Thread by @darchivist

Thanks @russ1mitchell for covering our recent paper describing the existence of corporate computational social media accounts (#fanbots) and their role in support of the Tesla narrative. Here’s a link to the article in @latimes latimes.com/business/story… and a short summary. 🧵 1/n

Looking at tweets from Tesla IPO to 2020, we found a set of accounts that did not exhibit human-like behavior. Using Botometer botometer.osome.iu.edu, an IS tool from Fil Menczer at UoIllinois, we identified these accounts as programmed users generating pro-Tesla content. 2/n

The #fanbots were active in the pro-Tesla movement using #TSLA and $TSLA, whereas accounts active in the counter-movement around #TSLAQ and $TSLAQ were human users. This imbalance suggests that the fanbots were a strategic resource supporting the Tesla narrative. 3/n

Several observations: first, this activity began early and its scale was significant. Initial tweets from #fanbots preceded public debate about use of propaganda on social media to shape political outcomes. #fanbots generated 23% of #TSLA tweets over our study period. 4/n

Second, the fanbots seem to matter: Adjusted stock returns were positive (and substantial) in the week following fanbot creation, though we are still exploring the mechanisms underlying this observation. 5/n

Third, we cannot prove that the fanbots changed the course of history. For some, the best explanation may always be – “Tesla was going to be successful anyway.” However, when $1T is at stake, it’s probably worth running down all the possible alternative explanations. 6/n

Fourth, we do not know where (or who) the fanbots came from. @elonmusk has never mentioned the use of bots to support Tesla and famously touts Tesla’s $0 budget for advertising. The use of #fanbots is a cheap intervention, with seemingly high returns on investment. 7/n

Fifth, though the use of #fanbots is not illegal, their potential influence on other social media users and on markets poses questions for regulators and policymakers. For example, should social media content generated by programmed users be labeled as such? 8/n

Sixth, we are not out to score points, merely trying to understand how entrepreneurial resource acquisition operates in the age of #Muskism. The paper will be presented at some conferences in the next couple of months. Happy to share it with those who are interested. 9/n

Finally, HT and thanks to many colleagues whose work inspired us to explore these issues, including @ginasue @marcventresca @STSNews @RhondaReger1 @pnhoward @vanishingcorp @RobertNEberhart @braydenk All usual caveats apply. 10/end