Meta has no plans to expand its Variance Reduction System (VRS), a newly announced anti-discrimination advertising tool in the US, to Australia and New Zealand a source with knowledge of the matter said.
In 2019, Meta was charged by the US Department of Justice (DOJ) with enabling housing discrimination through its “Special Ad Audiences” tool which ran contrary to the Fair Housing Act (FHA).
However, despite having an effective and presumably scalable system in place, B&T understands that Meta has no plans to roll out the system to Australia or New Zealand.
In June of this year, a court ruled that Meta’s Special Ad Audiences tool discriminated against Facebook users based on their race, colour, religion, sex, disability, familial status, and national origin.
According to court filings, the Lookalike Audience and Special Ad Audience tools used a machine-learning algorithm to find users who shared similarities with groups of users selected by an advertiser using several options provided by Facebook. Facebook allowed its algorithm to consider FHA-protected characteristics — including race, religion and sex — in finding Facebook users who “look like” the advertiser’s source audience and make them eligible to receive housing ads.
In response, Meta retired the Special Ad Audiences tool and launched VRS.
The new tool uses machine learning to ensure that the actual audience that sees an advert more closely reflects the eligible target audience for the ad.
“After the ad has been shown to a large enough group of people, the VRS measures aggregate demographic distribution of those who have seen the ad to understand how that audience compares with the demographic distribution of the eligible target audience selected by the advertiser,” the company said in a blog post.
It also uses ‘Bayesian Improved Surnamed Geocoding’ along with US census statistics to measure estimated race and ethnicity while respecting user privacy. Over the length of an ad campaign, VRS will keep measuring the audience’s demographic distribution and work to reduce the differences between audiences.
The system is already in use for housing adverts in the US and the company plans to expand it to employment and credit adverts over the coming year.
