Academia.eduAcademia.edu

As the Pirates Become CEOs: The Closing of the Open Internet. Daedalus, 145(1), 65–78. Winter 2016. http://doi.org/10.1162/DAED_a_00366

Zeynep Tufekci
This Paper
A short summary of this paper
37 Full PDFs related to this paper
As the Pirates Become ceos: The Closing of the Open Internet Zeynep Tufekci Abstract: The early Internet witnessed the flourishing of a digitally networked public sphere in which many people, including dissidents who had little to no access to mass media, found a voice as well as a place to connect with one another. As the Internet matures, its initial decentralized form has been increasingly replaced by a small number of ad-½nanced platforms, such as Facebook and Google, which structure the online experience of billions of people. These platforms often design, control, influence, and “optimize” the user experience accord- ing to their own internal values and priorities, sometimes using emergent methods such as algorithmic ½ltering and computational inference of private traits from computational social science. The shift to a small number of controlling platforms stems from a variety of dynamics, including network effects and the attractions of easier-to-use, closed platforms. This article considers these developments and their consequences for the vitality of the public sphere. I traveled to Cairo in the spring of 2011, a few months after the fall of President Hosni Mubarak. Egypt was unsettled but jubilant, and the rest of the Middle East had not yet fallen into war or renewed authoritarian- ism. One of the Egyptians I interviewed was a blogging pioneer whom I will call Hani.1 In the early 2000s, Hani had been among the ½rst to take advantage of the burst of freedom experienced by Egyptians before the ZEYNEP TUFEKCI is Assistant Pro- authorities fully caught on to the Internet’s revolu- fessor at the School of Information tionary potential. Many bloggers made it through the and Library Science, with an af½liate appointment in the Department of Mubarak era largely unscathed because the govern- Sociology, at the University of North ment could not keep up with or fully understand the Carolina, Chapel Hill. She is the au- new medium. Unfortunately, Hani had caught the at- thor of Beautiful Tear Gas: The Ecstatic, tention of the government; he was tried and sentenced Fragile Politics of Networked Protests in to years in prison for the crime of insulting Mubarak. the 21st Century (forthcoming 2016) Throughout his imprisonment, he remained de½ant. and coeditor of Inequity in the Technop- He was released in November 2010, just months be- olis: Race, Class, Gender, and the Digital fore a Facebook page would spark a revolution that Divide in Austin (with Joseph Straub- haar, Jeremiah Spence, and Roberta would dramatically change the country, the region, G. Lentz, 2012). She is also a contrib- and the world. uting opinion writer for The New York Before going to jail, Hani felt that his blog had been Times. a bustling crossroads of discussion. His voice reached © 2016 by Zeynep Tufekci doi:10.1162/DAED_a_00366 65 As the farther than he had ever thought possible. Square on January 25, 2011. One year prior, Pirates After his multiyear involuntary hiatus, he only about one hundred protesters met in Become CEOs: resumed blogging, he told me, with enthu- Tahrir Square, where they were surrounded The Closing siasm and excitement. But when he came and outnumbered by the police. But this of the Open Internet out of jail in late 2010, he found that his blog, year, the protest quickly swelled to include and much of the Egyptian blogosphere, hundreds of thousands of Egyptians who had become a comparative wasteland. occupied the square until Mubarak stepped “Where is everybody?” Hani answered down. To many activists I talked with, Face- himself: “They’re on Facebook.” book’s reach felt empowering. A survey of At the time I interviewed him, that did Tahrir protesters con½rmed that social me- not seem like such a bad development. Just dia had been essential to the early turnout a few months earlier, a Facebook group ti- that had triggered the avalanche of dissent.4 tled “We are All Khaled Saed”–named af- Egyptian use of Facebook continued to grow, ter a young Egyptian man who had been and it became plainly obvious that Face- tortured and killed by the police–had be- book had become a major player in the civ- come the organizational core of the revolu- ic sphere. Even the new military council that tion. The page was created in June 2010, a replaced Mubarak launched a Facebook few days after Saed’s death became public page. knowledge. The (then-anonymous) admin- But what did it mean for Facebook, a cor- istrator of the page was Wael Ghonim, a porate platform, to become so central to the Google employee and early adopter of the political life of the country? That was less Internet in the region. Ghonim had foreseen clear. Facebook’s potential to reach large num- With the advent of social media platforms bers of ordinary people: in just one month, in the mid-2000s, the “networked public his page gathered more than one hundred sphere”–the burgeoning civic space online5 thousand readers, and ordinary Egyptians that had been developed mostly through began using it to engage in political discus- blogs–expanded greatly, but with a simul- sion.2 In later interviews, some of those who taneous shift to commercial spaces.6 Many participated on the page told me that they scholars and civic activists worried about felt jubilant and liberated to be ½nally speak- how “sovereigns of cyberspace,” as Internet- ing about politics with other Egyptians. freedom advocate, journalist, and author Re- After the Tunisian revolution of early 2011, becca MacKinnon called these online plat- the “We are All Khaled Said” page became forms, would wield their power.7 Would a hotbed of conversation for Egyptians who they censor and restrict freedoms to serve longed for a similar upheaval. After much advertisers or governments with whom they discussion, including polls asking the page’s were trying to curry favor? Would they readers what they thought should be done, turn over user information to repressive re- Wael Ghonim created an event titled “The gimes? MacKinnon was prescient in iden- Revolution,” scheduled for January 25, 2011, tifying the core problem: the growth of pri- which was already a traditonal day of pro- vately owned spaces that functioned as pub- test in Egypt.3 