The Bad Ads Project

Online advertisements are an unavoidable fact of the modern web — they are embedded in and financially support the majority of content websites. As such, the content in these ads makes up a significant fraction of the information that people encounter as they browse the web. Depending on the ad, the content can range from useful (e.g., ads for products a user truly wants to buy) to annoying (e.g., poorly designed product ads) to deceptive (e.g., misleading clickbait headlines that look like real news headlines) to downright dangerous (e.g., scams). Despite its pervasiveness, the online ad ecosystem is opaque.

Different people may see vastly different online ads, and many different ad networks and other intermediaries are involved in the creation, placement, and delivery of ads. This lack of transparency makes it hard to publicly detect, audit, or analyze potentially problematic ad content, the advertisers and ad networks who distribute this content, and the sites it appears on.

To provide more transparency to this ecosystem – for end users, researchers, and regulators – the approach is a Public Interest Ad Archive.

Led by Innovation Lab Fellow, Franzi Rosner, the Bad Ads Project continued the investigation of problematic ads on the web with University of Washington’s Security and Privacy Lab. Research included a deep dive of the ad ecosystem to understand and characterize its technical mechanisms, the content pushed through it, the harms posed to consumers (including privacy, financial, attention, emotional, etc.) Since then, multiple areas of concern have been identified such as the high volume of misleading “native” advertising on online news websites, internet users’ dissatisfaction with clickbait techniques, and deceptive and manipulative design in political, health, and software ads.

The end result of the investigation- a prototype of the Public Interest Ad Archive.

The Public Interest Ad Archive aims to provide transparency to the online advertising ecosystem. The archive collects and makes available to researchers, journalists, and the public information about ads appearing on different sites over time, allowing investigations into specific advertisers or ad campaigns or ad networks, the potentially harmful impacts of particular types of ads, longitudinal studies or deep dives into the ad ecosystem around sensitive current events, etc.

Next up, the Bad Ads Project goes from memo to demo. The team is currently developing technical designs (e.g., in browsers) and/or policies to improve the ad ecosystem for end users, while balancing its financial importance for the content that depends on it. Big milestones ahead include:

• Completing a public-ready version of the archive to support data exploration
• Launching a citizen science campaign to collect and label data