the Takriz began moving into exile, fleeing the country, while the government tracked down who it could, imprisoning and torturing. The Takriz moved to the internet, anonymously maintaining websites with anti-government propaganda. The Takriz attribute to their own organization, and credibly, much of the early momentum of sentiment turning against the Tunisian government.

Offshoots of the Takriz also created propaganda efforts against the Tunisia government. Riadh "Astrubal" Guerfali was a law professor in France who was radicalized by the Takriz. As a propaganda stunt, Guerfali modified the famous Apple 1984 football advertisement to replace the face on the screen with the face of then-President Ben Ali. The original advertisement showed workers walking in lock-step towards a central room where they received the latest brain-washing from their leader - a face on a screen. Finally, an attractive woman runs into the room, swings a metal hammer around her head several times, and lets it loose at the screen, shattering it into fragments. The advertisement symbolized the personal computer releasing people from the bondage of the corporate computing structure. This appeal was powerful at the time, because there was a belief that personal computers would be freeing and empowering to individuals. By modifying this video to depict Ben Ali, Guerfali was clearly saying that Ben Ali was a dictator, that his power could be broken, and also by putting it in the context of a technology commercial, he seemed to be saying that technology would be part of the solution. In any case, this video represents an important example of propaganda being created and released on social media through the influence of the Takriz. It received a moderate number of hits on Youtube (13,127 hits as of May 8, 2012) and was actively discussed by prominent activists around the world.

Another video released by Guerfali was arguably more influential. Collaborating with a Tunisian exile, Sami Ben Gharbia, Guerfali scanned plane-spotter websites to identify images of the presidential plane at various airports around the world. A video montage (Nawaat, 2006), which superimposed airplane images on a Google map animation, effectively showed where the plane had been. The suspicion, aroused by the videos, that the first lady, Leila, was splurging on shopping sprees around the world was something that enraged many Tunisians, who already felt that the President and his family were corrupt and that the distribution of wealth in the country was unfair.

As I have described, social media was used by activists early on to create a grass-roots movement against the Tunisian government that was actually trans-national. And social media sites allowed for propaganda videos to gain popularity. Finally, primarily due to tensions over the economic situation, Bouzazi's death sparked a local protest, which due to videos of the protests spreading through social media, helped to inspire other protests around the country.

The next phase of the history of the revolutions began in the town of Kasserine. It was there that government security forces began using deadly force against civilians. The massacres at Kasserine were documented by bloggers after the fact and by some surviving mobile phone footage, although it was also reported that users of mobile phones were targeted by snipers there. Kasserine had more victims than any other city during the revolution, and according to Yasmine Ryan (an Al Jazeera correspondant), the massacres "turned a regional uprising into a nationwide revolution" (2011a). Just as the concept of self-immolation is abhorrent to most Americans, the behavior that led to the massacres in Kasserine seems equally bizarre to the Western viewpoint. What happened was that initially two young men, Mohamed Mbarki and Slah Dachraoui were killed by security police. Following their deaths, 200 people violated security force stipulations and formed a funeral procession to honor them. Police fired on the procession, resulting in more deaths. Another illegal funeral was formed to honor one of these victims. The police opened fire on this funeral as well, resulting in more deaths. Videos from mobile phones show mad rushes of brutally injured victims into a hospital, women howling in the background. Some of the footage purported to show protesters being shot by security forces. The roll of martyrs increased, and videos and images of atrocities began circulating on the internet, arguably increasing the momentum of the movement. The Facebook page "Tunisian Riots: English Information," for example, was one page that posted various media floating around the internet surrounding the uprisings (See Figure 1 on next page).

Figure 1

Social media played a new role towards the end of the revolution. According to Delany (2011, n.p.):

As the Revolution came to a head, the president prepared to flee the country and the police began to pull out of the streets, simple chaos became the greatest danger. Here again, social media channels gave people a way to organize themselves to protect their neighborhoods and stop the spread of destabilizing rumors. Some gathered on Facebook to form teams to clean up streets and shops, others organized to ration out food and bread. Meanwhile, neighborhood watch groups relayed information on snipers and armed militia groups or spread the word about looters so that they could be intercepted and thwarted.

Online channels helped people fight more insidious enemies as well: rumors and disinformation. Claims might spread of massive shootings in a neighborhood, for example, but people in the area could pipe in and say what was really happening. In other cases, rumors of poisoned water or cutoffs of electrical power might threaten to spark a public panic, but again Facebook and text messages let people pass along the truth. Overall, online social tools helped activists counter those who were trying to terrorize the population, helping to calm the entire situation down — they spread the message that people were helping to keep things in control.

Finally, on Friday, January 14, 2012, after about a month of protests, President Ben Ali fled the country with his family and took sanctuary in Saudi Arabia (BBC, 2011). His prime minister, Mohammed Ghannouchi assumed power as part of a transitional government, a move that worried many Tunisians (Walt, 2011). Eventually, it became clear that there was very little that was new about the new regime. According to Lina Ben Mhenni,

"The security system, and by this I mean the Ministry of the Interior and its police, has not changed at all. The police continue their violence against the people. Arbitrary arrests are carried on and torture is still practised[sic] inside police stations and prisons as has been shown by the reports of different human rights NGOs and associations. After a few weeks of revolutionary euphoria, Tunisia is once again a police state. Take a walk on the main streets of the capital and you will be shocked by the presence of the police. Police officers who apologized to Tunisians after 14 January are back to the old verbal and physical harassment" (Mhenni, 2010).

