Charlie Warzel, Buzzfeed:
Using publicly available information pulled from the APIs of USA Today, the New York Times, the Guardian, and BuzzFeed, researcher Joe Hovde compiled over 87,000 articles about Facebook published by the four outlets between 2006 and 2018. Then he ran a sentiment analysis on them, scoring words on a positive-to-negative scale of -5 to +5 — for example, a negative word like “fake” was scored -3, while a more positive word like “growth” was scored +2. The results were grim.
Hovde’s chart shows a steep increase in almost exclusively negative sentiment about Facebook beginning in late 2016, around the time of the presidential election. It also reveals a steady decline in positive sentiment between 2006 and 2016.
What this study seems to show is that the media is reacting solely to the remarkably shitty outcome of the 2016 American presidential election, arguably partially enabled by Facebook’s micro targeted ads. What it actually reveals is that Facebook — and Silicon Valley firms more generally — should have been covered with much more scrutiny and skepticism for years. The growing influence of algorithmically-tailored information based on mass data collection has always been worrying for now-obvious reasons, and more mainstream outlets should have explored that angle sooner and more frequently.