After being accused of using advertising techniques to allow advertisers to exclude customers based on their gender or family relationships the social media giant Meta has just launched a new Variance Reduction System (VRS) which claims that it would aid in the fair distribution of advertisements.

The new system for housing advertisements in the United States had been established, according to a blog post by Meta. This will expand over the next year to include US credit and employment adverts. 

Meta was accused of allowing real estate adverts to specify which racial or ethnic groups to exclude from their marketing during the trial.

For instance, the company forbids advertisers from targeting their housing, employment, or credit advertising based on age, gender, or ZIP code if they are based in or aiming to reach consumers in the US, Canada, or certain European nations.

In a previous statement, US Attorney Damian Williams stated that companies that develop and implement technology, deny users of housing opportunities based in whole or in part on protected characteristics that have violated the Fair Housing Act, just as when companies engage in discriminatory advertising using more traditional advertising methods.

With the new method, once an advertisement has been seen by a sufficient number of people, a machine learning algorithm kicks in and measures the advertisement's overall demographic distribution and modifies it as necessary to make it more relevant to the advertiser's desired target audience.

By measuring an audience's estimated race and ethnicity using Bayesian Improved Surname Geocoding (BISG), Meta claims that it can measure demographics while maintaining anonymity. Re-identification of people inside aggregated databases is likewise protected.