So the story of the Tunisia uprisings seems to end sadly for now. Most Tunisians are still dissatisfied with their government, which makes the term "Arab Spring" inaccurate when talking about Tunisia. The people of Tunisia successfully ousted their President, but it does not seem that they had much say in who replaced him.

In conclusion, through the activities of the Takriz and other early activists, social media was used to help create a grass-roots movement opposing the current government. During the riots, mobile phones were used to capture video and images that were then transmitted through social media, helping to bring awareness to the movement and to inspire additional protests. By the end of the movement, social media was being widely used as an alternative news source, combating rumors and threats by coordinating local efforts.

How Social Media Was Compromised by the Tunisian Government and How The Internet Allowed the West to Become Involved With Tunisian Politics

On July 5, 2010, before the revolution had begun, Tunisian blogger Slim Amamou noted that people were reporting strange-looking Gmail login pages. It seemed that a widespread phishing operation was occuring in the country. Mr. Amamou eliminated the possibility that the phishing problem could be the result of local malware on machines, and he collected data on the people affected. He found that the problem occurred both at home and on work computers, and when it happened at work, it happened to everyone in the office. At first, he thought it could be DNS poisoning, in which a hacker tricks a server into feeding up a phony web page instead of the real one. But, in fact, the situation was much worse. As he noted:

"there was something fishy going on : the IP address [for the Gmail login page] was correct!... To make such an attack, it takes full control of the Tunisian network, from the wires to the HTTP protocol. Those hackers were owning the whole country" (Reporters Without Borders, 2011).

Mr. Amamou monitored his computer's port 443 (the secure HTTP port) and found that it was blocked for 5 minutes every two hours each day like clockwork, during which time the phishing took place. When the secure https port was being blocked, people were forced to use a non-secure Gmail page, and from there they were diverted to a fake Gmail login page written using EasyPHP, an application for creating programs that run on a server. That page was sending their login information using a weak form of encryption in the form of a GET command. That is, the website would ask the server to retrieve a non-existent web address that in reality contained an encrypted form of the user's login information. Only someone who had access to the internet servers in the country would be in a position to harvest that information. According to blogger Steve Ragan, the government agency, ATI (Agence tunisienne d'Internet), the source of "all... privately held Tunisian ISP's" and "the main source of Internet access in the country" (Ragan, 2011a) was in such a position. He stated: "The likelihood that a group of criminals compromised the entire Tunisian infrastructure is virtually nonexistent." It is reasonable to assume that the Tunisian government was attempting to gain access to Gmail accounts.

Similar phishing problems were detected on login pages for Yahoo and Facebook, prompting Facebook's Chief Security Officer Joe Sullivan to implement what he said was a politically neutral technical security fix. In five days, solutions were released for all of Tunisia. Tunisian requests for Facebook were all routed to an HTTPS server. And "anyone who had logged out and then back in during the time when malicious code was running would be asked to identify... friends in photos to complete an account login" (wconeybear, 2008[?]).

On July 5, 2011, the hacktivist group Anonymous released a video on Youtube (See Figure 2). This was exactly one year after Amamou's blog posting, a coincidence that might be dramatic flair. A synthesized computer voice narrates the video:

"Dear citizens of the world, the time for truth has arrived – a time for freedom and transparency – a time for people to express themselves freely, and to be heard from anywhere across the world. Yet, the Tunisian government has decided it wants to control this present with falsehoods and misinformation and restrict the freedoms of their own people..." (Anonymousworldwar3, 2001).

What follows is a recruitment video and a chilling declaration of cyber-war, in which

Anonymous promises to target any organization involved in censorship in Tunisia. In the days following the beginning of "Operation Tunisia," Anonymous members participated in DDoS (Distributed Denial of Service) attacks against various government websites. Anonymous also wrote MonkeyScripts designed to strip away JavaScript code that the TSA seemed to be injecting into web forms to steal login information (Ragan, 2011a).

A crackdown followed Anonymous' activity. Hamadi Kaloutcha, a blogger activist, was arrested along with revolutionary rapper El General (Hamada Ben Aoun), whose music video "President, your people are dead" circulated widely on Facebook. Slim Amamou, who did investigations into the password hijacking, went missing, as did web activist Azyz Amamy (Reporters Without Borders, 2011). Eventually, all four of them were all released ("The South African," 2010; Anderson, 2011).

Discussion

Activists have always used the communication technologies available. Often communication technologies have "externalities," however, that is, unexpected political or social aspects. At the very least, the type of technology used is part of the story of a given revolution. Cost of publication, accessibility for authors, accessibility to readers, amount of imagery, possibility of encryption, and other features can be expected to influence the way a technology fits into the larger picture of a revolutionary movement. Social media is widely becoming the communication tool of choice for revolutionaries, and it was certainly influential during the Tunisian uprisings. So, what are the features of social media? In this section, I will discuss some of the features of social media as a revolutionary tool.

