CASE STUDY
Monitoring Election Advertising on Facebook
This project used the Australian Ad Observatory to monitor political advertising during the 2022 election. This project observed Facebook advertising to identify any political or issue advertising that wasn’t properly authorised and that may have tried to sneak through undetected.
The researchers did not locate any significant or widespread ‘dark ad’ campaigns throughout the election. With the caveat that they only examined ads from the 1,700 citizen scientists at the time, who had installed the ad plugin, and that they couldn’t see mobile-only ads.
This is not to say that there aren’t significant issues regarding false and misleading claims being made in advertising, or that the transparency provided by the platforms is adequate (far from it), but at least during this project the political ads seen could all be traced back to an authorised source.
A significant part of the work in the Ad Observatory has been to develop machine vision techniques to detect political logos and other signifiers that may help us locate unauthorised ads. It is great therefore to see that colleagues at The Guardian have also been experimenting with the use of machine vision to detect political messaging techniques, such as the use of novelty cheques, cute furry animals, and hi-vis workwear. With the continued fragmentation of our media landscape, these new techniques all play an important role in helping us understand the pulse of the political campaign and hold our politicians to account.
RESEARCHERS
Prof Daniel Angus
Chief Investigator
Prof Axel Bruns
Chief Investigator
