Media Crowds and Madness

Not what the teacher said to do
Making dreams come true
--Oingo Bingo

Prior to the 2016 presidential election Nate Silver was a media darling of the left. A political analyst with a background in statistics, Silver's stock increased with progressives when, based on his analysis of polling data, he assigned a higher degree confidence to a Barack Obama victory over Mitt Romney in the 2012 than many other 'experts.'

The allure for many Democrats seemed to be that not only did Silver predict that their side would win, but that he did so 'scientifically.' Like most people, Democrats tend to like the idea of scientific process when it supports their view.

The value of Silver's equity recently declined after his blog's analysis of 2016 presidential polling data suggested a high degree of confidence in a Hillary Clinton victory. Progressives who leaned on that prediction like they did in 2012 were sorely disappointed. Silver's 'science' let them down.

Silver, of course, would say (and I'm sure he did) that he was forecasting a probability, not a certainty. Although his forecasts suggested, say, a 75% percent chance of a Clinton win, there was still a one in four chance that she wouldn't.

Post election, Silver has written a series of pieces, including this one, that autopsy the phenomenon of the mainstream media's gargantuan miss of the 2016 presidential election. He concludes, per the title, that "there really was a media bubble." The use of past tense 'was' is interesting, as if the 'bubble,' better known as bias or slant, was a temporary condition not in place before or since the 2016 election cycle.

Drawing from Surowiecki's Wisdom of Crowds which. incidentally, follows a volume written 150 years earlier by Charles Mackay on Extraordinary Popular Delusions and the Madness of Crowds, Silver cites Surowiecki's argument that crowds tend to make good predictions when four conditions are in place. Crowds are 'smart' when there is diversity of opinion among participants, when people's opinions are independent of those around them, when knowledge is decentralized and local, and when private judgments can be aggregated into collective decision.

Silver suggests that political journalism scores poorly on the first three conditions. American newsrooms are not diverse. Over 92% of journalists have college degrees, up from 58% in 1971. More importantly, there seems little in the way of political diversity. Voting research suggests, for example, that journalists who vote Democrat outnumber those who vote Republican by at least four to one margins. Of the 59 major newspapers that endorsed presidential candidates in 2016, only two endorsed Trump.

Silver argues that independence among political journalists is also low. They attend the same conventions and debates, and gather in the same room for discussions among themselves. This facilitates "conventional wisdom being manufactured in real time." Reporting of conventional wisdom via social media leads to information cascades that reinforce how issues are framed through millions of impressions.

Political journalism is highly centralized, with most political news manufactured in NYC and DC. National reporters fly into local areas with pre-baked narratives and seek facts that support their point of view. That local newspapers are failing and media outlets with broad reach are gaining market share reinforce centralization of opinion-making.

In terms of improvement, Silver suggests that increased decentralization and improving diversity would be difficult. I am not so sure, particularly with respect to diversity. For example, why couldn't newsrooms intentionally hire reporters in a manner to achieve greater diversity from a political ideology standpoint? Outlets could market this diversity and reveal their ideological mix--perhaps as audited by an outside source. Outlets that want to be forthright about their bias would require journalists to disclose their voting and political contributions as part of their reporting.

Silver thinks there is more short term potential in changing the independence condition. He suggests that journalists could 'recalibrate themselves' to become more skeptical of consensus opinions and open to other viewpoints. Good luck with that. However, his recommendation does bear some similarity to Groseclose's idea that journalists could hang with others who are not like themselves in order to better understand other points of view.

There is also the view, of course, that the current state of media bias is just supply following demand. Consumers of information do not want diverse, independent, decentralized media coverage. Instead they seek out viewpoints that confirm their own viewpoints and minimize negative psychic income associated with realizing that their viewpoints may be wrong.

Cognitive dissonance w.r.t. things political facilitates and reinforces media bias among both producers and consumers.