Social media has the potential for misinformation but also for providing a balance to government propaganda. It has some of the features of the early newspaper publishers in France after the Revolution. The press in that time published propaganda of various kinds, outright lies and misinformation, caricature, over-simplification- leading some at the time to say that the free press was actually an enemy of liberty (Darnton and Roche, 1989, p.164). Social media, such as blogs, allow anyone the ability to (on many sites freely) publish on any topic. The vast quantity of material on the web that is not proofread provides a huge potential for misinformation. On the other hand, the fact that anyone can publish can prevent a monopoly on information by an authoritarian government. The features that make a free press wild, unruly, and potentially dangerous, make user-created content on the web wild, unruly, and potentially dangerous. But at the same time, they make it a place where one can cross-check multiple sources to try to arrive at the truth about a subject.

Social media allows for the embedding of images, visuals, and sounds. This is important, because imagery has always had revolutionary significance. Populist movements often arise among the uneducated, which makes a non-literary communication form desirable. But there is also truth to the saying that "a picture is worth a thousand words." Imagery has the potential to convey an accurate visual impression unlike other forms of communication. During the French Revolution, for example, imagery was often circulated in the form of prints (Darnton and Roche, 1989, p. 230). Numerous prints were made of the Storming of the Bastille. The figures in some of the drawings were, admittedly out of strict visual perspective, but the importance of the engravings were that they showed a significant moment in the life of the Revolution, and made you feel as though you had seen it with your own eyes- the next best thing to being there. Social media, similarly allows for protesters to share the sights and sounds of the revolution with each other, which increases a feeling of solidarity and empathy.

Social media also helped the Tunisian revolution take on trans-national dimensions. Prominent blogger Lina Ben Mhenni wrote in three different languages including English, making her blog posts readily accessible to Western media. A large number of tweets came out of Tunisia during the revolution, and anyone who wanted to try to influence the politics in Tunisia could help monitor social media for government agents or could comb through blogs and tweets and supply information to online activists. In addition, they could use their networking skills to coordinate a response if an activist went missing- that is, did not report back online at a given time. The most dramatic example of this trans-nationalism was the case of the Anonymous hacker war against the Tunisian government. Winston Churchill said about dictatorship in Nazi Germany in 1938, "such a state of society cannot long endure if brought into contact with the healthy outside world" (Churchill, 1938), and this is exactly what social media such as blogs may be doing: bringing people into contact with the outside world.

Finally, social media takes advantage of the all-the-time-connected culture that has arguably arisen around the mobile phone, e-mail, and social media. For this reason, there is a speed and immediacy to social media that is unusual with communication technology. Every day more and more people world-wide are becoming connected via social media, and they are sharing information at a rate that has caused the founder of Facebook, Mark Zuckerberg, to claim that observing the exponential trend of online sharing one can predict that eventually people will be wearing devices that automatically update their status information. Diffusion of information through a social media network can be rapid if the context is a topic that is important to a group of people. For example, the number of friends on Facebook per user was on average about 120 (Marlow, 2012) on March 9, 2009, which implies that Facebook networks are weak (not everyone you "friend" is your friend), but it also implies that your potential audience is large, as activity on the site is often defaulted to automatically share. A comment on a photo can instantly give it the attention of 120 possible viewers. At least one study has described the widespread practice of compulsive social media use as addictive (Flacey, 2012). But the addictive nature of sites like Facebook as it appeals to the social urge is effective at ensuring that relevant content propagates through it quickly.

Some people have argued that social media is not a very good choice for revolutionary communications. They have argued that the use of the commercialized social media sites is like "talking in a McDonalds across from the police" (Askanius and Nils, 2010). Facebook, for example, has a policy that requires that all users use their real name as their identity. This policy may be changing as Facebook continues to be used in more protests around the world, and as it becomes more evident that protest leaders could use the protection of anonymity. But I personally think this is unlikely given the broader Facebook corporate philosophy of a more "open and connected world." This is a vague slogan that seems to be a coherent touchstone for Mark Zuckeberg, the founder, and it has as one of its corollaries, the concept that every Facebook user must use his real name in order to promote "openness". I suspect that Zuckerberg, who has always been stubborn about his central tenets, will continue to insist on real names, to prevent Facebook from crumbling into a site like MySpace. The uninhibited nature of MySpace, for example, which does not require the use of real names, has resulted in it eventually losing the market. The foundation of using real names on Facebook seems to have enabled it to become well integrated with existing offline relationships. But why do activists use Facebook then if they are required to use their real names? Some activists turn to social media at first because it is simply familiar to them. But more importantly, most people who are using social media are on Facebook, which dominates the market. If you want to reach people that you do not have an offline relationship with, Facebook gives the broadest possible audience. Also, the fact that so many people are using Facebook for non-revolutionary, entertainment purposes may protect activists from the government simply shutting down the social media site. An oppressive government may hesitate before potentially radicalizing a generation of young people who want their internet access (Zuckerman, 2008). But, the problem of protecting the identity of protesters is a prickly one. And it is a problem, since, repressive governments sometimes hack and socially engineer accounts, as was seen in Tunisian, forcing online activists to be wary and monitor constantly for "government stooges" (Pollock, 2011) posing as sympathizers.

