Autonomy online: Jacques Ellul and the Facebook emotional manipulation study

This paper seeks to understand what “autonomy” means for Facebook, and what Facebook means for “autonomy.” The motivation for this investigation is, first, the recent revelation that Facebook was manipulating its users for the purposes of “research,” and, second, the reaction of Facebook users to this revelation.

Facebook’s number of “active users” ¹ has been steadily increasing, from 100 million in the 3rd quarter of 2008 to 1.317 billion in the 2nd quarter of 2014 (Statista, 2014). The online publication on 2 June of Facebook’s manipulation study, “Experimental evidence of massive-scale emotional contagion through social networks” (Kramer et al., 2014: 8788–8790) has appeared to have in no way interrupted the growth of Facebook as the top social media website in the world. This may come as a surprise, and yet, according to the philosophy of technology of Jacques Ellul, this is exactly the result we should have expected.

For Ellul (2003: 386), technology is “autonomous.” Though we believe ourselves to be in control of technology, it is technology that is in control of us, as technology operates evermore independently of our intentions and desires while our intentions and desires become more and more dependent upon technology. This does not mean, as both critics and supporters of Ellul alike have claimed (Menninger, 1981), that Ellul was either an anti-technology extremist warning us against some sort of eventual robot uprising or a luddite who ascribed to technology supernatural powers. Rather, by focusing on the concept of “technique,” Ellul aimed to show us how the methods and aims of technology are increasingly becoming the methods and aims of humanity. This is seen most clearly in Ellul’s analysis of “efficiency,” as the efficiency that technology is meant to provide for the sake of our demands has come at the price of our making ourselves more efficient for the sake of technology’s demands. In other words, we have undoubtedly benefitted from technology, but at the same time we have received these benefits not by technology conforming to a human-centric world, but by humanity conforming to a techno-centric world.

The power that technology wields over us not only manifests itself individually—think, for example, of the current ubiquity of smart phones on restaurant tables, never being left out of sight even when we are not using them—but also manifests itself politically. As Ellul (2003: 388) argues, against the “simple view” that “the state decides, technology obeys,” we instead “have to ask who in the state intervenes, and how the state intervenes, i.e., how a decision is reached and by whom in reality not in the idealist vision.” To make a political decision about technology requires knowledge of and experience with that technology, a knowledge and experience had, not by politicians or by voters, but by technology experts. There is thus no democracy when it comes to technology as our decisions are made, either directly or indirectly, by technicians, technicians for whom any possibility of democratic “checks and balances” can only come from other technicians (Ellul, 2003: 388–9). This is not to say of course that politicians cannot make decisions independently of technicians, but should these decisions turn out to hamper the development of technology the decision is almost always reversed, especially as the progress of technology and the progress of the state become increasingly intertwined. For example, when the US Congress tried to enact the Stop Online Piracy Act in 2012, “the protest against these bills culminated in a shutdown for a day of English-language Wikipedia and many other prominent websites. As a result of the negative publicity, these pieces of legislation quickly lost public support and became stalled in Congress” (Spinello, 2014: 116).

Technology is likewise autonomous with regards to ethics. In much the same way that it might appear wrong to say that technology can have power over us, it might appear wrong to say that technology can be good or evil, right or wrong, ethical or unethical. In both cases it seems as if we are anthropomorphizing technology, viewing technology as if it were itself a creator of values rather than only a medium for value-laden activities. For this reason we tend to treat not only technology as neutral, but also the research and development of technology as neutral. Just as it is claimed that “guns don’t kill people, people kill people,” it is similarly believed that the progress of technology cannot harm people, only the application of technology can harm people. A technician can therefore claim, as Ellul (2003: 394) puts it, that “his research, quite simply, is.”

It is precisely such a claim to neutrality that one finds in Facebook data scientist Adam D. I. Kramer’s (2014) “brief public explanation”—an explanation which of course came in the form of a Facebook status update—of the Facebook manipulation study. Kramer does not apologize for carrying out the study, for “very minimally deprioritizing a small percentage of content in News Feed,” but instead for how the study was publicized, as he writes (Kramer, 2014) that “my coauthors and I are very sorry for the way the paper described the research and any anxiety it caused.” Kramer sees the issue in question as not the study, but rather as how people came to find out about the study. Hence Kramer concludes by stating that any lessons he has learned are due to “the reaction to this paper” rather than being due to the research itself. In other words, using technology to manipulate people for the sake of technology is perfectly reasonable so long as researchers avoid the one offense committed by Kramer and his team, the offense that “we didn’t clearly state our motivations in the paper” (Kramer, 2014).

By distinguishing research from its results, technology is preserved as an amoral domain, maintaining the idea that it is only the human domain that can be morally judged. Yet the more we live our lives in and through technology the more the amorality ascribed to technology comes to dominate humanity. If technology just is, then the more that technology becomes central to my life becomes the more I just am. That our children are raised on tablets, that our smart phones are how we navigate the world, that our Thanksgiving dinners are cut short so that we can stand in line for hours for Black Friday deals on the latest devices, all of this is justified as “what everyone else is doing,” as what is now “normal.” In this way technology comes to move out of the amoral space we have given it and takes on moral dimensions, though not in the sense that we begin to apply norms to technology, but rather that technology begins to supply norms to us. In other words, technology does not have to conform to humanity, humanity has to conform to technology.