Hundreds of thousands of lic commons. Over time, the threats posed Egyptians accepted an “evite” to “The Rev- by this relationship may exceed even our olution,” displaying their dissent openly, earlier concerns about censorship. many perhaps for the ½rst time, to their on- Driven by structural dynamics and cor- line social networks. porate motivations, as well as by character- Emboldened by the outpouring of dissent, istics of the Internet, these new social plat- thousands of people assembled in Tahrir forms are remaking the Internet in a way 66 Dædalus, the Journal ofthe American Academy of Arts & Sciences that imperils the open architecture of the of seeing it as a way to make that text richer. Zeynep early Web that felt so intoxicatingly em- You’re encouraged to post one single hyper- Tufekci powering to many of its users. The conse- link and expose it to a quasi-democratic pro- quences are profound. This article examines cess of liking and plussing and hearting: Ad- where we are now, and then briefly traces ding several links to a piece of text is usually the dynamics that have led us here. not allowed. Hyperlinks are objectivized, iso- lated, stripped of their powers. I n 2015, Hossein Derakhshan–who has At the same time, these social networks tend been called the “grandfather” of the Iranian to treat native text and pictures–things that blogosphere–left prison after serving six are directly posted to them–with a lot more years of a nineteen-year sentence for blog- respect than those that reside on outside web ging, including long stretches of solitary pages. . . . A link to the pictures somewhere con½nement. But prison did not break him; outside Facebook . . . are much less visible to instead, he says, what nearly broke his heart Facebook itself, and therefore get far fewer was what he found online when he started likes. The cycle reinforces itself . . . Instagram blogging again.8 –owned by Facebook–doesn’t allow its au- After being released, Derakhshan learned diences to leave whatsoever. You can put up that he needed to adapt to the new digital a web address alongside your photos, but it environment and use the new commercial won’t go anywhere. Lots of people start their social networks. Up for innovation and daily online routine in these cul de sacs of change, he created a Facebook account and social media, and their journeys end there.9 posted a link to his blog. To his dismay, his post disappeared after just a few “likes.” There are billions of people on the Inter- Likes are the main currency in Facebook’s net, but a few services capture or shape most all-important algorithm that decides which of their activities. Take Facebook: it has 1.5 posts to display to other users, and which billion users, a billion of whom log in daily to hide. In the new world of social media, to see updates and news from the hundreds posts like Derakhshan’s could disappear of people they have “friended” on the plat- without being seen by more than a handful form.10 Or consider Google: more than one of people. Derakhshan was despondent billion people use the site to run more than about trying to learn the ropes of this new three billion Google searches per day. Face- world. But he soon realized that his per- book recently announced a program en- sonal grasp of the platform was not the only couraging publishers to upload articles to missing ingredient. Facebook’s servers to make them appear The new platforms were strangling access faster to the end-users. Google is planning to the hyperlink, directing users to content a similar gambit with “instant” articles of within their walls and regulating access to its own. As smartphones continue to claim the outside Web in very speci½c ways. Con- an increasingly large share of Internet users, tent like his, which was hosted outside of Google is also designing a new way to dis- Facebook’s territory, did not stand a chance. play pages on mobile devices.11 Google’s Derakhshan wrote the essay “The Web new scheme would shift more power to the We Have to Save” about his new experience company; though, as with all the other tran- of being online: sitions, it would offer bene½ts to users as well, which often serve to mask, or at least Nearly every social network now treats a link make palatable, the expansion of power. as just the same as it treats any other object– For an increasing number of people, the same as a photo, or a piece of text–instead Facebook and Google are the Internet, or at 145 (1) Winter 2016 67 As the least the framework that shapes their ex- a city, for example: When residential and Pirates perience of it.12 These platforms own the of½ce buildings are separate, and people live Become CEOs: most valuable troves of user data; control in far-flung suburbs, there are social, polit- The Closing the user experience; have the power to de- ical, and cultural consequences. Low walk- of the Open Internet cide winners and losers, through small ability may contribute to unhealthy life- changes to their policies and algorithms, styles. Or political polarization may increase in a variety of categories, including news, while people segregate by income levels and products, and books; and use their vast earn- race. ings to buy up potential competitors. Online, computer code offers a similar I talked with Derakhshan (online, since he structuring power. For example, Facebook is still in Iran) about his experiences, shar- requires mutual consent to interact, while ing my own research about the shift to a Twitter allows people to “follow” someone world of algorithmic walled gardens. Both else without being followed back. On Face- of us are aware that current social media book, friending someone requires acquies- platforms reach many more people than cence on both sides: the person making the Internet did in the heydays of blogging. the request and the person accepting it. On That is not the problem. Neither is it the Twitter, any public account can be followed existence of more frivolous or mundane with just a click, without having to formally content online; cute cat and baby images ask for permission. These structures are are part of the package. The problem is the formed through decisions made by the peo- shift in the architecture of the Internet. In ple who run, administer, and create the code ways both dramatic and subtle, the shift for these platforms, and are implemented has begun to create new profound and far- by in-house coders, resulting