Democrats, Cities, and the Media

This essay isn’t all that good or important, but it seems true

I believe that Democrats are structurally disadvantaged by where they are in power. It’s not just that they are electorally inefficient, and are easily gerrymandered. That could be fixed by electoral reforms. What is perhaps a larger disadvantage is that, since they rule over the big cities, their errors and inefficiencies are much more reported on than things out in the boonies.

This is made worse by how people reason. We have not yet adapted to the age of social media. We reason via anecdotes, not through data. When our groups were small, we could accurately infer what was going on in the world from it. Hilde’s cows fall sick, which leads us to infer that there are witches placing hexes on the cows. Here it is completely accurate, and allows us to find the witch, burn them, and save the cows. In today’s world, we can hear about cows falling sick every day, which leads to us burning a number of witches exceeding the social optimum.

In a more modern context, New York City is one of the safest large cities in America, yet everyone thinks it far more dangerous than it actually is. The explanation is simple – NYC is the media capital of the world, and with ten million people, there’s always going to be someone doing something crazy. If we believe that life is more dangerous than it actually is, we might imprison too many people — and we might be too harsh on Democrats. Media attention, and the accuracy of our world model, follows an inverted U-shaped curve.

There’s some related evidence. Campante and Do (2014) found that states which have their capitals far away from the center of the population of a state have much higher levels of corruption than those whose capital is near the population center. Their favored explanation is that there is less media scrutiny of state government, and in turn that citizens are less informed and less enthusiastic about voting. They are able to show (page 17) that states with isolated capitals have fewer stories about state-level politics, and that this is not simply due to size, but due to geographic isolation. The media reports on what happens in its vicinity, and does not investigate the hinterlands in the same way. I have no doubt that if you placed a reported on the beat in every podunk town in America, you could generate a stream of clickbait as wide as the year is long.

News is optimized to spread, not to reveal information. Things which are shocking and unusual are much more shared than things which are mundane, even when the mundane is more informative. This is actually somewhat counterintuitive, because we might think that unusual events are precisely the thing which the news would report on. (We can define unbiasedness in news following Armona, Gentzkow, Kamenica, and Shapiro, where maximally informative news is the thing which causes the greatest change in your actions given your priors). A naive view might say that of course the news should report on murders, for the same reason that the news should report more on storms than on sunny days, because it is the sharpest deviation away from our prior of “murders are unlikely to happen”. However, this is wrong. We have a prior that murders will be realized at a certain rate, and murders being realized doesn’t actually shift our prior at all. We should take no change in action, unlike when a storm comes and we choose to shelter inside. Ideally, the news would only report on the derivative of the murder rate, and not report on the level at all!

Since we are not able to comprehend the vastness of the times in which someone wasn’t murdered, we end up with a distorted view of the murder rate. I think this is a similar behavioral bias to what drives prospect theory. As recent experiments by Ryan Oprea suggest, humans make inaccurate choices when faced with complexity. We systematically overweight small possibilities. An improvement in our ability to gather information need not improve our understanding of the world, if our algorithms are biased.

This suggests that increased market power could get us closer to the social optimum. It’s akin to how market power in the market for oil has helped the world through fighting climate. (See this recent working paper for more!) Maybe the end of soft censorship was bad for us. Too late to put the genie back in the bottle.

Democrats are, to some degree, suffering from success. They are winning among the wealthy and successful. With that comes scrutiny, and no political party can stand scrutiny.

P. s. It is striking how my evidence for this is only a step above everyone else’s, though. Here I am decrying people’s reliance on anecdotes, and my evidence is one paper. I am not able to think in data, but need ideas to be condensed into discrete nibbles. I have no sense of whether the studies I cite are representative of a wider literature, or whether I am picking the studies which are optimized to spread. You can see this in action on twitter – I post a mixture of very useful studies, and very “punchy” studies, and it is the punchy studies which go viral (okay fine, “viral”) more often. Studies with funnier titles (after controlling for the importance of the paper, which they do by controlling for self-citation) get more citations. I find that the success of my blog posts is often driven simply by the titles, where my most informative blog posts get a tepid response, in spite of their greater importance.

I’m not sure what I can do about this, either for myself or for society. I do think we have something of a duty to be circumspect in the information we put out there. Yes, the demagogues will go viral. You have a duty to be truthful, and perhaps enough people will hear it, and be convinced.