Why Facebook should hire astronomers
DOI: 10.1063/PT.5.010287
On 3 June, the Proceedings of the National Academy of Sciences published the paper “Experimental evidence of massive-scale emotional contagion through social networks .” The scary-sounding paper described a study in which the news feeds of 689 003 Facebook users were stealthily manipulated to suppress either positive or negative posts.
Some Facebook users were angered by the study. They resented being used as unwitting lab rats in an experiment whose aim was to see whether the emotional tone of the posts in their news feed influences the emotional tone of the posts they write. In response to the uproar, Facebook’s chief operating officer, Sheryl Sandberg, half-apologized; PNAS editor-in-chief Inder Verma published an “Editorial expression of concern .”
I expect the negative reaction surprised the paper’s lead author, Facebook’s Adam Kramer. PNAS is a prestigious journal that issues weekly press releases to tout its contents. Kramer and his coauthors evidently wanted to share the results of their study with their peers and perhaps with the public, too.
What’s more, the authors’ motives in carrying out the study were far from sinister and perhaps blinded them to the paper’s potential for causing trouble. When news of the study broke, Kramer wrote (on his Facebook page, naturally):
The reason we did this research is because we care about the emotional impact of Facebook and the people that use our product. We felt that it was important to investigate the common worry that seeing friends post positive content leads to people feeling negative or left out.
When we see Frozen or other tear-jerking movies, we implicitly consent to have our emotions manipulated by filmmakers. Signing up to use Facebook entailed no such agreement. Deliberately making people sad without their consent will always seem creepy at best, immoral at worst.
But did Kramer and his coauthors need to resort to manipulation? In their experiment, some users saw 10% fewer negative posts than average, while others saw 10% fewer positive posts. Given the perturbation’s modest size, the experimenters could have instead made use of variations in behavior that arise naturally, just as astronomers do when they observe celestial phenomena.
My favorite example of exploiting natural variation dates from the 1970s. Radio astronomers suspected that the seemingly uniform background emission in their images was made up of myriad distant galaxies too faint to resolve. If so, the number of galaxies in the field of view would fluctuate randomly as the telescope scanned the sky. Those fluctuations could be measured and the mean density of galaxies deduced.
The Facebook universe surely abounds in natural variations that could be correlated with information within Facebook or with outside events (such as heat waves or World Cup soccer matches). But to exploit and analyze those variations, the company might need to hire astronomers.