in different reaching problems. In Derakhshan’s words, social and political environments for each a link is not just a link; it is a relationship. service. Facebook tends to have smaller net- The power of the Internet comes from our works made up of friends, family, and ac- relationships on it. And these relationships quaintances, while Twitter is better suited are increasingly mediated by the platforms for fan/celebrity relationships in which the that collect data about us; make judgments few can be followed by the many. Online about what is relevant, important, and vis- platforms are shaped not only by the code ible; and seek to shape our experiences for that structures visibility and access, but by commercial or political gain. computation and data as well. This combi- How did we get here? And how much nation gives online platforms powers for power is now concentrated in these plat- which there are no simple analogies in the forms? The answers to these questions are offline world. connected and offer hints of possible alter- The massive accumulation of user data native futures. has been written about extensively.14 There is an increasing amount of data about every- L egal scholar Lawrence Lessig has fa- one. More and more social, political, and ½nancial interactions are performed online. mously listed four forces that shape “cyber- space”: law, norms, markets, and code.13 He More and more people carry phones that compared his model to the offline world connect to the Internet and log their loca- where law, norms, markets, and architecture tion and activities. Everyday objects are in- play a major role in shaping society. Lessig creasingly acquiring sensors that collect in- analogized computer code, which de½nes formation even about passersby. Some of how online platforms work, to the role ar- these data are accessed by governments for chitecture plays offline. Take the layout of political purposes; some are used by com- 68 Dædalus, the Journal ofthe American Academy of Arts & Sciences panies and advertisers for marketing. Fi- tractable via direct questions, and far exceed Zeynep nancial institutions mine data to check the scope of information that could be gath- Tufekci credit-worthiness. Occasionally, the data are ered about a voter via traditional methods.17 leaked, hacked, or otherwise released for The computational inference generated reasons that can range from crime to politics by machine learning takes place during the to mischief. Ordinary people have very lit- process of sifting through many varieties of tle idea about who holds what kind of data data, with the proviso that the data are deep about them, or how the data are used. The and rich enough. Inferring political vari- amount of accumulated data and the asym- ables about a person does not require their metry of power between the people who participation in overtly political websites or are monitored and surveilled and the plat- conversations. For example, Facebook op- forms in which the data are held and mobi- erates mainly through likes: a one-click op- lized is a signi½cant problem, con½rmed by eration that signals a user’s approval of a polls revealing the public’s great uneasiness page, update, or person. The collection of about surveillance.15 these likes can be used to model, with sur- However, the involuntary accretion of prisingly high statistical reliability, a range massive amounts of data about people is of outcomes, including “sexual orientation, only the tip of the iceberg. In a networked ethnicity, religious and political views, per- society, computation brings another dimen- sonality traits, intelligence, happiness, use sion of asymmetric power. Through tech- of addictive substances, parental separation, niques that can be loosely collected under age, and gender.”18 the heading “computational inference”– This type of analytic power can go be- the application of statistical methods, mod- yond many of the traditional categories eling, and machine learning to vast troves used by demographers and advertisers to of data to make predictions–those who pro½le the public. By using only their social have gathered these data can infer from media imprints (again, not directly asking them information that has never even been questions of individuals), researchers have disclosed.16 been able to identify people who are likely In other words, aided by computation, big to become clinically depressed in the fu- data can now answer questions that have ture, even before the onset of clinical symp- never been asked about individuals who are toms.19 Much of this research is done with the sources of the data: the best of intentions: for example, as ear- The advent of big datasets that contain im- ly intervention for new mothers at risk for prints of actual behavior and social network postpartum depression.20 However, it is information–social interactions, conversa- easy to see the downsides of making infer- tions, friendship networks, history of reading ences using data in this fashion. Advertisers, and commenting on a variety of platforms for example, discovered that when women –along with advances in computational tech- feel “lonely, fat, and depressed” they are niques means that political campaigns (and more likely to purchase makeup, and that indeed, advertisers, corporations and others such women are ideal targets for “beauty with the access to these databases as well as interventions.”21 In other words, women technical resources) can model individual who are depressed and lonely can be more voter preferences and attributes at a high level easily sold makeup. It does not take much of precision, and crucially, often without ask- imagination to see that advertisers will ing the voter a single direct question. Strik- therefore want to use data gathered by on- ingly, the results of such models may match line platforms to ½nd out exactly who is feel- the quality of the answers that were only ex- ing “lonely, fat, and depressed” and mar- 145 (1) Winter 2016 69 As the ket to these targeted women at exactly these incorporate many biases. However, if we Pirates times. use social media data churned through com- Become CEOs: The increasing use of opaque computa- putational methods for hiring, we may The Closing tional methods known as machine learn- move from imperfect hiring systems that of the Open Internet ing–or “neural networks”–adds another we know discriminate against women, for layer of complexity to