How to Overcome Bias in Crowdsoucing Participants

Here are some problems that Sebastian Nickel brought up with regard to using collective intelligence for decision making:


1. This stuff pertains to public choice theory, which is defined by Wikipedia as: ” the use of modern economic tools to study problems that traditionally are in the province of political science. From the perspective of political science, it is the subset of positive political theory that models voters, politicians, and bureaucrats as mainly self-interested.”

2. An important point for much of the theory of public choice in democracies is the fact that the probability that any individual’s vote will affect the outcome of the election is generally negligible. For example, in U.S. presidential elections, the probability that your vote will affect the election’s outcome is comparable to that of winning the lottery seven times in a row. This is because your vote only makes a difference if all the other votes happen to be exactly tied, so that your vote turns out to be the deciding vote – or perhaps if your vote happens to be the one leading to an exact tie!

3. RATIONAL IGNORANCE: According to Wikipedia, “The term was coined by Anthony Downs, An Economic Theory of Democracy”. Wiki sez: “Rational ignorance occurs when the cost of educating oneself on an issue exceeds the potential benefit that the knowledge would provide.” Applied to democracy, voters are to be expected to be rationally ignorant about the issues they vote on. This is because, since their vote almost certainly won’t change policy, they have nothing to gain from making an informed vote rather than an uninformed one. On the other hand, the cost of informing themselves is considerable. Some people enjoy informing themselves about such issues, but they are a small minority.

4. Empirically, voters really are very ignorant about politics and economics. They are least ignorant about the aspects of politics that are the most entertaining.

5. Donald Wittman, in The Myth of Democratic Failure, has defended democracy against many of the usual criticisms from public choice theorists. One argument he makes in the book is that, because of the miracle of aggregation, rational ignorance is not actually a problem: provided that people are merely ignorant and not systematically biased, their mistakes should cancel out. Also, if you assume that 5% of the electorate is well-informed on the relevant issues, you could argue that those 5% effectively decide the outcome of the election, whereas the votes of the other 95%, which are randomly distributed, cancel each other out.

6. Bryan Caplan, who credits Donald Wittman with “waking [him] from [his] dogmatic slumbers in political economy”, has compiled empirical evidence that voters are, in fact, systematically biased, thus concluding that Wittman’s argument is valid but not sound. In his book The Myth of the Rational Voter, Caplan identifies four main widespread biases about economics: “Make-work bias”, “Anti-foreign bias”, “Pessimistic bias”, “Anti-market bias”.

7. The term RATIONAL IRRATIONALITY, coined by Bryan Caplan, sounds all kinds of oxymoronic, but the two instances of “irrationality” in the phrase refer to two different kinds of rationality:

a. Epistemic rationality: “forming beliefs in truth-conductive ways”

b. Instrumental rationality: “choosing effective means to attain one’s goals, given one’s beliefs”

8. So while the idea of rational ignorance is that it is *instrumentally* rational to be ignorant, the idea of rational irrationality is that it is *instrumentally* rational to be *epistemically* irrational.

9. The costs of overcoming ignorance are generally high because studying the relevant issues takes time and effort. The costs of overcoming irrationality are generally high whenever people simply *enjoy* being irrational. Bryan Caplan argues that people have “preferences over beliefs”: there are beliefs that people just like to hold whether they are true or not, and giving them up would be emotionally costly to them – often extremely costly. So if the expected rewards from giving them up aren’t high enough, they won’t.

I think the solution is using the Crowdsourcing Utopia project as a form of direct democracy to identify the most cost-effective solution to each public policy problem that government is intended to solve.

 Averaging guesses as to the effectiveness of a particular public policy solution to a problem would easily overcome plain ignorance as it would basically be seen as random noise and cancel itself out.  The few people with the right answer would pile on top of each other so that you could identify the right answer.  However, the problem of systemic bias remains.

The way to overcome this is to identify biased guessers and remove their guesses from the pool (aka filter out their biased guesses).  This would be done using the equivalent of a captcha.  Captcha are used to distinguish dumb robots from smarter humans.  The captchas used in the crowdsourcing utopia project wouldn’t be just fuzzy strings of text.  These captcha’s would be questions with a single, hard, correct and numerical answer that a biased person would have avoided integrating into their thought process in an attempt to avoid cognitive dissonance.

Let’s look at the 4 biases to be identified and filtered out.

Make-work bias

Caplan refers to the make-work bias as a “tendency to underestimate the economic benefits from conserving labor.”[1] Caplan claims that there is a tendency to equate economic growth with job creation. However, this is not necessarily true, since real economic growth is a product of increases in the productivity of labor. Dislocation and unemployment can be caused by productivity gains making certain jobs no longer necessary. All things being equal, economic rationality would require that these people make use of their talents elsewhere. Caplan makes special emphasis of the movement away from farming over the past two hundred years—from 95% of Americans as farmers to just 3%—as an illustrative example.As an economy industrialises, increased labour productivity in agriculture means less labour is needed to produce a given quantity of agricultural goods, freeing up labour (a scarce resource) to be employed in the production of manufactured goods and services.
To identify a voter with this bias, you could ask them,
“What percentage of Americans were farmers in the year 1800?” and
“What percent of Americans are farmers, today?”
People that are aware of these numbers are more likely to have realized that those lost jobs were a good thing because they resulted in 92% of the population being made available to provide new goods and services that people living in the 1800’s could never have dreamed of.

Anti-foreign bias

 Caplan refers to the anti-foreign bias as a “tendency to underestimate the economic benefits of interaction with foreigners.”[2] People systematically see their country of origin as in competition with other nations and are thus averse to free trade with them. Foreigners are seen as the “enemy” even if the two governments are at a lasting peace. The principles of comparative advantage allow two countries to benefit a great deal from trade. The degree of benefit is rarely equalized, but it is always positive for both parties. Caplan notes how the anti-foreign bias can be rooted in pseudo-racist attitudes: For Americans, trading with Japan and Mexico is more controversial than trading with Canada and England, the latter of whom speak our language and look like white Americans.You could identify anti-foreign bias by asking the voter, “What percent tariff should be imposed on goods from Mexico?” and “what percent tariff should be imposed on goods from Canada?”  If the tariff imposed on Canada is lower, then the reason is likely that they are racist to some extent.  Hence they should be filtered from the voting pool.

Pessimistic bias

Caplan refers to the pessimistic bias as a “tendency to overestimate the severity of economic problems and underestimate the (recent) past, present, and future performance of the economy.”[3]The public generally perceives economic conditions as declining. Caplan alleges that there is often little or no evidence to back up such perceptions.
To identify pessimistic bias, you could just ask the voter what the employment rate is or what last quarter GDP was. People with the pessimistic bias will overestimate the unemployment rate and underestimate GDP.

Anti-market bias

Caplan refers to the anti-market bias as a “tendency to underestimate the benefits of the market mechanism.”[4] In Caplan’s view, the populace tends to view themselves as victims of the market, rather than participants of it. Corporations, and even small-scale suppliers, are seen as greedy monopolists that prey on the consumer. Caplan argues that all trade is a two-way street. Cheating people is bad for business and the existence of multiple firms offering similar products demonstrates there is competition, not monopoly power.

To identify anti-market bias, you could just the voter if they would support using government price controls to keep prices down.  People with anti-market bias would support price controls without realizing that these inevitably lead to shortages.  THen you could filter their votes out.