Highly informed voters are also highly biased. That’s a serious problem for democracy, but also for any other system of political decision-making in big groups.
Two new books, Against Democracy by Georgetown moral philosopher Jason Brennan and Democracy for Realists: Why Elections Do Not Produce Responsive Government by Princeton University political scientist Christopher Achen and Vanderbilt University political scientist Larry Bartels, deal intellectual hammer blows to the political system so many of us take for granted: democracy.
Achen and Bartels show that most voters are ignorant of ideology and other basic political and economic facts (Brennan dubs these voters “hobbits”), while the remainder vote on the basis of their unshakable partisan identities (Brennan dubs these voters “hooligans”). Hardly anyone approximates the ideal that Brennan dubs “Vulcans”: unbiased and knowledgeable about politics and economics, and also conscientious in voting.
Because voters are so flawed, democracy does not actually bring about the “will of the people,” and as Brennan argues, most people’s political wills are corrupt anyway, and so we would not want a political system that gave them power.
Brennan infers from this evidence and a moral argument about the overriding importance of competent government that “epistocracy” (rule of the knowledgeable) may well be superior to democracy.
However, I argue that the problem of motivated reasoning implies that empowering the politically knowledgeable is likely to lead to just as severe problems as democracy with universal, equal suffrage.
To show this, we need to look more closely at the problems of confirmation bias and motivated reasoning in politics.
Confirmation bias refers to an unconscious process whereby people tend to discount evidence that goes against their existing views and overrate evidence that apparently confirms them. Because partisan identities are so strong in the United States and most other democracies, partisans suffer from severe confirmation bias. They are extremely unlikely to change their minds about candidates or policies in the face of new evidence, and in fact their issue positions flow almost wholly from their partisan identities, rather than the other way around.
Two good, recent examples of this phenomenon are partisans’ changing views about trade and about the importance of personal ethics in a candidate. Figure 1 shows how Republicans used to view foreign trade more positively than Democrats, but Democrats became much more pro-trade when Obama took office and began negotiating trade agreements.
Figures 2 and 3 show how Republicans were, by one measure, as free-trading as Democrats as recently as May 2015, just before Donald Trump began his presidential campaign, but once Trump began pounding the lectern with anti-trade rhetoric, Republicans turned sharply against trade.
Did anything in the world change to make free trade less desirable for Republicans or conservatives? No. What changed was that the candidate to whom Republicans were committed opposed trade, and Republicans dutifully changed their minds on the issue.
The other example of partisan rationalization comes from a survey of views on personal ethics in politics (figure 4). In 2011, white evangelical Protestants were the least likely religious group to say that an elected official who acts immorally in his personal life can still behave ethically in his professional life. By 2016, white evangelical Protestants were the most likely group to believe that personal ethics don’t matter for professional life.
Why the huge change? The vast majority of white evangelical Protestants are Republicans, and now that the Republican presidential candidate is an immoral man, Republicans have changed their views on whether that trait matters.
Motivated reasoning is a related problem, referring to how voters with strong partisan or ideological views go to great lengths to reinterpret new evidence to fit their worldview, rather than adjusting their worldview. A recent but much-cited experiment on motivated reasoning from Yale social psychologist Dan Kahan and collaborators showed that “politics makes you bad at math.”
When confronted with counter-ideological evidence from a fake study on gun control, both progressives and conservatives of high mathematical ability were unable to interpret the data correctly.
Figure 5 shows one of the key charts from the paper. The blue lines are for progressives’ responses under the two experimental conditions (gun control increases or reduces crime, indicated with solid and dashed lines, respectively), and the red lines are for conservatives’ responses under the two experimental conditions.
Note that high-numeracy progressives are no more likely than low-numeracy progressives to understand that the data show gun control increases crime when that is the correct interpretation, and high-numeracy conservatives are barely more likely than low-numeracy conservatives to understand that the data show gun control decreases crime when that is the correct interpretation. However, numeracy does aid both conservatives and progressives in interpreting the data correctly when those interpretations are congenial to their existing worldviews.
And here is where epistocracy runs aground. Epistocracy works by somehow up-weighting the votes of more politically knowledgeable voters. Brennan argues that this system should be superior to democracy because it at least rids us of the policy problems caused by political ignorance and those types of bias more common among the uneducated and unintelligent (such as nationalistic bias).
However, another likely consequence of up-weighting the votes of the more politically knowledgeable or the more intelligent is to give partisan and ideological confirmation bias and motivated reasoning an even bigger role in our politics.
After all, it is the intelligent and informed who are really good at self-deception. Brennan concedes that strength of partisan identity correlates positively with political knowledge. If we empower the knowledgeable, we empower the partisans.
If we let pure independents with lots of intelligence and political knowledge make our decisions, it is reasonable to suppose those decisions would be more competent. But those voters are rare; and, practically speaking, there is no way to filter voters for strength of myside bias or motivated reasoning. (Partisans will pretend to be independents if incentivized to do so.)
Perhaps epistocracy would incentivize independents with low knowledge to acquire more knowledge in order to have more voting power, but who is to say that this process will not turn these hobbits into hooligans rather than Vulcans?
Political decision-making in large groups is likely to be terrible whether we use democracy or epistocracy.