predictions made example, to ones whose workings are hid- with big data sets. These are systems that den from us, but nonetheless still discrimi- “learn to learn” how to classify individuals nate. This could mean using systems that (or whatever type of cases they are present- discriminate only against women who are ed with) into various categories. Machine- statistically likely to become pregnant soon. learning systems are often provided with a This type of discrimination would not be “training set”: a database in which cases visible to employers because neither the are marked with the correct answers. women being hired nor the women not be- For example, to train a machine-learning ing hired would be pregnant at the time of system, an employer might provide it with the hiring, and because a machine-learning a list of employees he has classi½ed as either system does not display decision-making “high-performance” or “unsatisfactory,” variables that are easily interpretable, even accompanied by social-media data about by its engineers. Social media platforms in- all employees in the database. Without re- creasingly hold the kind of data that can be ceiving direct instruction or a recipe about used in these ways. what makes a worker either high-perfor- mance or unsatisfactory, the system learns the set of associations that are linked to each While this combination of big data and computation obviously creates signi½cant outcome, and how to use that knowledge challenges, there are additional, equally to classify new employees. On the surface, daunting issues. When combined with the this looks a lot like many other methods power of “code” as architecture, in the sense that employers use to discriminate among ½rst identi½ed by Lessig,22 platforms can al- potential hires. But there is a twist: a ma- so nudge behavior, quietly and impercept- chine-learning system often does not pro- ibly, and sometimes in ways that are not di- vide any human-understandable clues to rectly visible even to the people who run the why it classi½es the way it does. In fact, if platforms. Facebook, for instance, uses an we knew exactly what it was doing, there algorithm to order the news feed that shows would be no need for the “machine-learn- its 1.5 billion users’ status updates. These ing” part: we could just program the criteria may range from updates that are purely per- ourselves. In reality, though, all that a man- sonal in nature to news articles. Increasing- ager might know is that the system places ly, for many population segments ranging potential hires into one category or the oth- from younger people in developed coun- er, without any understanding of what parts tries to populations just coming online in of the social media big data set were used poorer countries, Facebook has become the as signals for a particular outcome. number one source of news.23 In poorer For all a hiring manager knows, such a countries, many people are not even aware system might classify applicants based on that there is an Internet outside of Face- criteria such as statistical likelihood of ex- book,24 and many others choose to stay periencing depression in the future (even if completely within Facebook’s realm.25 As undiagnosed at the time of evaluation) or David Clark explains in his essay in this issue, the possibility of impending parenthood. Facebook has helped ensure this through It is well known that current hiring systems promotion of its stripped-down Facebook 70 Dædalus, the Journal ofthe American Academy of Arts & Sciences app–0.facebook.com–which, in agree- rolls, that single message caused 340,000 Zeynep ment with mobile service providers in many additional people to turn out to vote in the Tufekci developing countries, does not incur data 2010 U.S. congressional elections.28 In an- charges for users.26 other experiment, Facebook randomly se- In my research, I have encountered many lected whether users saw posts with slightly people whose Internet routine resembles more upbeat words versus more downbeat the following: If on a desktop computer, a ones: the result was correspondingly slight- user launches a browser and types “Face- ly more upbeat or downbeat posts by those book” into Google’s search box, likely un- same users. Dubbed the “emotional conta- aware that the url bar at the top of the gion” study, this incident sparked interna- browser is a separate and faster way to get tional interest in Facebook’s power to shape there. Google brings up Facebook as the ½rst the user’s experience.29 link, and the user clicks on Facebook and The power to shape experience (or per- proceeds to interact mostly within the site. haps elections) is not limited to Facebook; If using a mobile platform, which is increas- there are other powerful platforms. For ex- ingly the norm, a user will simply launch ample, Google rankings are hugely conse- the Facebook app and rarely encounter the quential. A politician can be greatly helped open Web at all. or greatly hurt if Google chooses to highlight This tendency to stay within Facebook is or hide, say, a link to a corruption scandal what gives Facebook’s architectural deci- on the ½rst page of its results. A recent study sions such power, and invisibly so. In one showed that slight changes to search rank- study, 62.5 percent of users had no idea that ings can shift the voting preferences of un- the algorithm controlling their feed exist- decided voters, and that these shifts can be ed, let alone how it worked.27 This study hidden so that people show no awareness used a small sample in the United States, of the manipulation.30 where the subjects were likely more educat- For a small taste of how platform choices ed about the Internet than many other popu- affect the civic sphere, consider the case of lations globally, creating a potentially un- the protests in Ferguson, Missouri, in Au- representatively low estimate. The news gust 2014. What started as a community feed is a world with its own laws of phys- shaken over the police killing of a young ics, and the deities that rule it are Facebook man under murky circumstances grew into programmers. In this world, some types of major protests after the police responded information are nudged and helped to to initial small-scale–and completely non- spread more, while others are discouraged. violent, according to journalists on the scene There is great power in what we do (and do –protests by residents with overwhelming not) see