The Conventional Economic Model:

1) People are rational maximizers,
2) Correct information helps them maximize (false information has no value to them because they get no utility out of believing false things),
3) Competitive Markets will generate the optimal amount of correct information. Information will not be perfect because:
a) this is probably not possible, but
b) even if it were, it would be prohibitively expensive the squeeze out the last bit of truth, and
c) people who develop information must earn a return on their efforts so they must know a little something that not everyone else knows. This last point implies that while one cannot make excessive profits by becoming an expert, experts do earn a higher return on the basis of their specific knowledge (just enough to compensate them for the effort of becoming experts).

This view that there are forces at work ensuring accurate information is probably shared by computer scientists. So for example, there is an underlying presumption embedded in Google and Wikipedia that large masses of people will help ensure the triumph of the truth.

An Alternative View:

1) While people may maximize utility, they get disutility from hearing information that is at variance with their previously held beliefs (as opposed to welcoming the information as rescuing them from their ignorance, which might have caused them to make bad decisions), so they do not seek it out.

2) Separate from the above emotional reason, humans seem to have a cognitive bias that consistently underestimates the value of seeking disconfirming information. The four card experiment is a good illustration of this.

Together, 1) and 2) can be called Confirmation Bias.

Consequences of 1) and 2):

Consequence A: As described in "The Market for News", it may be most profitable for news media to segment the market - cater to people's preconceptions than to their desire for the truth. More media competition will not solve this problem - it may make it worse. We were probably better off when ABC, CBS and NBC had a quasi monopoly and determined what news to present based on their sense of professionalism than we are in the current environment where they are desperately competing for audience.

Consequence B: A successful campaign (political or advertising) appeals to people's previously held beliefs; it does not try to alter their beliefs with new information. The goal is to present your candidate (or product) as someone the voter (customer) would already be supporting (buying) if only they had known about the candidate (or product) previously.

3) "Neighbors are the best campaigners" I don't know if the below is true (so the fact that I included it is an example of confirmatory bias), but here is a quote from a website on selling your home, "You might get lucky and find a buyer, but the main objective of your open house is to inform your neighbors and enlist them in your sales campaign. Remember: "Over 50% of homes are bought by friends and relatives of people who already live or work in the neighborhood."

People believe information more readily if it comes from people they already trust, even if the information is in an arena completely distinct from their usual interaction with these people. This means that a political campaign can be modelled as a series of discrete local interactions. For a national campaign one has to layer over this the "news" coming from the media, but for a local campaign, this local interaction may be most of the campaign.

4) For reasons that are unclear, people like to believe the worst of politicians. A salacious story about a politician has more "legs" than a virtuous story.

Consequence of 3) and 4):

Consequence C: Word of mouth matters a lot - especially in a local campaign. One can, therefore, run a campaign on two levels: the undercampaign and the public campaign. The undercampaign can be particularly dirty because a) more salacious "news" will travel faster and farther, and b) the allegations are never put in print so they cannot be refuted.

Why Political Outcomes are So Bad:

1) Elections are always one time, unique, events. Even rational decision makers with a lot of time on their hands would have problems making good forecasts in this environment. It would be very difficult in this environment for someone to learn from their mistakes.

2) Who gets elected has minimal effect on people's lives, especially as the government becomes smaller and smaller.

i) Strictly speaking, therefore, it does not pay for people to spend a lot of time seeking out information and trying hard to make the right decision. In fact, I would argue that we are lucky that people spend as much time on politics as they do. This is in part due to civic responsibility and in part due to the competitive nature of politics (it is like watching a sports event).

ii) They do not lose a lot by making a non-optimal decision that conforms to their own preconceptions. In other words, people can vote their prejudices and not suffer any consequences. There is no brake, therefore, on confirmation bias.

3) The free-rider problem: Economists and political scientists have spent a lot of time worrying about why people vote at all (given that the probability of their influencing the outcome is near zero). In line with other research, I am proposing an expressive motive: people vote to register their beliefs. This motive, however, is usually presented as a good thing because it counters the free-rider problem, but there is a downside that I emphasize here. Since people know that they really can't influence the election, they are free to "register" whatever preconceived beliefs that they have. They get utility from voting, but they do not get utility from questioning their previously held beliefs, and there is nothing in the election process that forces them to do this.