Authored by Kalev Leetaru via RealClear Politics (emphasis ours),
Last week, Facebook and its fact-checking partners began applying a warning label to some posts claiming that the American economy had entered a recession, deeming them false and warning that further such posts would result in sanctions. The ensuing backlash received widespread media attention, but the reality is more complex, reflecting the arcane nuances and inconsistency of the fact-checking enterprise.
As with most fact-checking controversies, the story that first emerged was both simple and Orwellian: Anyone who dared admit that the United States had entered a recession had their posts immediately flagged as false by shadowy “fact-checkers” working in concert with the White House to silence dissent. In reality, posts that received the label revolved not around whether or not the U.S. was in a recession, but on the specific claim that the White House “will redefine what a recession is so they don’t have to admit they caused a recession.”
In other words, rather than silencing dissenting views on whether the U.S. has entered a recession, Facebook’s label applied specifically to the narrow claim that the White House was attempting to redefine what a recession is. A post stating that the U.S. met the common definition of a recession would not be flagged – only those posts accusing the administration of changing the definition.
That said, given that social platforms are not required to release datasets of the posts they flag, it is not possible to determine how well Facebook’s automated content-matching systems worked in this case. Given the creativity of human language, moderation systems that rely on matching the text of posts against known fact checks tend to simply measure how similar their word usage is, rather than parse the underlying arguments being made. Thus, it is conceivable that other posts making recession-related claims may have been incorrectly flagged by Facebook. Similarly, if Facebook assumed that all screen captures of the White House’s recession page were related to the false claim, it could also flag a lot of unrelated posts.
At the same time, the fact-checker cited by Facebook, PolitiFact, emphasizes that the National Bureau of Economic Research “is the official arbiter of when U.S. recessions begin and end.” Yet in 2019, in rating a claim by Elizabeth Warren as “Half True,” PolitiFact argued that the “technical definition of an overall economic recession was met when the industrial production index from the Federal Reserve fell for two quarters in a row.”
In fact, as Phil Magness notes, the author of last week’s PolitiFact fact check citing NBER as the official arbiter of recessions wrote in a 2015 fact check of Donald Trump that “the general rule of thumb is that it takes two quarters of negative growth to signal a recession” in rejecting Trump’s recessionary claims.
Magness also cites several previous statements by Democratic politicians that asserted the U.S. had entered a recession under then-President Trump, arguing that Facebook’s failure to label those as false demonstrates political bias. Yet those claims related to whether the U.S. was in a recession, whereas the current fact check relates to whether the White House is changing the definition of a recession, rather than fact-checking whether the U.S. is in fact in a recession. This is a critical yet seemingly lost distinction.
Fact-checkers have, however, delved into more existential questions on the economy many times in the past year, including controversial questions such as whether increased consumer prices were due in part to companies charging more than needed to offset inflationary pressures.
In the end, this latest fact-checking controversy offers a reminder of how nuanced and controversial the fact-checking landscape is becoming as it moves beyond easily refuted urban legends to complex debates like the causes of inflation and whether the U.S. has entered a recession. Even seemingly uncontroversial questions like whether the White House is “redefining” recession are transformed into existential censorship battles that have little to do with the actual claim being reviewed. Perhaps fact-checkers would be better served returning to the basics and avoiding undecided questions for which there is no way of rendering a decisive verdict.