A Twitter account from a supposed Democratic party activist was suspended amidst accusations that the account is fake. The account belonged to Erica Marsh – Twitter handle @ericareport – who claimed to be a field organizer for the Biden campaign and a “volunteer for the Obama Foundation.” However, when The Washington Post reporter Drew Harwell raised concerns to Twitter, the account was suspended with no response. In the past few months, the account was already flagged by other researchers as possibly fake.
In the span of just 8 months, @ericareport gathered over 100,000 followers through “rage baiting,” where a user garners attention by posting content meant to stoke controversy. In his analysis, Drew Harwell found that she was “especially popular with conservatives, who promoted her as a perfect symbol of how overly theatrical and inane liberals can be.”
The use of deceptive accounts on social media meant to manipulate political discourse did not start with Erica Marsh.
- In 2020, Dean Browning, a local conservative politician in Lehigh County, Pennsylvania, surprised Twitter users when he responded to his own tweet claiming to be a “black gay guy,” apparently forgetting to log into a fake account before responding to himself (the technique Browning was failing at in this instance is known as “sock puppeting”).
- In 2017, Democratic party operatives and researchers conducted what they dubbed an “experiment” during the Senate election in Alabama, called Project Birmingham. The operation involved creating a Facebook page called "Alabama Conservative Politics" and producing content geared toward Republicans in that state. According to an internal document obtained by The Washington Post in 2018, the page’s creators "planted the idea that the [Republican Roy] Moore campaign was amplified on social media by a Russian botnet […] then tied that botnet to the Moore campaign digital director, making it appear as if he had purchased the accounts." According to the same document, the stated goal was to "enrage and energize Democrats" and "depress turnout" among Republicans.
Fakepedia, from rage bating to troll farm
Rage baiting: The act of posting content deliberately meant to provoke outrage with controversial or incendiary statements, with the goal of increasing traffic and engagement (which in many cases translates to revenue).
Troll farm: A troll farm is an organized group of internet trolls (people who post insincere, inflammatory, or accusatory remarks on social media) with the goal of manipulating public discourse. Troll farms can be ad hoc groups of like-minded individuals, but can also be groups employed by all manner of organizations and institutions.
Conservative political communications firm Turning Points USA employed trolls out of Phoenix, Arizona (including minors) to spread doubt about the integrity of the 2020 election and the COVID-19 pandemic. Former Brazilian president Jair Bolsonaro was revealed to be operating an "office of hate," which according to Bloomberg "rain[ed] hate and vitriol on all who oppose[d]" him while in office.
The Internet Research Agency (IRA) - a Russian government troll farm based in St. Petersburg - created fake accounts on social media to manipulate public discourse in the run-up to the 2016 US election.
Uncensored chatbot: Chatbots are prone to offering inaccurate, unhelpful, and offensive writing when answering queries. In 2016, for instance, a chatbot on Twitter named Tay, created by Microsoft, veered into lewd and racist comments within 24 hours of going online, forcing the tech giant to shut it down. Just this year, a lawyer in New York presented a bogus case precedent to a judge after ChatGPT offered it as an answer to his query. With these possibilities in mind, chatbot creators have instituted guardrails to mitigate these results, with the goal of eliminating them altogether.
Uncensored chatbots independently replicated versions of established technologies like ChatGPT, with little to no guardrails implemented, leaving open the possibility for misinformation and morally shocking content (e.g. instructions for committing violence to others or suicide, descriptions of child pornography).
