The rise of democracy is a bumpy process, making irregular jumps forward e.g. at the end of World War I and after the fall of the Berlin Wall. The reason is that the introduction of democracy is not a Pareto improvement, because the old dominant coalition ruling the country loses out. It therefore requires a revolutionary moment. We examine the complexities of preference aggregation, highlighting theoretical challenges such as the Condorcet paradox and Arrow’s impossibility theorem. The median voter theorem applied to voting on the tax rate shows that democracy leads to a more progressive and Kaldor-Hicks efficient tax system than a dictatorship. However, the median voter model might describe reality less accurately than the voter paradox model, where the critical issue is whether a voter turns out to cast their ballot rather than for which candidate she votes. We also discuss how lobbying can distort democratic outcomes, favoring concentrated interest groups.