Random errors, dirty information, and politics

Rational voters' assessments of candidates and policy proposals are unbiased, but affected by random errors. "Clean" information decreases these errors, while "dirty" information increases them. In politics, most voting procedures weigh random individual errors asymmetrically. Thus, such errors do not counterbalance one another in the aggregate. They systematically affect politics. This illuminates the roles of political propaganda and interest groups. It helps to explain various puzzles in Public Choice, e.g., the frequent use of inefficient policy instruments. Institutional conditions are identified that shape the aggregate impact of individual errors and the politicians' incentives to produce dirty information.