The Madness of the Crowd
Par Tim Hwang (mars 2017)
As the Trump Administration enters its first hundred days, the 2016 election and its unexpected result remains a central topic of discussion among journalists, researchers, and the public at large.
It is notable the degree to which Trump’s victory has propelled a broader, wholesale evaluation of the defects of the modern media ecosystem. Whether it is “fake news,” the influence of “filter bubbles,” or the online emergence of the “alt-right,” the internet has been cast as a familiar villain: enabling and empowering extreme views, and producing a “post-fact” society.
This isn’t the first time that the internet has figured prominently in a presidential win. Among commentators on the left, the collective pessimism about the technological forces powering Trump’s 2016 victory are matched in mirror image by the collective optimism about the technological forces driving Obama’s 2008 victory. As Arianna Huffington put it simply then, “Were it not for the Internet, Barack Obama would not be president. Were it not for the Internet, Barack Obama would not have been the nominee.”
But whereas Obama was seen as a sign that the new media ecosystem wrought by the internet was functioning beautifully (one commentator praised it as “a perfect medium for genuine grass-roots political movements”), the Trump win has been blamed on a media ecosystem in deep failure mode. We could chalk these accounts up to simple partisanship, but that would ignore a whole constellation of other incidents that should raise real concerns about the weaknesses of the public sphere that the contemporary internet has established.
This troubled internet has been around for years. Fears about filter bubbles facilitating the rise of the alt-right can and should be linked to existing concerns about the forces producing insular, extreme communities like the ones driving the Gamergate controversy. Fears about the impotence of facts in political debate match existing frustrations about the inability for documentary evidence in police killings—widely distributed through social media—to produce real change. Similarly, fears about organized mobs of Trump supporters systematically silencing political opponents online are just the latest data point in a long-standing critique of the failure of social media platforms to halt harassment.
One critical anchor point is the centrality of the wisdom of the crowd to the intellectual firmament of Web 2.0: the idea that the broad freedom to communicate enabled by the internet tends to produce beneficial outcomes for society. This position celebrated user-generated content, encouraged platforms for collective participation, and advocated the openness of data.
Inspired by the success of projects like the open-source operating system Linux and the explosion of platforms like Wikipedia, a generation of internet commentators espoused the benefits of crowd-sourced problem-solving. Anthony D. Williams and Don Tapscott’s Wikinomics (2006) touted the economic potential of the crowd. Clay Shirky’s Here Comes Everybody (2008) highlighted how open systems powered by volunteer contributions could create social change. Yochai Benkler’s The Wealth of Networks (2006) posited a cooperative form of socioeconomic production unleashed by the structure of the open web called “commons-based peer production.”
Such notions inspired movements like “Gov 2.0” and projects like the Sunlight Foundation, which sought to publish government data in order to reduce corruption and enable the creation of valuable new services by third parties. It also inspired a range of citizen journalism projects, empowering a new fourth estate.
The platforms inspired by the “wisdom of the crowd” represented an experiment. They tested the hypothesis that large groups of people can self-organize to produce knowledge effectively and ultimately arrive at positive outcomes.
In recent years, however, a number of underlying assumptions in this framework have been challenged, as these platforms have increasingly produced outcomes quite opposite to what their designers had in mind. With the benefit of hindsight, we can start to diagnose why. In particular, there have been four major “divergences” between how the vision of the wisdom of the crowd optimistically predicted people would act online and how they actually behaved.
First, the wisdom of the crowd assumes that each member of the crowd will sift through information to make independent observations and contributions. If not, it hopes that at least a majority will, such that a competitive marketplace of ideas will be able to arrive at the best result.
Second, collective intelligence requires aggregating many individual observations. To that end, it assumes a sufficient diversity of viewpoints. However, open platforms did not generate or actively cultivate this kind of diversity, instead more passively relying on the ostensible availability of these tools to all.
Third, collective intelligence assumes that wrong information will be systematically weeded out as it conflicts with the mass of observations being made by others. Quite the opposite played out in practice, as it ended up being much easier to share information than to evaluate its accuracy. Hoaxes spread very effectively through the crowd, from bogus medical beliefs and conspiracy theories to faked celebrity deaths and clickbait headlines.
Fourth, collective intelligence was assumed to be a vehicle for positive social change because broad participation would make wrongdoing more difficult to hide. Though this latter point turned out to be arguably true, transparency alone was not the powerful disinfectant it was assumed to be.
The ability to capture police violence on smartphones did not result in increased convictions or changes to the underlying policies of law enforcement. The Edward Snowden revelations failed to produce substantial surveillance reform in the United States. The leak of Donald Trump’s Access Hollywood recording failed to change the political momentum of the 2016 election. And so on. As Aaron Swartz warned us in 2009, “reality doesn’t live in the databases.”
Ultimately, the aspirations of collective intelligence underlying a generation of online platforms proved far more narrow and limited in practice. The wisdom of the crowd turned out to be susceptible to the influence of recommendation algorithms, the designs of bad actors, in-built biases of users, and the strength of incumbent institutions, among other forces.
The resulting ecosystem feels deeply out of control. The promise of a collective search for the truth gave way to a pernicious ecosystem of fake news. The promise of a broad participatory culture gave way to campaigns of harassment and atomized, deeply insular communities. The promise of greater public accountability gave way to waves of outrage with little real change. Trump 2016 and Obama 2008 are through-the-looking-glass versions of one another, with the benefits from one era giving rise to the failures of the next.
To the extent that the vision of the wisdom of the crowd was naive, it was naive because it assumed that the internet was a spontaneous reactor for a certain kind of collective behavior. It mistook what should have been an agenda, a ongoing program for the design of the web, for the way things already were. It assumed users had the time and education to contribute and evaluate scads of information. It assumed a level of class, race, and gender diversity in online participation that never materialized. It assumed a gentility of collaboration and discussion among people that only ever existed in certain contexts. It assumed that the simple revelation of facts would produce social change.
In short, the wisdom of the crowd didn’t describe where we were, so much as paint a picture of where we should have been going.
The vision of collective participation embedded in the idea of the wisdom of the crowd rests on the belief in the unique potential of the web and what it might achieve. Even as the technology evolves, that vision—and a renewed defense of it—must guide us as we enter the next decade.