from our friends and acquaintanc- force, including the use of attack dogs and es, and increasingly, this is greatly influ- tear gas. A few national journalists, as well as enced by platform design and code. ordinary citizens with smartphones, start- Facebook’s own research has shown the ed tweeting from the scene of the initial pro- power of its designers’ architectural choic- tests. The burgeoning unrest and conflict es. In one Facebook experiment, randomly soon grew into major Twitter discussions selected users received a neutral message to that later sparked the attention of the main- “go vote,” while others, also randomly se- stream news media. About three million lected, saw a slightly more social version of tweets were sent before the mass media be- the encouragement, noting also which of gan covering events in Ferguson. The na- their friends voted using small thumbnails tionwide movement that grew from these of their pro½le photos. Matched with voter events is often referred to as the “Black Lives 145 (1) Winter 2016 71 As the Matter” movement, named after the Twit- independent relationship between people? Pirates ter hashtag. Why are they dictating who sees what? Become CEOs: However, on the ½rst night of the pro- Some aspects of the answer are decep- The Closing tests, the topic was mostly invisible on Face- tively simple, and at the same time deeply of the Open Internet book’s algorithmically controlled news structural. The open Internet that held so feed.31 Instead, the “ice bucket challenge,” much generative power took a turn toward in which people poured cold buckets of wa- ad-½nanced platforms, while the dangers ter over their heads and, in some cases, lurking for ordinary users from the Inter- donated to an als charity, dominated the net’s open and trusting design were not Facebook news feed. This was not a situa- counteracted, causing people to flee to safer tion that reflected Facebook users’ lack of and more user-friendly platforms. In com- interest in the Ferguson protests; rather, it bination, these two developments encour- was an indication that it is hard to “like” aged, enabled, and forced the creation of –Facebook’s dominant algorithmic signal massive, quasi-monopolistic platforms, –such disturbing news, while it is easy to while incentivizing the platforms to use give a thumbs-up to a charity drive. Once a their massive troves of data with the power topic is buried by an algorithm, this be- of computational inference to become bet- comes a self-feeding cycle: fewer people ter spy machines, geared toward ad delivery, are able to see it in the ½rst place, with few- the source of their ½nancing. er still choosing to share it further, causing From Wikipedia to question-and-answer the algorithm to bury it deeper. On Twit- sites to countless numbers of sites and blogs ter’s platform, in which users see all posts that provide a public service (but not pay- from the people they follow in chronologi- ment for their creators), the Internet offers cal order, the topic grew to dominate dis- direct proof that people enjoy sharing their cussion, trending locally, nationally, and creative and personal output with others.34 globally, catching attention of journalists If there were ever a need to expand our con- and broader publics. On Facebook, it barely ception of humanity beyond the restricted surfaced. Given the importance of online “homo economicus” who works only for platforms and public attention to political his or her bene½t, the explosion of user- movements, burying such news is highly generated content on the Internet has pro- consequential.32 Had our media been exclu- vided major evidence.35 However, creative sively controlled by an algorithm in which and altruistic output alone does not provide “liking” were the main emotive input, the ½nancing for servers, coders, and database long and hard national conversation about management. As the public Internet scaled race and policing in America that was gen- up and grew in numbers of participants, erated by the Ferguson protests might have many websites faced a dilemma: whether never transformed into a national move- to charge their users, or to sell users’ eye- ment.33 balls to advertisers. It was a crucial turning point: were people How did we get here? Was it inevitable? going to be the customers, or were they go- ing to be the product sold? Almost all of the Tracing this path requires combining and probing the two questions posed by Hani major platforms went with advertising. As and Derakhshan, two people who blogged Ethan Zuckerman, then a staff member of under repressive regimes and who were re- one of the Internet’s earliest user-generated leased from prison ½ve years apart. Why is platforms, tripod.com, explains: everyone on Facebook now? And why are Advertising became the default business mod- these platforms killing the hyperlink as an el on the web, “the entire economic founda- 72 Dædalus, the Journal ofthe American Academy of Arts & Sciences tion of our industry,” because it was the eas- platforms, such as Facebook, tend to quickly Zeynep iest model for a web startup to implement, dominate their market and become near- Tufekci and the easiest to market to investors. Web monopolies. This is also why everyone lists startups could contract their revenue growth their wares on Ebay, where all the buyers are, to an ad network and focus on building an and advertises on Google, where all the eye- audience. If revenues were insuf½cient to cov- balls go. The fact that a lot of people already er the costs of providing the content or serv- have Facebook accounts means that consid- ice, it didn’t matter–what mattered was au- erations of network externalities will result dience growth, as a site with tens of millions in existing people staying put, or new people of loyal users would surely ½nd a way to gen- joining in anyway, even if they have qualms erate revenue.36 about the privacy issues.37 These decisions were made partly out of While network externalities made it pos- idealism: a free website-hosting platform sible for platforms to become very large, like Tripod also allowed Thai dissidents to the ad-½nancing model meant that a mid- circulate otherwise censored content with- sized platform, even one with hundreds of out worrying about paying for the site. It millions of users, faced great challenges, made more sense at the time to have ads since ads on the Internet are not worth than to charge