Should we worry about Facebook rigging elections?

Recently, I had a conversation with a young, idealistic acquaintance who thought that Facebook had the potential to rig elections. He was indirectly referring to this paper. Here are a few reasons why I’m not terribly worried about Facebook being able to rig elections:

  • The effect sizes measured in the paper were really small. For example, “For each close friend who received the social message, a user was 0.224% (null 95% CI -0.181% to 0.174%) more likely to vote than they would have been had their close friend received no message.” Yes, I know that sometimes a small difference goes a long way, but not for landslide elections, and not without feedback cycles.
  • Human societies are complex, unpredictable systems. Economists may lower the interest rates to get some predicted outcome, but the total opposite may happen. The best laid plans of men can backfire. Why are elections any different? The conceit of many “data scientists” is that they think they can do psychohistory just because they have Big Data. So what if you have “predictably irrational” behaviour in laboratory settings? That doesn’t mean you know how to reliably manipulate individuals — let alone society — at large.
  • There is a much bigger problem: many presidential candidates already seem to be in the pocket of lobbyists. Why are we worried about a social network being able to tip elections when either candidate has already been bought over? We should be much more concerned about campaign finance reform.

Instead of worrying about exactly who wins, we should try to safeguard our laws and institutions such that the harm that can be done by any president is limited, inasmuch as that is possible.