information needed to update the database. Some states instead opted for a bottom-up approach, in which local jurisdictions continue to maintain their own registration lists but also provide periodic updates to a separate statewide system. Other states have adopted a hybrid architecture that combines elements of both the top-down and the bottom-up approach. Kentucky and Michigan had already implemented statewide voter registration databases before the enactment of HAVA, but most states have had to implement new systems to comply with HAVA.
Does HAVA mandate a particular architectural approach to the implementation of VRDs? This issue has been argued both in the affirmative and in the negative at length, and the committee takes no position on this question. HAVA does require that the control of the VRD be maintained at the statewide level. Thus, the committee believes that the particular technical architecture used is less significant than the structure of the actual workflow used within a state. For instance, a county or municipality that remains in full control over the registration process and over who is on or off the voting rolls is not comporting itself within the spirit of HAVA, which mandates statewide control. As a result, any assessment of whether a system conforms to the requirements and expectations of HAVA should consider the processes whereby information is entered into the system, verified, and maintained.
It should also be noted that although guidance regarding database structures or system attributes has been promulgated through the EAC’s Voluntary Guidance on Implementation of Statewide Voter Registration Lists,4 the guidance remains voluntary, and the agency charged with enforcing HAVA—the Department of Justice—has not issued guidelines or regulations of its own. Thus, state election officials may proceed at their own risk that some design decision might be challenged later as not being HAVA-compliant.