Colorado’s Creation of RLAs to Verify Election Results

Wayne Williams, a former Colorado Secretary of State, explains how the state moved from unverified machine counts to statewide risk-limiting audits. He starts with the wider election process, arguing that every voting method has weaknesses, and says machine tabulation is more accurate than hand tallies but still needs independent checking. Older audits sampled a few precincts, so they could miss errors elsewhere in the state.

He then walks through the method. Sample size depends on how close a race is rather than on the total number of ballots, and each sampled paper ballot is compared against a machine-generated cast vote record, which the new voting systems had to produce. Colorado passed the law in 2009 but did not reach statewide use until 2017, after counties adopted compliant systems and an Election Assistance Commission pilot helped. Staff worked with small counties on new deadlines, used an open-source sampling tool built with an outside group, and drew random seeds from ten-sided dice rolled in public. The risk limit tightened from nine percent to three over time, and failed rounds lead to more sampling and eventually a full hand count.

His closing point is that an audit does not solve every election problem, such as ineligible voting or voter suppression, but it gives the public a verifiable basis for trust. He encourages other states to review their own post-election audits and adopt the parts of Colorado's approach that fit their systems.