Micro-Targeting – Just Like the Post Office

Facebook’s excuses and promises are misleading: We do not need a security fix, but new rules for political micro-targeting

“We have a responsibility to not just build tools, but to make sure those tools are used for good”, Mark Zuckerberg explained to the US-Congress. Sounds like an excuse. In fact, we should be rather alarmed.

Zuckerberg seems to think that the whole story about Cambridge Analytica and Russian propaganda on Facebook is about security breaches, data privacy and technological fixes (for which Facebook indeed should claim responsibility). But what we should be much more concerned about is the manipulation of political opinion.

Mark Zuckerberg would like to determine these rules himself – pretending that it’s just a matter of installing the right technological fix. Let’s not let him get away with this!

With the advent of online micro-targeting, new rules for political advertising are becoming more and more of a real necessity. Certainly, Mark Zuckerberg would like to determine these rules himself – pretending that it’s just a matter of installing the right technological fix. That’s what his words before the US-Congress imply. Let’s not let him get away with this! We – society and politics – should discuss and make the necessary rules. Zuckerberg’s job should be to comply with what has been determined.

Micro-targeting: A short introduction
Micro-targeting is basically personalised advertisement. At Facebook, those advertisements are known as dark posts – “dark” because the specific content and the number of such ads are not visible to the general public. There is nothing illegal about micro-targeting. When, during an election campaign, local politicians come to visit voters at their homes, this is regarded rather as sincere interest in the voters’ needs and opinions. But this is not always so. This spring, it became known that the former monopolist Deutsche Post, which is still the major provider of postal services in Germany, sold demographic data on a very micro-level both to the conservative party (CDU) and the liberal (or libertarian) party, FDP. With the help of that data, potential swing-voters could be identified as a target for visits by politicians. Even though, from the point of data protection, there was nothing illegal about this procedure, the story produced a public outcry in Germany.

Problems
Micro-targeting has been an established practice for some time. But with the advent of Big Data and social media, the scope and relevance of micro-targeting has reached a new level. Social media makes it possible to assess more thorough data about individual preferences and, moreover, about interpersonal relationships than ever before. With the help of artificial intelligence, vast amounts of data can be analyzed in in order to gain insights about individual profiles and ways to manipulate people. Also, with the help of artificial intelligence, individually tailored messages can be composed by robots. The results can, again with the help of social media, be delivered to highly specified groups and subgroups of voters and consumers with little cost.

Researchers investigating the impact of micro-targeting in political advertising stress that, from the point of view of public interest, there are both risks and benefits associated with the practice. “Political micro-targeting has a positive potential for activating and engaging people. Tailored messages can have some appeal with voters, since they might be understood as more personally relevant”, Tom Dobber and Natali Helberger (University of Amsterdam) write in the Internet Policy Review.

On the downside, there is the risk that voters’ opinions can be manipulated with highly sophisticated instruments – without that manipulation becomes noticeable. But the line between legitimate information and illegitimate manipulation is hard to draw both on a conceptual and an empirical level. Other risks are easier to grasp. One concerns the fragmentation of the public sphere. In a scholarly article published in the Utrecht Law Review, a team of Belgian and Dutch researchers comes up with the following example:

[A] politician has information that suggests that Alice dislikes immigrants. The politician shows Alice personalised ads. Those ads say that the politician plans to curtail immigration. The politician [also] has a profile of Bob that suggests that Bob has more progressive views. The ad targeted at Bob says the politician will fight the discrimination of immigrants in the job market. The ad does not mention the plan to limit immigration. […] Hence, without technically lying, the politician could say something different to each individual.

The example is telling insofar as it stresses what is specific about political advertisement. A political party can be regarded as a very complex product – much more complex than most of the products we usually buy! Because of this complexity, political advertising particularly benefits from being tailored to the needs of individual consumers. But what makes parties different from products is that ‘buying’ a party (by means of voting) has a much broader effect than consuming a product. Products might well have functions that we don’t need.

Intermediaries who provide data and allow access to voters via their networks gain “unprecedented power to set prices and dictate the terms upon political parties”

It doesn’t have much impact to the rest of the world if we don’t use these functions. In political advertising, the same procedure qualifies as rather unethical. To stay with the example: Fighting discrimination of immigrants in the job market might be a feature that Alice doesn’t care for. But by voting for the party (because of the plans to limit immigration), she makes a contribution to the effect that that very function exerts an influence upon the world. Thus, micro-targeting makes it possible for politicians to influence opinion by giving an incorrect impression about important aspects of the party’s program.

Another aspect which is highly problematic about online micro-targeting: Intermediaries who provide data and allow access to voters via their networks gain “unprecedented power to set prices and dictate the terms upon political parties”, as the article in the Utrecht Law Review points out. This power would even put them into a position “to provide services to political parties at their own rate and discretion” (see the Netopia report “The Citizens’ Internet”, p. 33 for further examples and literature on this topic).

Solutions
On the technology front, not really much is to be expected in terms of solutions for these problems. Cambridge Analytica was able to collect the data of 87 million or more Facebook users with the help of an app developed by the researcher Aleksandr Kogan. As a reaction to the public accusations, Facebook has announced a series of changes regarding the way third-party developers can interact with Facebook via APIs. (APIs are the digital interfaces through which apps and third parties can interact with and extract data from the platform.) Will this help? Maybe not too much. Not listed among the changes is, for instance, the Graph API. A team of researchers from the Technical University of Munich demonstrated just recently, how, with the help of the Graph API, individual data can be extracted from Facebook in order to target people with individually tailored messages – disregarding data protection rules which are explicitly prohibiting such practice. This, ironically, merely underlines what Zuckerberg himself confessed in a CNN-interview recently: “You never, ever solve security. It’s an arms race.”

But, as said before: the case “Cambridge Analytica” is neither about security, nor about mere technological fixes. Also, mere transparency, as announced by Facebook as a strategy to deal with challenges in political advertising, will certainly not do the job in respect to the mentioned problems. Still, there are at least some minor proposals of what can be done. Tom Dobber and Natali Helberger point out in the Internet Policy Review:

“Right now, we have the somewhat peculiar situation that consumers as addresses of commercial micro-targeting enjoy rather far-reaching rights to information, fairness in advertising, protection from misleading claims and unfair practices, but that these rules are not applicable to political micro-targeting.”

Protection from misleading claims and rights to information – that’s just one example of where a perhaps rather easy, non-technological fix could be brought into operation. Rules to ensure that all political parties equally have access to advertising in social networks for a reasonable price would be another candidate for a useful solution. In fact, this would be nothing else than a translation of well-established regulation concerning political campaign advertising on TV to the realm of social media (see again The Citizens’s Internet, p. 6).

Let’s bother no longer with Zuckerberg’s excuses and promises. There is work to do. Facebook’s CEO probably will be of little use getting it done.