Skeptics of social media's impact like Malcolm Gladwell are quick to point out that social media's strength is that it employs a network of weak connections, something that makes it inadequate at inspiring people to do high-risk activism (Gladwell, 2010). According to him, this form of social structure is resilient and effective at mobilizing people who do not know each other very well to take actions that do not inconvenience themselves very much. He argues that the past form of activism, the form of activism that was used during the Civil Rights sit in protests, for example, involved hard connections, offline relationships, and discipline. In addition, he points out, protests of that kind required a hierarchical command structure. For example, during the Montgomery bus boycott, a coordinated group of local black churches maintained a free alternative bus-line on a prompt timetable. In summary, he claims, social media can create a stir, but it cannot be used to motivate strangers to take real risks on the streets. This opinion existed, interestingly enough, in Tunisia. In spite of the fact that blogger Lina Ben Mhenni traveled to the sites she blogged about, she received a good deal of criticism and even death and rape threats from people outside of her blogging circle while being considered for the Nobel peace prize. She has repeatedly deflected praise away from herself, while emphasizing the necessity of the protesters on the ground. This sounds like an echo of Wael Ghonim when he appeared on Al Jazeera (Mackey, 2011), eyes bloodshot, soon after his release from prison, where he was interred for starting a Facebook Page "We Are All Khaled Said" that ignited the uprising in Egypt. "I am not a hero, I only used the keyboard, the real heroes are the ones on the ground," he said with emotion. We should take notice that prominent online activists push away the title of hero, impatiently. It is clear that a protest does not exist unless people are on the ground protesting.

Still, social media did play an important role in coordinating, informing, propagandizing, and creating empathy. I do not think it is possible to dismiss the fact that social media played a role. First of all, the fact that social media is especially good at inspiring low-cost activism among strangers does not prevent it from being a factor in coordinating and even sparking protests. For activists in the Takriz group, as an example, social media was something that they gradually adopted after many years of strong offline relationships, and was used for communication and coordination. Social media reinforced offline relationships. But the point that I think Gladwell would find more interesting is that social media may directly increase the likelihood of a revolution taking place. It may provide a mechanism for letting people know how much popular support there is for a movement. Social media cannot produce a revolution that does not already have popular support. However, it can prevent government controlled media from convincing a population that there is not support for a movement when there really is (Kiss and Rosa-Garcia, 2011). It is in this way that growing consensus on a network of weak connections may spill over onto the streets in real protest. Finally, social media allowed for the sharing of information and imagery in a way that was rapidly accessible to a large segment of the population. It is very difficult to know whether or not the riots in Tunisia would have occurred if social media had not been available. What I think is certain is that the revolution would have taken a slightly different form, and here I think Gladwell's observation is salient again. At the end of the protests, President Ben Ali left the country, but he was immediately replaced by his Prime Minister, a move that was disliked by the vast majority of people in the country. Maybe social media can harness the power of democracy (the kind that Plato talks about) to force change as long as the objective is simple enough that everyone can agree on it. "Out!" [to President Ben Ali] was the rallying cry of the Tunisian Revolution (Mhenni 2011). This was a simple enough slogan to be agreed upon by most people who, in general, viewed the regime as being corrupt, power-hungry, and repressive. However, perhaps Gladwell was right, and to replace the old system with a new one would have required a more articulated manifesto with specific mandates for change and coordinated manpower to enforce those changes. In other words, maybe what was needed was the product of a disciplined, hierarchical social structure with chains of authority. Perhaps social media teased people out onto the streets but did not provide a mechanism for creating real change.

Conclusion

Social media played a significant role throughout the history of the uprising. Many Tunisians were already using social media before the uprisings, which made it an apparently obvious choice for activists. Revolutionary media posted by bloggers, artists, and ordinary people stirred emotions and seem to have inspired riots. Social media gave people an alternative news source to the government controlled media. It was used to help orchestrate a movement that destabilized the existing power structure. However, it did not provide an on-the-ground organization for replacing the old regime. Clearly, social media was not just a force for positive social change. Even Tunisia's government, which was not that technologically sophisticated (e.g. compared to most European countries, the United States, Japan, and China), was able to wage a campaign to crackdown on dissidents by stealing passwords and user identities, and social engineering people's online networks.

Works Referenced

Abouzeid, Rania. (2011). Bouazizi: The Man Who Set Himself and Tunisia on Fire. Time Magazine. Retrieved from http://www.time.com/time/magazine/article/0,9171,2044723,00.html

Amamou, Slim. (2010). Mass Gmail phishing in Tunisia. Global Voices Online. Retrieved from http://advocacy.globalvoicesonline.org/2010/07/05/mass-gmail-phishing-in-tunisia/

Anderson, Nate. (2011). Tweeting tyrants out of Tunisia: the global Internet at its best. Ars Technica. Retrieved from http://arstechnica.com/tech-policy/2011/01/tweeting-tyrants-out-of-tunisia/

Anonymousworldwar3 (2011 January 5). ANONYMOUS – OPERATION TUNISIA – A Press Release. Retrieved April 23, 2012, from http://www.youtube.com/watch?v=BFLaBRk9wY0

Askanius, Tina; Gustafsson, Nils. (2010). Mainstreaming the Alternative: The Changing Media Practices of Protest Movements. 2(2): 23-41.

