Social media is being used to target the disenfranchised in swing states. During the last presidential election polarising content was concentrated in swing states to influence the outcome.
The argument
In the run up to the 2016 US presidential election, operatives of the Internet Research Agency, a Russian company associated with Putin, posed as Americans by posting messages and ads on social media. The impersonation is the first thing to note about the operation: material that would carry little weight as foreign messaging was designed to reach people as the opinions of fellow citizens, which is what allowed it to circulate at all. The purpose of some of that content can be read directly off its substance. They used practices like directing people to vote via text, which is not even a valid option anywhere in the US. A message of that kind cannot persuade anyone of a political position; its only possible effect is that someone who follows it does not cast a vote, which makes the intention to suppress turnout explicit rather than inferred. The distribution of the material points the same way. Average levels of misinformation via social media including Twitter and news platforms were higher in swing states than in uncontested states during the 2016 presidential campaign. This holds even when weighted for the relative size of the user population in each state, so the concentration is not an artefact of where more users happen to live. Effort was being directed to the places where a small change in turnout would change the result. Nor was such targeting confined to foreign actors. The Trump campaign team also incorporated a strategy to deter millions of African-Americans from voting in 2016. This included a category named ‘Deterrence’, which was later described by Trump’s chief data scientist as containing people that the administration hoped would not show up to vote.
Premises
Counter-arguments
A study of more than 16,000 Twitter users found that only a small fraction of fake news spread during the 2016 presidential election. Only a small number saw the majority of misinformation, and those that saw it were typically older and politically conservative. Therefore the effect of fake news, and in particular its effect on swing states may have been less pervasive on social media during the election than originally thought.
Rejecting the premises
[Rejecting P1] The operation is documented, but the premise establishes that false content was posted rather than that anyone acted on it — the volume was small against the ordinary flow of political material, and the study evidence indicates that exposure was concentrated among a narrow, already-committed group. [Rejecting P2] Higher misinformation in contested states is what any political messaging looks like, since attention follows electoral value — so the finding is consistent with targeting voters generally rather than with targeting the disenfranchised, which is the specific claim the position makes. [Rejecting P3] The campaign category cited is likewise about persuasion and turnout modelling, which every modern campaign performs, and the argument never establishes the identification the position needs — that those reached were disenfranchised people rather than simply voters in the states that mattered.