Surely I can’t be the only one skeptical of Eliezer’s Gaff Theory? Indeed, journalists would often applaud policies with very little support if it fitted in with their political beliefs. Instead, I suspect that politicans don’t want to support a policy until it moves over that 50% threshold, at which point they switch en-mass to supporting it, not because of the media, but because they want votes.
Any policy-centered theory like this runs into the problem that voters don’t vote based on policy. From a NY Times discussion of why Sanders voters in the 2016 primary generally supported Clinton’s policies more than Clinton voters did, and vice versa:
Decades of social-scientific evidence show that voting behavior is primarily a product of inherited partisan loyalties, social identities and symbolic attachments. Over time, engaged citizens may construct policy preferences and ideologies that rationalize their choices, but those issues are seldom fundamental.
I think Cecie is right that press coverage of things like “gaffes” and “scandals” often focus intensely on the recursive issue of whether a gaffe will make someone “controversial” (unelectable, offensive, unpopular, etc.), neglecting or entirely skipping over the question of whether the gaffe was factually accurate. The last few times I’ve actually looked at news stories about controversial things, whether during elections or not, the coverage was very strongly and explicitly focused on perception over substance.
One way of putting it is that journalists care a lot about what the voters(/consumers/public) think, and this emphasis on electability over substance can then run away with itself. Another way of putting it is that voters care a lot about what other voters think. We’re all perpetually nervously glancing around to try to spot signs of what everyone else’s plans and preferences are, and what the new social norms and expectations are; journalists then pick up on this nervous curiosity and give the voting public what it wants, and also have an outsized impact on those norms and expectations because they’re an unusually centralized and concentrated source of evidence about norms/expectations.
Regardless of how this pattern develops, once it’s in place it tends to be self-reinforcing, because any additional coverage of “controversies” both establishes more firmly what things are versus aren’t controversial (including what’s inside the Overton window policy-wise), and establishes more firmly that it’s normal to focus on controversialness and what-other-people-are-thinking-about-this over substance.
Two places where it looks like I disagree with Cecie (and agree with Eliezer and possibly Simplicio) are that voters aren’t actually voting rationally, and in particular, voters aren’t just trying to model other voters for strategic reasons; they’re also emotionally invested in the idea of supporting strong, impressive, successful, high-status people. Gaffes don’t just provide evidence that candidates or policies will fail; they provide evidence that it will be embarrassing, low-status, or otherwise unpleasant to be perceived as supporting those candidates or policies. Even the minority who support underdogs want to feel like they’ve found a strong, impressive, successful, high-status underdog who’s got a real shot at victory against all odds, and loses only due to bias/unfairness and not due to social-status-lowering personal failings.
I think the idea that gaffes are serving as a proxy not only for acceptability-to-others but also for weakness (which in turn provides some evidence about emotional appeal-to-others) helps make a lot more sense of Donald Trump’s appeal. See also Eliezer on Gruber and on Trump.
To test whether this is an important factor, we might look at how well “losers” perform in non-FPTP systems. In FPTP, it makes sense to shun candidates based purely on the fact that you’re worried other people might shun them. In other systems, this makes less sense, so we can better isolate how much of the effect is about overactive primate coalition-building instincts versus sensible voting strategy.
Any policy-centered theory like this runs into the problem that voters don’t vote based on policy. From a NY Times discussion of why Sanders voters in the 2016 primary generally supported Clinton’s policies more than Clinton voters did, and vice versa:
I think Cecie is right that press coverage of things like “gaffes” and “scandals” often focus intensely on the recursive issue of whether a gaffe will make someone “controversial” (unelectable, offensive, unpopular, etc.), neglecting or entirely skipping over the question of whether the gaffe was factually accurate. The last few times I’ve actually looked at news stories about controversial things, whether during elections or not, the coverage was very strongly and explicitly focused on perception over substance.
One way of putting it is that journalists care a lot about what the voters(/consumers/public) think, and this emphasis on electability over substance can then run away with itself. Another way of putting it is that voters care a lot about what other voters think. We’re all perpetually nervously glancing around to try to spot signs of what everyone else’s plans and preferences are, and what the new social norms and expectations are; journalists then pick up on this nervous curiosity and give the voting public what it wants, and also have an outsized impact on those norms and expectations because they’re an unusually centralized and concentrated source of evidence about norms/expectations.
Regardless of how this pattern develops, once it’s in place it tends to be self-reinforcing, because any additional coverage of “controversies” both establishes more firmly what things are versus aren’t controversial (including what’s inside the Overton window policy-wise), and establishes more firmly that it’s normal to focus on controversialness and what-other-people-are-thinking-about-this over substance.
Two places where it looks like I disagree with Cecie (and agree with Eliezer and possibly Simplicio) are that voters aren’t actually voting rationally, and in particular, voters aren’t just trying to model other voters for strategic reasons; they’re also emotionally invested in the idea of supporting strong, impressive, successful, high-status people. Gaffes don’t just provide evidence that candidates or policies will fail; they provide evidence that it will be embarrassing, low-status, or otherwise unpleasant to be perceived as supporting those candidates or policies. Even the minority who support underdogs want to feel like they’ve found a strong, impressive, successful, high-status underdog who’s got a real shot at victory against all odds, and loses only due to bias/unfairness and not due to social-status-lowering personal failings.
I think the idea that gaffes are serving as a proxy not only for acceptability-to-others but also for weakness (which in turn provides some evidence about emotional appeal-to-others) helps make a lot more sense of Donald Trump’s appeal. See also Eliezer on Gruber and on Trump.
To test whether this is an important factor, we might look at how well “losers” perform in non-FPTP systems. In FPTP, it makes sense to shun candidates based purely on the fact that you’re worried other people might shun them. In other systems, this makes less sense, so we can better isolate how much of the effect is about overactive primate coalition-building instincts versus sensible voting strategy.