BBC News. (2011). Tunisia: Ex-President Ben Ali flees to Saudi Arabia. BBC News. Retrieved from http://www.bbc.co.uk/news/world-africa-12198106

Chrisafis, Angelique. (2011). Tunisia's most influential bloggers prepare for historic elections. The Guardian. Retrieved from http://www.guardian.co.uk/world/2011/oct/22/tunisian-elections-bloggers

Churchill, Winston. (1938). The Light are Going Out. Retrieved on April 26, 2012 from http://teachingamericanhistory.org/library/index.asp?document=1910

Darnton, Robert; Roche, Daniel. (1989). Revolution in Print. Berkeley and Los Angeles, California: University of California Press.

Delany, Colin. (2011). How Social Media Accelerated Tunisia's Revolution: An Inside View. Huffington Post. Retrieved from http://www.epolitics.com/2011/02/10/how-social-media-accelerated-tunisias-revolution-an-inside-view/

Dorsey, James M. (2011). How Egyptian and Tunisian youth hacked the Arab Spring. The Turbulent World of Middle East Soccer. Retrieved from http://mideastsoccer.blogspot.com/search/label/Tunisia

Flacy, Mike. (2012). Study: Is Facebook more addictive than alcohol or cigarettes?. Digital Trends. Retrieved from http://www.digitaltrends.com/social-media/study-is-facebook-more-addictive-than-alcohol-or-cigarettes/

Gladwell, Malcolm. (2010). Small Change: Why the Revolution will not be tweeted. The New Yorker. Retrieved from http://www.newyorker.com/reporting/2010/10/04/101004fa_fact_gladwell

Invisible Children Inc. (2012). KONY 2012. Retrieved April 26, 2012 from http://www.youtube.com/watch?v=Y4MnpzG5Sqc

kapdkjumb. (2011). el general, the voice of Tunisia, english subtitles. Retrieved April 24, 2012, from http://www.youtube.com/watch?v=IeGlJ7OouR0

Kirkpatrick, David. (2010). The Facebook Effect: The Inside Story of the Company That is Connecting the World. New York, NY: Simon & Schuster.

Kiss, Hubert Janos; Rosa-Garcia, Alfonso. (2011). Why do Facebook and Twitter facilitate revolutions more than TV and radio? Munich Personal RePEc Archive Paper No. 33496, posted19 September 2011.

Macko, Robert. (2011). Updates on Day 14 of Egypt Protests. The New York Times. Retrieved from http://thelede.blogs.nytimes.com/2011/02/07/latest-updates-on-day-14-of-egypt-protests/#google-executive-and-blogger-gives-emotional-interview

Madrigal, Alexis. (2011). The Inside Story of How Facebook Responded to Tunisian Hacks. The Atlantic. Retrieved from http://www.theatlantic.com/technology/archive/2011/01/the-inside-story-of-how-facebook-responded-to-tunisian-hacks/70044/

Marlow, Cameron. (2009). Maintained Relationships on Facebook. Retrieved from http://www.facebook.com/note.php?note_id=55257228858

Mhenni, Lina Ben. (2010). Tunisia: Unemployed Man's Suicide Attempt Sparks Riots. Global Voices Online. Retrieved from http://globalvoicesonline.org/2010/12/23/tunisia-unemployed-mans-suicide-attempt-sparks-riots/

Mhenni, Lina Ben. (2011). Tunisian elections: Beware, beware, my hunger and my anger. The Guardian. Retrieved from http://www.guardian.co.uk/commentisfree/2011/oct/22/tunisian-elections-2011-tunisia1

Nawaat. (2006). Dans la tete d'Aziza.... Retrieved April 24, 2012, from http://www.youtube.com/watch?v=GsF8qQmLYo0&feature=player_embedded

Pollard, Ruth. (2012). Facing up to tyranny: how a girl and her blog helped change the world. The Sydney Morning Herald. Retrieved from http://www.smh.com.au/technology/technology-news/facing-up-to-tyranny-how-a-girl-and-her-blog-helped-change-the-world-20120120-1qa4g.html

Pollock, John. (2011). Streetbook: How Egyptian and Tunisian youth hacked the Arab Spring. Technology Review. Retrieved from http://www.technologyreview.com/web/38379/

Ragan, Steve. (2011a). Anonymous offers support to Tunisian protestors (Update 2). The Tech Herald. Retrieved from http://www.thetechherald.com/articles/Anonymous-offers-support-to-Tunisian-protestors-(Update-2)/12403/

Ragan, Steve. (2011b). Tunisian government harvesting usernames and passwords. The Tech Herald. Retrieved from http://www.fastcompany.com/1715575/tunisian-government-hacking-facebook-gmail-anonymous

Reporters Without Borders. (2011). Wave of Arrests of Bloggers and Activists. Reporters Without Borders. Retrieved from http://en.rsf.org/tunisia-wave-of-arrests-of-bloggers-and-07-01-2011,39238.html

Ryan, Yasmine. (2011a). The massacre behind the revolution. Aljazeera. Retrieved from http://www.aljazeera.com/indepth/features/2011/02/2011215123229922898.html

Ryan, Yasmine. (2011b). The tragic life of a street vendor. Aljazeera. Retrieved from http://www.aljazeera.com/indepth/features/2011/01/2011126121815985483.html