users. But once advertising much.38 An ad-dependent platform can on- became the way to make money, almost ly survive if it serves enormous numbers everything flowed from it, especially when of people. For example, Wall Street’s in- combined with another key feature of on- vestors have soured on Twitter because it line platforms: network effects. only has about three hundred million us- Network effects, also called network ex- ers. For most products, hundreds of mil- ternalities, are the tendency of the value of lions of users would appear to be a huge suc- some products or services to increase as cess. In an ad-½nanced online world, that’s more people use them, and to become less barely enough to get by. worthwhile when they are not used by oth- But there is one key path for online ads ers, even if the less popular product or ser- to become more valuable for platforms. If vice is objectively better, cheaper, faster, or platforms accumulate a great amount of more diverse in its offerings. For many on- data on their users, and harness computa- line applications, everyone wants to be tional inference to “understand” them on where everyone else is. This dynamic al- behalf of their advertisers, then the ads, lows many online platforms that manage which have a higher chance of leading to a to get ahead of their competition to com- purchase, are worth a lot more. These adver- pletely dominate their niche: tisers could include both corporate entities selling products and political campaigns The more people own fax machines, for ex- marketing politicians. Platforms can also ample, the more useful each one becomes. use their architectural power to create an That is also why there is a single standard for environment that is more advertiser-friend- fax machines–would you switch to a brand ly. Until quite recently, for example, Face- new, faster fax machine standard if there book allowed likes as the only signal (aside was nobody else you could fax with your ma- from making comments) that users could chine? Research shows that the presence of send about a page or status update. While network externalities trumps product pref- Facebook recently expanded choices in a erence or quality; many people will chose a few countries to include a few more “one- service that has more users compared to the click” options such as “like,” “love,” “ha- one that is otherwise better for them. Such ha,” “yay,” “wow,” “sad,” and “angry,” the 145 (1) Winter 2016 73 As the expanded list is still heavily geared toward protocols that underpins almost all Internet Pirates positivity, with only two that are typically commerce. The bug “heartbleed” allowed Become CEOs: associated with negativity: angry and sad. an attacker to read parts of a computer’s The Closing Overall, many of the issues identi½ed in memory that the program should not ordi- of the Open Internet this article are a direct consequence of this narily have access to, and to learn crucial combination: Internet platforms are ½- private information, including stored pass- nanced by ads that demand great scale, and words. While it is almost too ridiculous to they are fueled by network effects that al- believe, the Openssl architecture, used by low such scale through the emergence of about two-thirds of all web pages, includ- monopolies. These quasi monopolies then ing almost all major banks, is maintained have incentives to collect and process vast by a group of only a dozen people, all but amounts of data on their users to make the one of whom are volunteers.39 The crisis ads more effective for the advertisers, while with Openssl was but one example of crit- also controlling the experience of the users ical parts of the Internet’s infrastructure to keep the platform advertising-friendly, that provide security for ordinary users be- and to keep the user from leaving the plat- ing tended by almost nobody. There is very form. little energy or resources dedicated to tend- ing the commons of the Internet, and the The other major development over the resulting environment has made ordinary Web navigation increasingly dif½cult and past decade from the user side has been the lack of attention and resources to ensure user data increasingly insecure. For regular that the open Web–the one in which the users, remaining within trusted walled gar- hyperlink and address bar, rather than a dens, like Facebook or Google’s new pro- closed platform and its algorithmic and ar- posed Web architecture, is a reasonable chitectural choices, dominate navigation– choice. This is exactly the scenario warned remains a secure and navigable place for or- against by scholars. dinary users. This shift toward the walled gardens is Many of the early protocols that de½ned only increasing as the next billions come the Internet were developed for use by a online: people with less technical literacy, trusting, small, and closed community of less powerful devices, shakier Internet con- academic and military research staff. How- nections, and often mobile-only access. In ever, on the current scale of billions of peo- developing nations, the walled gardens of ple, the Internet’s insecurity, and the pro- huge online platforms have many draws. liferation of malware, spam, and untrust- Network effects means that their expatriate worthy sites, has caused many to retreat to relatives and friends are most likely to be easier-to-use, relatively safe platforms. The on the biggest platforms. A controlled en- ad-½nancing model means that almost all vironment makes the Internet more navi- commercial websites have installed exten- gable. Bigger platforms offer better trans- sive ad-tracking software on their sites, lation and localization services, something which is not distinguishable, in effects or volunteer sites have more dif½culty provid- operation, from malware dedicated to spy- ing. Google helps order the chaotic, seem- ing. Navigating the ordinary, open Internet ingly endless, choices effectively, while Face- now seems treacherous and feels slow (since book offers a way to manage the flow of in- the sites are loaded with ads and tracking formation from a user’s social networks, software). albeit algorithmically curated within an ad- In 2014, for example, a massive vulnera- delivery platform. And thus, the Internet bility was found in “Openssl,” one of the giants continue to grow, and have become 74 Dædalus, the Journal ofthe American Academy of Arts & Sciences the dominant landscape of the