Meta rolls out new ad variance system to create "equitable distribution" of ads

Representative Image

After being accused of ad practices that enabled advertisers to exclude users by gender or family ties, Meta has now rolled out a new Variance Reduction System (VRS) that the social media giant says will help in equitable distribution of ads.

In a blog post announcing the system, Meta said that they had launched the new system for housing ads in the United States. Over the coming year, this will grow to include US employment and credit ads.

The new system is part of the settlement that Meta reached with the US Department of Justice (DOJ). The trial accused Meta of allowing housing advertisers to specify which ethnic or cultural groups to exclude from its campaigns.

“When a company develops and deploys technology that deprives users of housing opportunities based in whole or in part on protected characteristics, it has violated the Fair Housing Act, just as when companies engage in discriminatory advertising using more traditional advertising methods," US Attorney Damian Williams had said in an earlier statement.

With the new system, once an ad has been shown to a large enough group of people, a machine learning algorithm kicks in and measures the aggregate demographic distribution of the ad and tweaks it accordingly, so that the ad is in line with an advertiser's intended target audience.

To protect privacy, Meta says it uses a method of measurement called Bayesian Improved Surname Geocoding (BISG) that measures an audience's estimated race and ethnicity. It also protects against re-identification of individuals within aggregated datasets.

"The VRS builds on our longstanding efforts to help protect against discrimination. This includes restricting certain targeting options for campaigns that advertise housing, employment or credit ads," wrote Meta.

"For example, we don’t allow advertisers that are either based in or trying to reach people in the US, Canada and certain European countries from targeting their housing, employment or credit ads based on age, gender or ZIP code."