Ryan, Yasmine. (2011c). How Tunisia's revolution began. Aljazeera. Retrieved from http://www.aljazeera.com/indepth/features/2011/01/201111684242518839.html

The South African Centre of International P.E.N. (2010). Tunisia: Boukaddous and seven others released. Retrieved from http://www.sapen.co.za/site/Writers-in-Prison/tunisia-boukaddous-and-seven-others-released.html

Tunisian Riots English Information. (n.d.). In Facebook. [Group page]. Retrieved April 24, 2012, from http://www.facebook.com/pages/Tunisian-Riots-English-information/185807121448908

Ungerleider, Neal. (2011). Tunisian Government Allegedly Hacking Facebook, Gmail Accounts of Dissidents and Journalists. Fast Company. Retrieved from http://www.thetechherald.com/articles/Tunisian-government-harvesting-usernames-and-passwords/12429/

Walt, Vivienne. (2011). Tension Grips Tunisia's Capital After Leader Flees. Time Magazine. Retrieved from http://www.time.com/time/world/article/0,8599,2042622,00.html

wconeybear. (2008 [seems like erroneous date]). Anonymous Operation Tunisia rages, US Govt grows worried. Myce. Retrieved from http://www.myce.com/news/anonymous-operation-tunisia-rages-on-us-govt-grows-concerned-38617/

Zuckerberg, Mark. (2009). An Open Letter From Facebook Founder Mark Zuckerberg. The Facebook Blog. Retrieved from http://blog.facebook.com/blog.php?post=190423927130.

Zuckerman, Ethan. (2008). The Cute Cat Theory Talk at ETech. my heart's in accra. Retrieved from http://www.ethanzuckerman.com/blog/2008/03/08/the-cute-cat-theory-talk-at-etech/

Memory Light Game:

An Embedded Processor Design Project

A Technical Report

in STS 4600

Presented to

The Faculty of the

School of Engineering and Applied Science

University of Virginia

In Partial Fulfillment of the Requirements for the Degree

Bachelor of Science in Computer Engineering

Daniel Toler

May 11, 2012

Technical Project Team Members:

Torrest Bates

Gautham Kurup

Daniel Toler

On my honor as a University student, I have neither given nor received unauthorized aid on this assignment as defined by the Honor Guidelines for Thesis-Related Assignments

Signed: ___________________________________________________ Date:_____________________________________________________

Approved: _________________________________________________

Professor Mircea Stan of the Department of Electrical and Computer Engineering

Table of Contents

I. Abstract

II. Motivation Behind the Processor Design Project

III. Team Organization and Design Decisions

IV. The Classic 5 Stage RISC Pipeline

V. Development Experiences

VI. Challenges Faced

VII. Lessons Learned

VIII. Conclusions and Future Work

IX. Works Cited

Abstract

The main goal of this project was to design a processor in VHDL based on a provided instruction set architecture (called SAYEH) and to port that design to an FPGA board. A multicycle VHDL implementation of the processor was provided by the creators of the ISA, and our team tried to convert the pipelined processor VHDL design from last semester's class so that it implemented the SAYEH ISA. Although we did partially convert the pipelined processor to support SAYEH, it did not work properly with the provided memory, and it lacked some functionality. Simultaneously, with the development of the pipelined processor, we also developed a single cycle which successfully met the SAYEH ISA, seemed to work perfectly in simulation, and worked, at least partly, on the board, although we had limited time to test it. In addition, we developed a memory game involving LEDs to run on the single cycle.

Motivation Behind the Processor Design Project

Advanced Digital Logic Design is the second semester of a year-long design project that partially satisfies the requirements for completion of a Bachelor of Science in Computer Engineering at UVA. The design project is currently transitioning away from being simply a processor design project to being also an embedded system design project. According to Professor Mircea Stan, this reflects the reality that only a small cadre of companies actually designs processors nowadays, whereas many companies design embedded systems. The project this semester was intended to involve designing and optimizing a pipe-lined processor and then incorporating it into an embedded system. However, by about mid-way through the semester it became clear that no one was making sufficient progress for the class to achieve both goals. Rather than discard the idea of an embedded system, however, our group decided to get a processor working no matter what (even if it was a simple single cycle) and then interface with some form of input/output (I/O) to form a simple embedded system. To make the project more fun, we decided that our embedded system should realize a computer game of some kind. Thus, our personal motivations for the project included:

Graduating from the University of Virginia

Understanding many of the difficulties of a large digital logic design project

Becoming as proficient as possible at digital logic design

Having as much fun as possible

We haven't taken any of these goals lightly, and we think we've achieved all of them at least to some extent.

Team Organization and Design Decisions

We divided team responsibilities in an unusual way. As a response to our failure to meet many of the early class deadlines (along with most of the class), we chose to split development into two separate paths. Torrest Bates felt that the pipelined processor he had designed the previous semester was close enough to functioning that the best course was for him to continue modifying it to satisfy the SAYEH ISA, which was the processor ISA we were instructed to implement (for full details on the SAYEH ISA, see Navabi, p.188, 2007). Due to the complexity of Torrest's code and a lack of understanding by the rest of the group, it was decided that Daniel Toler would begin designing a single cycle processor from scratch satisfying the SAYEH ISA. This was the simplest possible processor design, and it could be later transitioned to a pipelined version. Gautham Kurup was tasked with providing assistance for both development paths as well as having greater responsibility for lab reports. We used VHDL as our HDL of choice throughout the project. Although we admired some of the features that made Verilog code concise, we never seriously considered using it, since we were familiar with VHDL from the previous class.