Internet for many of the current tech giants. The influ- Zeynep most people. ence of network effects is especially power- Tufekci ful for user-generated platforms, since what I n his prescient book The Future of the In- partially powers them is user investment. People have spent a lot of time and effort ternet–And How to Stop It, Jonathan Zittrain warned about these problems, and predict- building up their positive feedback on eBay ed that unless addressed, they would lead and cultivating their social networks on to the collapse of the open, generative In- Facebook. It is unlikely that competition ternet in favor of closed systems.40 Legal alone–even competition offered with bet- scholar Tim Wu looked into past informa- ter terms and services–can dislodge these tion systems and pointed out that many be- powerful platforms, given the costs sunk came dominated by monopolies.41 As early into them by their users. as 2003, Deborah L. Spar–now president The path toward change is uphill, but of Barnard College–predicted that insur- the ½rst step requires the public recognition gent technologies would pass from the “pi- of what dissidents in repressive regimes– rates” that use technologies to disrupt order often the canaries in mines–have already to the hands of powerful commercial and discovered: the power of the Internet de- governmental bodies who use it to consol- rives from our ability to freely connect with idate power.42 The Internet, in some ways, each other. These developments are not seems set on this path, although we have changing only from one type of program not yet passed the point of no return. or site to another; they are shifting to a new Because ad-based ½nancing quickly de- regime in which our relationships are me- volves into large-scale, monopolistic sys- diated by forces trying to mine our data, tems working on behalf of advertisers, to mostly in order to sell a few more ads slightly change directions, we ½rst must change how more effectively, but also open to a host of we ½nance the Internet’s platforms, includ- other political uses.43 From politics to cul- ing ½nancing potential challengers to cur- ture, much power resides with owners of rently dominant platforms. Alternative data, especially those possessing command models of ½nancing were developed in the of computation and online architecture. early days of the Internet, but these were It is not too late to change this path, but quashed, in part because they may have been to do so requires an open-eyed and realistic too early for mass adoption, but also be- look at the forces that have brought us here cause banks and websites resisted their im- –½nancing models, the need for tending plementation. Second, the Internet’s com- the security of the Internet’s commons, de- mons needs tending, which will also require mand for usability, and the shift to mobile– substantial resources and ½nancing as well. and asking how to generate an alternative A global system whose security depends so model that can scale-up. That demand still much on volunteer work will, inevitably, be- exists: the ½rst billion Internet users have come a dif½cult-to-navigate, insecure, and experienced, and remember, the admittedly unpleasant experience, and will result in chaotic early Internet, built upon the energy people retreating to safer platforms that and euphoria of people discovering both cushion the user experience while also con- information and each other. Now that the trolling it. Third, we must recognize that due Internet is approaching three billion users, to network effects, unregulated markets the question facing us is whether their (one of the mechanisms of Lessig’s original Internet experience will much differ from four forces) do not work well on the Internet a tightly regulated coffeehouse within a gi- for certain kinds of platforms, including gantic shopping mall. 145 (1) Winter 2016 75 As the endnotes Pirates 1 Become He did not ask me to keep his identity secret, but I am not using his name on principle, to avoid CEOs: my arguments getting tangled with his views as a result of Google searches run by clumsy repres- The Closing sive regimes. of the Open 2 Internet Jennifer Preston, “Movement Began With Outrage and a Facebook Page That Gave It an Outlet,” The New York Times, February 5, 2011, http://www.nytimes.com/2011/02/06/world/middleeast/ 06face.html. 3 Wael Ghonim, Revolution 2.0: The Power of the People is Greater Than the People in Power–A Memoir (Boston: Houghton Mifflin Harcourt, 2012); and the author’s private conversation with Wael Ghonim (2015). 4 Zeynep Tufekci and Christopher Wilson, “Social Media and the Decision to Participate in Politi- cal Protest: Observations From Tahrir Square,” Journal of Communication 62 (2) (2012): 363–379, http://doi.org/10.1111/j.1460-2466.2012.01629.x. 5 Yochai Benkler, The Wealth of Networks: How Social Production Transforms Markets and Freedom (New Haven, Conn.: Yale University Press, 2007). 6 Steven Johnson, “Can Anything Take Down the Facebook Juggernaut?” Wired, May 16, 2012, http://www.wired.com/2012/05/mf_facebook/. 7 Rebecca MacKinnon, Consent of the Networked: The Worldwide Struggle for Internet Freedom (New York: Basic Books, 2012). 8 Hossein Derakhshan, “The Web We Have to Save: The Rich, Diverse, Free Web that I Loved–and Spent Years in an Iranian Jail for–is Dying. Why is Nobody Stopping It?” July 2014, https:// medium.com/matter/the-web-we-have-to-save-2eb1fe15a426. 9 Ibid. 10 Don Clark and Robert McMillan, “Facebook, Amazon and Other Tech Giants Tighten Grip on Internet Economy,” The Wall Street Journal, November 5, 2015, http://www.wsj.com/articles/ giants-tighten-grip-on-internet-economy-1446771732. 11 Joshua Benton, “Get amp’d: Here’s What Publishers Need to Know about Google’s New Plan to Speed Up Your Website,” Nieman Lab, October 7, 2015, http://www.niemanlab.org/2015/ 10/get-ampd-heres-what-publishers-need-to-know-about-googles-new-plan-to-speed-up -your-website/. 12 Leo Mirani, “Millions of Facebook Users have No Idea They’re Using the Internet,” Quartz, February 9, 2015, http://qz.com/333313/milliions-of-facebook-users-have-no-idea-theyre-using -the-internet/. 13 Lawrence Lessig, Code: And Other Laws of Cyberspace, Version 2.0 (New York: Basic Books, 2006). 14 See, for example, Viktor Mayer-Schönberger and Kenneth Cukier, Big Data: A Revolution that Will Transform How We Live, Work, and Think (Boston: Houghton Mifflin Harcourt, 2013). 