Political microtargeting could undermine electoral democracy


I  n early 2018, the Facebook/Cambridge Analytica scandal made headlines and introduced a large public  to the possibilities of targeted political advertising online. On the face of it, the scandal concerned a massive data breach, caused by a data analytics firm that had harvested the data of 50 million Facebook profiles without authorisation. Yet the firm had worked for both the Trump election campaign in the US and the Leave campaign before the Brexit referendum in the UK. Therefore the scandal turned the spotlight on the practice of microtargeting and its potential for political manipulation. Can microtargeting endanger or undermine electoral democracy in the long run?

Political microtargeting uses available data on geographic, demographic, behavioural or psychological characteristics of subgroups or individuals to decide about the best addressees and most effective content of campaign communication. The idea originates in commercial marketing, and all major online platforms today base their business model on it. They promise (mostly commercial) clients to spend their scarce advertising money on messages to people who are likely to be receptive to them. The essential difference is between broadcasting something to everyone and narrowcasting it only to those actually interested in it. Since political campaigns have limited financial resources, microtargeting seems to make a lot of sense for politicians and parties as well. Even voters could find it beneficial to only receive messages on issues they care about, as the former CEO of Cambridge Analytica suggested in an inquiry by a committee of the British House of Commons.


Exemple tiré du site Dealers United

Much of the current uneasiness about political microtargeting seems to stem from the perception that it actually does more than effectively allocate spending on advertising. Indeed Cambridge Analytica’s sales pitch before the eruption of the scandal claimed that it was able to map psychological models to data from Facebook and increase the effectiveness of adverts by adjusting content and presentation to the psychological make-up of individuals.

Whether it is actually possible to reliably predict individuals’ personality and interests based on Facebook likes and similar data is still controversial. But even provided that is the case, the actual impact of targeted adverts on individual political behaviour is very uncertain.

In a commercial context, psychological targeting aims at increasing the number of clicks on adverts and eventually the number of purchases. But choosing to buy a particular product is quite different from choosing to vote for a particular party, candidate or referendum option. When you see an ad for running shoes, the fact that you are targeted with it suggests that you have been identified as the kind of person who likes running and may therefore be persuaded to soon buy new shoes of the advertised brand. Commercial advertising is about getting you to make a purchase that you had not necessarily planned to make and/or to opt for a particular brand of essentially the same product once you are set on a purchase.

Political elections by contrast offer options for only one type of product and have a fixed date at which the decision has to be made. It is a very complex type of product, bundling ideological beliefs, specific pledges and politicians who claim to have competences relevant to governing. Therefore it is a very long way from persuading people to click on a political advert to persuading them to pick a party or candidate on election day that they would otherwise not have picked.

The analogy to commercial advertising suggests that targeted political adverts may be more effective in getting groups of people to vote who had not necessarily intended to. If one party or candidate manages to do that better than their competitors, it will certainly influence the election result. But it is hard to see why the use of targeted advertising to this aim should be considered more manipulative than other attempts to reach out to habitual non-voters.

Does the uncertainty of a direct impact on the psychology of individual voting decisions mean that microtargeting is actually of little consequence for electoral democracy? Not at all. The crucial point is that targeted advertising is per definition non-public communication. This poses no problem in the economic realm, which is made up of private decisions about purchases, savings and investments. But it is of major concern in democratic politics, where political communication and decision-making concern collectivities and used to be exposed to public scrutiny.

Such scrutiny has a disciplining effect. First of all, it functions as a check against outright lies. They are likely to get noticed and countered in the same general communication space in which they are told, which makes them less effective in the first place. But the potential for public scrutiny also restricts possibilities to tell every potential voter exactly what they want to hear. Public campaign communication can either remain very general to cater to the diversity of potential voters or address subgroups with more concrete messages, which those addressed may like but other potential voters not. Either way, public political communication has the side-effect of conveying that a democracy cannot give everyone exactly what they would like to get.

Publicité d'une entreprise vantant les mérites du micro-targetingTargeted advertising is based on the opposite idea and since it is not public, heavy reliance on it in political campaigning could foster irreconcilable expectations in the electorate. The result could be a ‘digital clientelism’, in which everybody is promised what they want, but political outcomes are constantly disappointing for that very reason.

In sum, political microtargeting could indirectly pose a serious threat to the effectiveness of electoral democracy. That means that the massive use of individual-level data for advertising purposes is a problematic development no matter whether the individuals feel that they have nothing to hide and do not care. But a thorough analysis of possible implications of microtargeting can help to identify useful countermeasures. A legal requirement to simultaneously publish targeted advertising content on official campaign websites and in a public archive maintained by the platforms could for example re-invoke the disciplining effect of public scrutiny. The obligation to inform targeted people on what grounds they have been selected for seeing a specific advert would be another measure. Legal limitations to the current business model of online platforms could be even more radical. In any case, it is high time to initiate regulation that adjusts legal standards for political campaign communication to the new technological capabilities of big data analytics.

reagissez !

Ajouter un commentaire