The Classic 5 Stage RISC Pipeline

In the 1980's when David Patterson and David Ditzel wrote "The Case for the Reduced Instruction Set Computer," ( Patterson and Ditzel, 1980) memory was expensive. Because of that, along with other reasons, the dominant design paradigm was the CISC architecture. CISC instruction sets were large and contained many special purpose instructions intended to make code more compact. Today, however, memory is inexpensive, and there is evidence that a longer RISC program executes in roughly the same time as a shorter CISC program, since RISC instructions generally do not take as long to execute as CISC instructions. Hence RISC, as the paper argued long ago, seems more appropriate for modern processors. A RISC architecture uses a small set of simple, orthogonal commands that together provide all the needed functionality for a processor. Interestingly, the prevalent x86 architecture accepts CISC instructions but internally breaks them down into smaller RISC instructions for processing. The pipelined processor, like the RISC paradigm, is fairly modern also. When hardware was expensive, multi-cycle processors were popular. This was because in a multi-cycle, the various stages of processing take place in stages of time. The processing of a single instruction takes multiple clock cycles, but hardware area is conserved because some hardware modules are reused for multiple stages in the processing. As a side note, multi-cycle processors may be making a comeback in the form of tiny FPGA's that simulate a three-dimensional circuit by toggling between multiple stored "layers" each clock cycle. But, in any case, almost every processor nowadays is a pipelined processor. This is partially a result of the fact that transistors are much cheaper than they used to be. Pipe-lined processors take up more space in hardware (than for example multi-cycles), but they also allow for parallelism by causing each instruction to be processed in stages. Multiple instructions can be executed at roughly the same time, each of them in a different stage of execution. Indeed, the more finely grained you can subdivide the processing of an instruction, the more parallelism you can extract, although there is probably a limit.

An analogy using an every-day system helps to clarify the benefits of a pipelined architecture (Hennessy & Patterson, 2008). When people are doing their laundry, if there are multiple loads (A through D in the figures on the next page), and one washer and one dryer, after the first load is washed, it would be faster (Figure 1) to put in the next load into the washer while the first load is being dried rather than waiting for the dryer to finish with the first load (Figure 2). This way, during the time the dryer is being used, another useful task is also being performed (the next load being washed). Overall, this results in reduced time for all the loads

Almost any serious processor today is not only pipelined but also superscalar, which exploits a different kind of parallelism to increase the number of instructions per clock cycle. However, for academic purposes, the classic five stage RISC pipeline is commonly studied. MIPS (Hennessy, Jouppi, Baskett, and Gill, 1981) is a well-known classic RISC pipeline (different versions of MIPS have 5 or 6 stages), but our task in this class was to implement SAYEH, a five stage pipelined processor designed by Zainalabedin Navabi (2007). A classic five stage RISC pipeline like SAYEH has the following stages: Instruction Fetch, Instruction Decode, Execute, Memory, and Write Back. Each of the stages is separated by a memory element called a buffer, so that instructions can advance from one stage to the next each clock cycle. In the Instruction Fetch stage, the instruction is simply retrieved from instruction memory. In the Decode stage, the operands in the command are interpreted and appropriate register values are retrieved from the register file. In the Execute stage, the operands are used to perform calculations or operations of some kind. In the Memory stage, data is read to or from memory. And in the Write Back stage, data is written back to the register file. A good pipeline should be well-balanced in the sense that signals should propagate through the worst-case path of every stage in about the same time. The length of the clock cycle of the processor, which determines its speed, is set by the worst-case time of all of the stages, so it makes sense to minimize the worst-case time.

There are several basic kinds of instructions that we might want a processor to execute. We want instructions for reading from and writing to memory, the register file, and input and output devices. We want instructions for performing calculations or operations on data we have gotten from somewhere else. And we want instructions for jumping to different places in the program, possibly based on the result of a comparison or calculation. The most basic kind of processor, a single cycle, would execute any one of these instructions in a single clock cycle, start to finish. But, as we have already described, a pipelined processor executes these instructions piecemeal, and multiple instructions can be in the pipeline at the same time. As you might expect, this can result in complications for certain combinations of instructions.

For example, a jump instruction doesn't actually change the current position in the program until it reaches the Execute stage. Meanwhile, two other instructions enter the pipeline behind it. If we did not do anything special about the situation, those two instructions would execute completely before we made the jump to another section of code. This could result in much undesired program behavior. This problem is called a "control hazard," and the simplest approach to fixing it is to reset the buffers holding the two instructions so that they simply disappear, and after a lapse of two clock cycles, the program jumps to the new position in code. There are much better ways of dealing with the control hazard problem, but this provides a simple example of what is termed "hazard handling." There are two other types of hazards: "data hazard" and "structural hazard". A data hazard occurs when one instruction relies on a data value that is modified by another instruction later in the pipeline. A structural hazard occurs when two different stages need to access the same resource at the same time, such as memory. These hazards cause the pipeline to fail under certain situations unless they are addressed. They have to be dealt with; they cannot simply be ignored! And they illustrate some of the difficulties of designing a pipelined processor.

