[论文解读] Ad Delivery Algorithms: The Hidden Arbiters of Political Messaging

Muhammad Ali, Piotr Sapieżyński+3 more|arXiv (Cornell University)|Dec 9, 2019

Social Media and Politics参考文献 26被引用 33

一句话总结

The paper experimentally analyzes Facebook’s ad delivery algorithms and shows that political ad content, even with identical targeting, influences which users see the ads and at what cost, potentially creating political filter bubbles.

ABSTRACT

Political campaigns are increasingly turning to digital advertising to reach voters. These platforms empower advertisers to target messages to platform users with great precision, including through inferences about those users' political affiliations. However, prior work has shown that platforms' ad delivery algorithms can selectively deliver ads within these target audiences in ways that can lead to demographic skews along race and gender lines, often without an advertiser's knowledge. In this study, we investigate the impact of Facebook's ad delivery algorithms on political ads. We run a series of political ads on Facebook and measure how Facebook delivers those ads to different groups, depending on an ad's content (e.g., the political viewpoint featured) and targeting criteria. We find that Facebook's ad delivery algorithms effectively differentiate the price of reaching a user based on their inferred political alignment with the advertised content, inhibiting political campaigns' ability to reach voters with diverse political views. This effect is most acute when advertisers use small budgets, as Facebook's delivery algorithm tends to preferentially deliver to the users who are, according to Facebook's estimation, most relevant. Our findings point to advertising platforms' potential role in political polarization and creating informational filter bubbles. Furthermore, some large ad platforms have recently changed their policies to restrict the targeting tools they offer to political campaigns; our findings show that such reforms will be insufficient if the goal is to ensure that political ads are shown to users of diverse political views. Our findings add urgency to calls for more meaningful public transparency into the political advertising ecosystem.

研究动机与目标

• Assess whether Facebook’s ad delivery skews political ad delivery along political lines despite identical targeting and budgets.
• Evaluate how ad content and targeting interact to influence reach and costs for political campaigns.
• Investigate whether delivery skews persist under different campaign objectives and budgets.
• Provide evidence to inform policy discussions on transparency in political advertising platforms.

提出的方法

• Run controlled political ad campaigns on Facebook for Democratic and Republican candidates under carefully matched targeting and budgets.
• Use Custom Audiences and detailed targeting to infer delivery skew by political leaning through proxies from voter and donor records.
• Exhaust audiences by forcing delivery to all target users to assess reach.
• Compare delivery and pricing when ads are neutral versus when they resemble candidate campaigns.
• Analyze Facebook’s reporting to determine exposure differences and costs across sub-populations.

实验结果

研究问题

• RQ1 Does Facebook preferentially deliver political ads to users aligned with the ad content's inferred political stance, even with identical targeting?
• RQ2 How does ad delivery affect reach and pricing across different political leanings and audience types (voters, donors)?
• RQ3 Are the observed delivery skews amplified by Facebook’s political targeting features or by the ad content itself?
• RQ4 Do reforms restricting targeting tools mitigate the delivery biases observed in political ads?

主要发现

• Ads with Democratic content reached a higher share of Democrats than ads with Republican content reached of Democrats, despite identical targeting.
• Neutral control ads produced a more balanced reach across political groups compared to candidate-specific ads.
• Delivery skew was stronger when targeting by Facebook’s political inference features (e.g., Liberal vs Conservative), with substantial differences in audience reach.
• Ad delivery skew also influenced cost, with certain audiences being more expensive to reach depending on alignment with the ad content.
• Skews persisted even when the ad content was neutral but framed to appear as a candidate page, suggesting a priori platform-driven delivery decisions.
• The study raises concerns that delivery algorithms can contribute to political polarization and informational filter bubbles, and that platform reforms may be insufficient without transparency.