Product Strategy to Limit COVID-19 Health Misinformation in Facebook's Feed

You are a product manager at Meta working on the Facebook News Feed. During the COVID-19 pandemic, false and misleading claims about the virus spread quickly through people's feeds. How would you fight this misinformation? Walk through how you would define the problem, which kinds of false claims you would go after first, what you would change in the product and how you would know it is working. ```hint Not every false claim is equal COVID-19 falsehoods range from fake cures to rumors about local cases to distorted statistics. Compare them on the harm they can cause, how fast they spread and how easily they can be checked before you decide where to intervene. ``` ### Clarifying Questions - Where is the line between misinformation and legitimate debate: claims that contradict public-health authorities, unverified rumors about local cases, or also contested opinions about lockdown policy? - Does the scope stop at posts in the feed, or does it also cover shared links, comments, and group posts and reshares that surface in the feed? - Is the goal to cut how often people see false claims, how often they are posted and reshared, or both? How much reach for borderline but legitimate posts are we willing to give up? - Which countries and languages come first, given that rumors and the trusted health authorities differ by market? ### What a Strong Answer Covers - A working definition of COVID-19 misinformation and a breakdown of claim types (cures and treatments, vaccines, local outbreak rumors, statistics, official guidance) ranked by harm and prevalence - The people involved (posters, resharers, viewers, fact-checkers, health authorities) and the points in a post's life where the product can act: when it is posted, when it is ranked and distributed, and when it is viewed - A prioritized set of interventions covering detection, the choice between removing, demoting and labeling, and user-facing changes, with the speed-versus-accuracy and safety-versus-expression trade-offs stated - Success metrics centered on how much misinformation people see, with guardrails for wrongful enforcement and for the reach of legitimate health information - Risks such as shifting scientific consensus, evasion by bad actors, weak coverage in smaller languages, and privacy when rumors name real people ### Follow-up Questions - Partway through the pandemic, health guidance on a topic changes and a claim you had labeled false becomes plausible. How should the product handle that? - You cannot review every post. How would you estimate how much COVID-19 misinformation people actually see? - A rumor that residents of one apartment building have tested positive spreads through a city within hours. How would you respond without exposing anyone's health information?