A discussion of the theory behind our pipelined RISC processor would not be complete without mentioning memory. Our processor uses a Von Neumann architecture, which means that instructions and data are both stored in the same memory. Memory accesses are made every time an instruction is fetched, and they are also made every time an instruction reads data from memory or writes data to memory. Since the processor has to wait to proceed (stall) until memory returns data, any way to speed up the memory would be desirable. Caches offer a partial solution for increasing memory speed. One important factor that differentiates cache memory from main memory is that caches are located on the processor chip itself, while main memory (RAM) is located outside the CPU. The close proximity of cache to the functional areas of the CUO means the CPU can access the data much quicker than having to wait for it from the main memory. Caches work by providing the illusion of fast access to main memory by storing sections of main memory for ready access. In other words, caches provide the means to take advantage of locality, which is the concept that data or instruction read once has a good chance of being read again, so it is brought into the cache which can send it out to the CPU faster than main memory can, when the CPU requests it again.

Development Experience

In the pipeline development path, we started off using a processor that Torrest Bates had designed from the previous semester, satisfying an ISA defined by Professor Ronald Williams. In transitioning to the new ISA, we changed the length of the instruction memory, the control module for the ALU, and the bit width of the registers. We also created a window pointer, and designed a new register file that allowed for writing to the high or low 8 bits of a register. We made numerous other minor changes to conform the processor to the SAYEH ISA.

In the single cycle development path, the processor had two separate memories: instruction memory and data memory. Instruction memory was internal, and consisted of a number of case statements that looked like this: when addr=x"0001" => data <= x"00FF"; The provided test bench assumed that instruction memory and data memory were not separate, so we thought using it for the single cycle was out of the question. In retrospect, we probably could have modified the test bench or our single cycle to take advantage of the test bench’s ability to compile from assembly to machine code. Instead, we wrote our own custom compiler that read an assembly file and outputted the VHDL case statements to be placed in Instruction Memory. The single cycle development path avoided some of the endless debugging and mysterious problems of the pipeline development path, but it was still surprisingly time consuming to get the single cycle to work perfectly. Our TA, Thumarros Rugthum helped us to get the single cycle on the FPGA board and to configure a setup in the Quartus software that allowed us to use 16 of the onboard switches as input, and 16 of the onboard LED's as output (See Figure 3). Tests of the I/O setup seemed to show that the configuration worked. Sometimes we assigned the processor clock to the internal clock on the board, but for much of the on-board testing we assigned the clock to a push button to give us single-stepping control. This worked acceptably well, although problems with switch "bouncing" would sometimes cause the processor to freeze up.

For both processors, our testing procedure in simulation involved examining ModelSim waveforms (Figure 3), forcing values or trying out different programs to try to make the processor fail. When a waveform value was different from what we expected, we would look at the area of code that seemed responsible for the behavior. (This was one way that modular development was helpful by allowing us to isolate behavior in a relatively few lines of code.) We also prototyped extensively before incorporating a new feature into either one of our processors. We tested individual components, and then carefully incorporated them one at a time, testing as we did so.

Challenges Faced

One of the major difficulties we faced during the project was figuring out how to divide the work effectively on our team. Idiosyncratic coding styles proved a major challenge to collaboration. Also, varying skill levels and expertise made it difficult to know how to assign work.

We also initially faced some minor difficulties with the development software we were using. The block diagram feature of ModelSim, which allowed for visually laying out the circuit design, didn't work on everyone's computers, even after re-installation. We also discovered that rolling back to an earlier version of our project was not as simple as restoring files in our DropBox folder. It was necessary to force a recompilation, because otherwise ModelSim simulated the past version of the HDL, causing some confusion. The Quartus II software was finicky, sometimes giving mysterious error messages that would disappear later for an unknown reason.

We also had difficulty writing assembly code for the SAYEH ISA, since some SAYEH instructions are translated as 8 bit instructions and some are translated as 16 bit instructions into machine code. Our custom compiler packed instructions in such a way that the least amount of memory would be used for a given program. Sometimes two 8 bit instructions would be packed together. And sometimes one 8 bit instruction would be followed by a 16 bit instruction. This made it difficult to estimate how many actual lines of machine code would be generated by a given number of lines of assembly code. This created a real difficulty when writing assembly programs that did branching. Calculating the distance to a particular instruction in terms of machine code was irritating. Also, since branching always jumps execution to the beginning of a 16 bit line of machine code, if there are two 8 bit instructions side by side, the only way to jump to the second instruction is to replace it with a no-op, forcing it onto its own line. But this had to be done manually. We had a plan in the works to update our custom compiler so that it would allow branches and jumps relative to the original assembly line number, but we never implemented it.

We faced a number of challenges in the actual design of the processor on both development paths. One particular problem took a long time to figure out, due to a lack of understanding about how the SAYEH memory test bench worked. The SAYEH memory test bench is level sensitive, outputting data instantly through the instruction fetch stage. As a result, the fetch stage acted in a le