< Back to Blog

Controlling the Power of Social Media Data

Set-1-Blog-images-Ipad

Nearly 10 years ago, I ran research and development for Data Persona, a startup that provided personalization as a service for media web sites, using customer data obtained from several sources including social media. To test the product, we built a dummy Facebook app and asked our company’s employees to install it. When we checked our database, we were amazed to discover that Facebook had given us good personalization information (e.g., groups and likes) not only about me, but also about all my Facebook friends. At the time I assumed it was just a bug in the Facebook API. I did not complain because I was happy to get more user data to train our personalization algorithm.

It seems Data Persona was not the only beneficiary of Facebook’s “generosity.” In 2014 a psychologist created a personality quiz Facebook app and convinced 250,000 people to install it. From those 250,000 installations he was able to harvest data about more than 50 million Facebook users. This data eventually ended up in the hands of Cambridge Analytica, which used it to create psychometric profiles of American voters, and then deployed that information to enable targeted campaign ads for the U.S. Presidential election. This has created a stir in the media, as the world now discovers the potential capability of social media to influence human behavior.

Full disclosure: I am an active Facebook user. I post photos of my travels, and I read my news feed every day. As news of Cambridge Analytica’s influence on U.S. election advertising broke, I expected it to have some impact on the conversation among my Facebook friends, up to and including decisions to leave Facebook entirely. But, much to my surprise, there was no reaction at all – just the usual stream of links, rants and photos. So, I posted my own observation: “I am amazed by the lack of discussion (on my feed at least) about Facebook’s inability to preserve users’ privacy.” The responses I got were mainly bored indifference: “I always assumed Facebook was using data about me, but I consider it a fair deal in return for getting a free service.” As one friend put it, “Once you install freeware, you are the product.”

I believe this argument misses the point. The 250,000 Facebook users who actually installed the personality quiz app certainly should not be surprised that their social data was sold. That’s the deal they implicitly made when they installed the app. But, what about their friends who comprise the rest of the 50 million people whose social data was sold? They did not knowingly give up their social data to Cambridge Analytica, and certainly had no idea that this could happen as a result of accepting a friend invitation from one of the app installers.

Facebook has admitted that this kind of data usage should never have been allowed to happen and has promised to take measures to ensure that it cannot happen again. However, I believe this self-regulation is not enough. In this case, government regulation is required for the following reasons:

  1. Governments regulate areas where public safety is at risk, e.g., emission controls for cars and safety standards for nuclear reactors. Social media data fits into this category, because:
    1. It provides such a complete picture of who you are, based on perfect and permanent tracking of all your likes, dislikes, conversations and affiliations.
    2. Analytical tools have progressed to the stage where they can use this data to not only predict your actions (predictive analytics), but also to persuade you to take a specified action by controlling the information to which you are exposed (persuasive analytics).
    3. The persuasive power of social media data puts it in same category as other aspects of public welfare and safety that the government regulates.
  2. Governments regulate monopolies to ensure fair practices. Social media (like much of high tech) is effectively a monopoly, due to its winner-take-all nature. Potential competitors who manage to overcome this network effect are typically purchased by the 500-pound gorilla before they can develop into full-fledged competition.
  3. The alternative to government regulation would be self-regulation. I do not believe this can work because social media companies have an inherent conflict of interest that clouds their judgement. Facebook makes its money by harvesting your social information and selling it to advertisers in the form of highly targeted ads. The more data Facebook can harvest, the better the ad targeting, and one of the best ways to determine your interests is to examine the interests of your friends, because birds of a feather flock together. This kind of logic led Facebook to make information about app users’ friends available via their application programming interface (API). Even if they have fixed this security hole, they are still subject to the same conflict of interest that makes such holes possible. For example, Facebook’s troubles have not stopped Snapchat from announcing the launch of third party apps on their social media platform.

Given the power of social media data and the inability of social media companies to self-regulate, government regulation is needed. To be clear, I do not mean that the government should collect and analyze all personal data from social networks and then decide who can access it. I mean that the government should ensure that such powerful data is not aggregated at all, via regulation and inspection. I am aware of how inefficient and confining such regulation can be, yet I see no better alternative, short of ditching social media entirely, and that isn’t happening, to judge from my Facebook friends’ response to Cambridge Analytica.