Studies that found Meta algorithms do not drive misinformation are misleading, say researchers

A view of Facebook parent company Meta's headquarters in Dublin.

Researchers have claimed to have debunked studies funded by Facebook owner Meta that found its algorithms are not major drivers of misinformation online.

The team led by UCD assistant professor Przemek Grabowicz found while such platforms can limit untrustworthy content online, they often choose not to.

Published in the journal Science, the researchers said Meta-funded research was conducted during a short period when the company introduced a new, more rigorous news algorithm rather than its standard one, and the researchers did not account for this change in the algorithm.

They said this helped create the perception that Facebook and Instagram news feeds were “largely reliable sources of trustworthy news”.

This “rang alarm bells” for the team when they realised the previous research was conducted during the same time Facebook made a short-term change to its news algorithm.

Starting around November 2020, the time of the US presidential election, Meta introduced 63 “break glass” changes to Facebook’s news feed which were expressly designed to diminish the visibility of untrustworthy news surrounding the 2020 US presidential election.

The researchers said these changes were successful, cutting user views of misinformation by around 24%, but these changes were only temporary.

“[The Meta-funded] paper gives the impression that the standard Facebook algorithm is good at stopping misinformation, which is questionable,” Dr Grabowicz said.

They said studies such as the one funded by Meta must be “pre-registered”, meaning companies could know what researchers are looking for ahead of time.

But, at the same time, social media companies are not required to notify anyone when making significant changes to their algorithms.

The researchers wrote: “This can lead to situations where social media companies could conceivably change their algorithms to improve their public image if they know they are being studied.” 

They conclude social media companies can mitigate the spread of misinformation through modifying their algorithms but “may not have the financial incentives to do so”.

Issues around social media algorithms and the content they put in front of users have been to the fore in Ireland recently. Coimisiún na Meán chair Jeremy Godfrey told the Irish Examiner late last year one of their chief concerns centred on “toxic feeds”.

The Irish Council for Civil Liberties commissioned research earlier this year which showed a strong support for regulating social media algorithms more strictly.

However, there was dismay when toxic algorithms weren’t included in the regulator's updated Online Safety Code in May.

What's your view on this issue?

You can tell us here

The watchdog said at the time it would be best placed to tackle the potential dangers of such systems through the EU’s landmark Digital Safety Act, for which it is charged to hold Irish-based tech giants to account.

The researchers in this study highlighted the Digital Services Act as having a role to play here, adding that currently "there is no guarantee that algorithms beneficial to the public will be in place again in the future".