Political Micro-targeting: An Experiment on Facebook's Political Targeting Tags.

-- By KuanHsinHuang - 06 Dec 2019

I An Observation of Facebook’s Ad-targeting System

Political micro-targeting ads on social media has become a potent tool for campaigns and even disinformation operations. Although it is an oversimplification to contend that one can simply incite behaviors of human beings by simply “pushing buttons,” there is no doubt that the fundamental values in deliberative democracy have largely come under threat by such practices.

In addition to such normative concerns, this paper provides a set of descriptive observations to further support the initiatives of curbing political micro-targeting ads on social media. The approach taken here is to do experiments on Facebook’s ad-targeting system (Facebook Ads Manager). By inserting an array of political issues in the system, this paper explored whether such issues are available as a “tag” for advertiser to target with. For instance, the phrase “Conspiracy Theory” has been established by Facebook as a tag which advertisers may choose to include in their micro-targeting settings. After testing the availability of an array political issues, this paper finds that Facebook has failed to maintain a coherent and unbiased system of “target-ability” in political issues.

II The Results

This experiment searched the “target-ability” of topics such as social issues, controversies, as well as taboos and stereotypical symbols. The results of this experiment are as follows:

Targetable issues

• Taboo: Conspiracy Theory; 4Chan; Breaking the taboo.

• Right-leaning: National Rifle Association; Anti-abortion movement; Climate change denial; Make America Great Again; Sean Hannity and Witch-Hunt.

• Left-leaning: Same-sex marriage; Racial integration; Bernie Sanders; Right of asylum; Police brutality; Military–industrial complex; Separation of church and state; Hating Breitbart; Ruth Bader Ginsburg; Impeachment; The New Jim Crow; Black Lives Matter; and Occupy Wall Street.

• Others: WikiLeaks? ; Edward Snowden; Stand for Israel; and Tariff.

Not Targetable issues

• Metoo; Mueller report; Prison–industrial complex; Charlottesville; Nihilism.

Here's a table that compares targetability among opposite/related issues

III Implications

Through this preliminary experiment, we observe much asymmetries and arbitrary. Asymmetry happens when only one particular side of an issue is targetable and the other side is not. For instance, among slogans of different campaigns in the 2016 U.S. presidential election, only Trump’s “Make America Great Again” was established as a targetable option. Same goes with NRA, same-sex marriage, and “Hating Breitbart” where there are no direct opposite tags available for targeting. Although determined political advertiser may still be capable of targeting the opposite side by targeting with other issue tags that speaks to the similar group of people, the absence of direct tags still makes a difference. Secondly, the social significance of an issues is not reflected in this system. It is unclear why “Metoo” movements and the “Mueller report” are not made targetable while other issues of similar or even less weight are.

This leads us to a series of questions: Why are some political issues available for targeting while others are not? What is the underling rationale? What is the process of making a political issue into a targeting option? Who are making those decisions? And what (if any) standards are implemented in such decision process? Clearly, Facebook fails to maintain coherency in the targeting options it chooses to provide. And its decision-making process is a “black box” that blocks the public from understanding the mechanism of this dominant tool for political influence. This situation is similar to the early days of Facebook’s content moderation, where decisions were made arbitrarily in absence of standards and established process. Although the newly launched “Oversight Board” still faces much criticism, at least content moderation has begun to move towards a more institutionalized process.

However, in the context of political moderating, implementing a systematic decision process still does not solve the problem. This is because the pursuit of symmetry and coherency will most likely lead to an ever-expanding coverage of targetable issues, which runs directly against our initial and ultimate goal of curbing political targeting. We should minimize the poison, not extend it just to make sure that everything is equally contaminated.

What we should do is: instead of making in more political issues available for targeting to match symmetry and coherency, we should abolish all the existing targeting tags of political nature. For example, the cure is not adding in “Feel the Bern,” “I’m with her,” and the like to balance out the existing “Make America Great Again,” but simply stop offering the ability to single out Trump supporters and make them exclusively subject to a particular paid political influence.

Conclusion

This paper attempts to raise two points: (1) The current micro-targeting system that is widely used to inject political influence is in itself biased, inconsistent and dubious, which adds another ground to reject political micro-targeting alongside normative concerns of deliberative democracy. (2) The solution should be abolishing all targetable political tags, which will not only resolve the asymmetry and coherency problem, but also ultimately mitigate the chances for political micro-targeting.