This idea returns us to my claim at the outset that the reaction of Facebook users to the revelation of the Facebook manipulation study was predictable. Indeed it is quite likely that most people found out about the study through Facebook, and expressed their outrage (or lack thereof) to this discovery on Facebook. If a friend tells you that she has been manipulating you, even if it was in the hopes of trying to help you, the nature of your friendship will likely be damaged, if not simply ended. Yet if a Facebook data scientist tells you that he has been manipulating you, even if it was in the hopes of learning “how to provide a better service” (Kramer, 2014), the nature of your relationship to Facebook will likely be unchanged. The irony here is that, as Facebook is now the medium through which we interact most with our “friends,” it is probable that it is through Facebook that you would discover your friend’s manipulation and that it is through Facebook that you would respond to this discovery, whether by status updates, direct messages, or by “unfriending” her. We can take the risk of finding new friends, but so long as Facebook is the primary medium through which we make friends and keep friends, we cannot take the risk of finding new social networks. Yet as one’s popularity can now actually be measured in the number of friends one has on Facebook, ² it is likely that even if you should end your relationship with your manipulating friend, you would still remain “friended” to her on Facebook nonetheless.

This is the new normal of friendship, the new normal defined, not by the human users of Facebook, but by the technological means made available to humans by Facebook. The normative power of Facebook does not end at friendship however. As Facebook becomes more and more integrated into every facet of our daily lives—providing us with not only a medium for maintaining friendships, but also for news, for shopping, for entertainment—Facebook becomes further impervious to criticism and external influence. To remove oneself from Facebook’s social network is to risk not only becoming disconnected from friends and relatives, but to risk becoming disconnected from our now networked society.

While for humans we typically define autonomy in terms of agency, for Ellul, as we have seen, technology’s autonomy is defined instead by its ability to operate as a closed system, a system independent of our agency. Consequently, Facebook’s autonomy expands while the autonomy of its users contracts when, for example, we respond to the “Terms and Conditions” of websites like Facebook by laughing off concerns about not having read them before accepting them. The reason for this is not that we trust Facebook to not ask us to agree to terms that, if we did read them, we would find disagreeable, but that we know that our need to use Facebook outweighs any disagreements we might have with the site’s terms. Indeed we fully expect that we would disagree with Facebook’s terms if we did read them, for which reason we instead opt for this “out of sight, out of mind” approach, ceding our autonomy to Facebook in the process. While on South Park the ramifications of this approach can lead to our becoming components in an “Apple HumancentiPad,” ³ on Facebook this approach has led to our becoming “a Facebook lab rat.” As Kashmir Hill (2014b) points out in her article for Forbes entitled “10 Other Facebook Experiments on Users, Rated on a Highly-Scientific WTF Scale,” Facebook’s manipulation study is but one of many such experiments. In fact, according to former Facebook data scientist Andrew Ledvina, “experiments are run on every user at some point in their tenure on the site” (Hill, 2014a), and yet we only know about any of these experiments “because it’s been published” (Hill, 2014b).

This last point is especially relevant to the argument that the way to respond to Facebook’s experiments is by instituting review boards or oversight committees. Ryan Calo (2013), for example, suggests that companies like Facebook should create “Consumer Subject Review Boards (CSRBs)” that would operate like the business-equivalent of academic Institutional Review Boards and be composed of “a small internal committee comprised of employees with diverse training (law, engineering) and operated according to predetermined rules.” Similarly, Matthew Salganik (2014) calls for the creation of “Human-Subjects Research Oversight Committees (HSROCs) that would review, improve, and ultimately accept or reject plans for human-subjects research” based on the five principles that they be “restricted in scope, focused on balancing risks and benefits, transparent, dynamic, and diverse.” danah boyd (2014) on the other hand argues instead that “any company that manipulates user data create an ethics board,” a board that, “rather than simply consisting of board members” would “consist of scholars and users.”

That Facebook not only publishes their research, but uses such words in the titles of their publications as “massive-scale” and “contagion,” should be sufficient to tell us that these boards and committees would have little impact on either the growth of Facebook’s autonomy or the diminishment of ours. As Ledvina makes clear, experimenting on users “is just how a website works, everyone does this and everyone knows that everyone does this,” and he further adds, “I don’t see why people are all up in arms over this thing all of a sudden” (Hill, 2014a). Hence the idea that the research and development of technology can be policed by committees of either insiders, outsiders, or by both is simply untenable. Websites manipulate users, and as we have seen and as Ledvina refers to, we continue to use websites that manipulate us. So long as data scientists like Kramer and Ledvina view such manipulation as necessary to improve technology, and so long as users want technology to be improved, there is no reason for this situation to change.

This is precisely the relationship to technology that Ellul has described. As Ellul (2003: 395) writes, though “hitherto, man has always tried to refer his actions to a superior value,” this is no longer the case, as “this situation is vanishing for the sake of technology,” since “man in our society both discerns this autonomy demanded by the system (which can progress only if autonomous) and grants this system autonomy by accepting it as legitimate in itself.” Our primary worries concerning technology are not moral but practical, as the complaints about what technology should be doing for us drown out any complaints about what technology is doing to us. The revelation of Facebook’s manipulation study must therefore lead to soul-searching, not at Facebook, but amongst ourselves, as the dilemma we are facing here must be recognized as existential before it can meaningfully be posed as ethical. To do so we must ask ourselves whether we want to continue to live in a world where technological development not only overshadows human development, it is seen as the one true path to human development. Unfortunately, it is likely that, in much the same way that I found the call for papers that led me to write this paper because it was posted on Facebook, the only way to get an answer to this question would be to put it in the form of a Facebook poll.