15 George Gao, “What Americans Think about nsa Surveillance, National Security and Privacy,” Pew Research Center, May 29, 2015, http://www.pewresearch.org/fact-tank/2015/05/29/what -americans-think-about-nsa-surveillance-national-security-and-privacy/. 16 Zeynep Tufekci, “Engineering the Public: Big Data, Surveillance and Computational Politics,” First Monday 19 (7) (2014), http://dx.doi.org/10.5210/fm.v19i7.4901. 17 Ibid. 18 Michal Kosinski, David Stillwell, and Thore Graepel, “Private Traits and Attributes are Predic- table from Digital Records of Human Behavior,” Proceedings of the National Academy of Sciences 110 (15) (2013): 5802–5805, http://doi.org/10.1073/pnas.1218772110. 19 Munmun De Choudhury, Michael Gamon, Scott Counts, and Eric Horvitz, “Predicting Depres- sion via Social Media,” in Proceedings of the Seventh International AAAI Conference on Weblogs and 76 Dædalus, the Journal ofthe American Academy of Arts & Sciences Social Media (Palo Alto, Calif.: Association for the Advancement of Arti½cial Intelligence, 2013), Zeynep http://www.aaai.org/ocs/index.php/ICWSM/ICWSM13/paper/viewFile/6124/6351. Tufekci 20 Munmun De Choudhury, Scott Counts, Eric Horvitz, and Aaron Hoff, “Characterizing and Predicting Postpartum Depression from Shared Facebook Data,” in Proceedings of the 17th ACM Conference on Computer Supported Cooperative Work & Social Computing (New York: Association for Computing Machinery, 2014), 626–638, http://doi.org/10.1145/2531602.2531675. 21 Lucia Moses, “Marketers Should Take Note of When Women Feel Least Attractive,” AdWeek, October 2, 2013, http://www.adweek.com/news/advertising-branding/marketers-should-take -note-when-women-feel-least-attractive-152753. 22 Lessig, Code: And Other Laws of Cyberspace, Version 2.0. 23 Amy Mitchell, Jeffrey Gottfried, and Katerina Eva Matsa, “Millennials and Political News,” Pew Research Center, June 1, 2015, http://www.journalism.org/2015/06/01/millennials-political -news/. 24 Mirani, “Millions of Facebook Users have No Idea They’re Using the Internet.” 25 World Wide Web Foundation, Women’s Rights Online: Translating Access into Empowerment (Wash- ington, D.C.: World Wide Web Foundation, 2015), http://webfoundation.org/wp-content/ uploads/2015/10/WomensRightsOnlineWF_Oct2015.pdf. 26 See David D. Clark, “The Contingent Internet,” Dædalus 145 (1) (Winter 2016), 9–17. 27 Motahhare Eslami, Aimee Rickman, Kristen Vaccaro, Amirhossein Aleyasen, Andy Vuong, Karrie Karahalios, Kevin Hamilton, and Christian Sandvig, “‘I Always Assumed That I Wasn’t Really That Close to [Her]’: Reasoning about Invisible Algorithms in the News Feed,” in Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems (New York: Association for Com- puting Machinery, 2015), 153–162, http://www.researchgate.net/pro½le/Kevin_Hamilton/ publication/275353888__I_always_assumed_that_I_wasn’t_really_that_close_to_her___ Reasoning_about_Invisible_Algorithms_in_News_Feeds/links/553aa2fd0cf245bdd764475f.pdf. 28 Robert M. Bond, Christopher J. Fariss, Jason J. Jones, Adam D. I. Kramer, Cameron Marlow, Jaime E. Settle, and James H. Fowler, “A 61-Million-Person Experiment in Social Influence and Political Mobilization,” Nature 489 (7415) (2012): 295–298, http://doi.org/10.1038/nature11421; and Jonathan Zittrain, “Facebook Could Decide an Election Without Anyone Ever Finding Out,” The New Republic, June 1, 2014, https://newrepublic.com/article/117878/information-½duciary -solution-facebook-digital-gerrymandering. 29 Lorenzo Coviello, Yunkyu Sohn, Adam D. I. Kramer, Cameron Marlow, Massimo Franceschetti, Nicholas A. Christakis, and James H. Fowler, “Detecting Emotional Contagion in Massive So- cial Networks,” PLoS ONE 9 (3) (2014): e90315, http://doi.org/10.1371/journal.pone.0090315. 30 Robert Epstein and Ronald E. Robertson, “The Search Engine Manipulation Effect (seme) and Its Possible Impact on the Outcomes of Elections,” Proceedings of the National Academy of Sciences 112 (33) (2015): E4512–E4521, http://doi.org/10.1073/pnas.1419828112. 31 Zeynep Tufekci, “The Medium and the Movement: Digital Tools, Social Movement Politics, and the End of the Free Rider Problem,” Policy & Internet 6 (2) (2014): 202–208, http://doi.org/10 .1002/1944-2866.POI362. 32 Zeynep Tufekci, “Algorithmic Harms beyond Facebook and Google: Emergent Challenges of Computational Agency,” Colorado Technology Law Journal [formerly Journal on Telecommunications and High Technology Law] 13 (2) (2015): 203–218; and Zeynep Tufekci and Deen Freelon, “In- troduction to the Special Issue on New Media and Social Unrest,” American Behavioural Scientist 57 (7) (2013): 843–847. 33 Tufekci, “Algorithmic Harms beyond Facebook and Google.” 34 Benkler, The Wealth of Networks. 35 Yochai Benkler, The Penguin and the Leviathan: How Cooperation Triumphs over Self-Interest (New York: Crown Publishing Group, 2011). 145 (1) Winter 2016 77 As the 36 Ethan Zuckerman, “The Internet’s Original Sin,” The Atlantic, August 14, 2014, http://www Pirates .theatlantic.com/technology/archive/2014/08/advertising-is-the-internets-original-sin/ Become 376041/. CEOs: The Closing 37 Zeynep Tufekci, “Facebook, Network Externalities, Regulation,” Technosociology, May 26, of the Open 2010, http://technosociology.org/?p=137. Internet 38 Zeynep Tufekci, “Mark Zuckerberg, Let Me Pay for Facebook,” The New York Times, June 4, 2015, http://www.nytimes.com/2015/06/04/opinion/zeynep-tufekci-mark-zuckerberg-let-me -pay-for-facebook.html. 39 Dan Goodin, “Critical Crypto Bug in Openssl Opens Two-Thirds of the Web to Eavesdropping,” Ars Technica, April 7, 2014, http://arstechnica.com/security/2014/04/critical-crypto-bug-in -openssl-opens-two-thirds-of-the-web-to-eavesdropping/; and Jose Pagliery, “Your Internet Se- curity Relies on a Few Volunteers,” cnn Money, April 18, 2014, http://money.cnn.com/2014/ 04/18/technology/security/heartbleed-volunteers/index.html. 40 See Jonathan Zittrain, The Future of the Internet–And How to Stop It (New Haven, Conn.: Yale Uni- versity Press, 2008). 41 Tim Wu, The Master Switch: The Rise and Fall of Information Empires (New York: Knopf Doubleday Publishing Group, 2010). 42 Debora L. Spar, Ruling the Waves: From the Compass to the Internet, a History of Business and Politics along the Technological Frontier (New York: Mariner Books, 2003). 43 Tufekci, “Engineering the Public”; and Zittrain, “Facebook Could Decide an Election Without Anyone Ever Finding Out.” 78 Dædalus, the Journal ofthe American Academy of Arts & Sciences