List Maintenance Is a Policy

A voter register has to be maintained. People move, die and change names, and a list that is never corrected becomes inaccurate in ways that serve nobody.

The provision governing list maintenance requires that correction and constrains how it may be done, which is the right structure and a difficult balance to strike.

The statute forbids removal for not voting alone, requires notice before removal for change of address, and bars systematic purges close to an election.

Each of those is a considered protection responding to a documented abuse, and together they describe a process that is meant to be slow and reversible.

What the statute cannot control is the quality of the data that triggers the process in the first place.

There is a timing rule worth understanding because it is the one most often misdescribed in public argument.

Matching a register against another database is the standard method, and matching is where the errors are made, because names are not unique and records are not clean.

Systematic removals are barred within ninety days of a federal election, which sounds like a broad protection and covers only the systematic kind.

The administration data records how many registrations are removed and on what ground, and the ground categories are broad enough to conceal a great deal of variation in method.

Individualised removals continue throughout, so a register can still change in the final weeks, one record at a time, without breaching anything.

A jurisdiction matching on full name and date of birth produces very different results from one matching on name alone, and both report the removal the same way.

That distinction is invisible from outside and is the difference between a list being maintained and a list being adjusted.

The asymmetry in consequences is the part that gets least attention. A register containing a stale entry causes almost no harm; nobody votes in a dead person's name at scale.

A register missing a live voter causes a specific harm to a specific person, at a moment when it cannot be fixed, which is exactly when they discover it.

Those two errors are treated as equivalent in most descriptions of accuracy, and they are not equivalent at all.

The register carries the federal guidance and its revisions, which is the only continuous public account of how the standards have moved.

The notice requirement is the practical safeguard and it depends on the mail reaching a person who has, by hypothesis, possibly moved.

A notice sent to an address the system already suspects is wrong is a safeguard with a predictable failure mode, and it is the principal one in the scheme.

The provisional ballot is the backstop, which returns the problem to a mechanism this publication has examined and found to vary as widely as the purge rules do.

The enacted history shows the framework built in response to registration practices that were plainly exclusionary, and built as a floor rather than a ceiling.

States may do more, and the ones doing more are not the ones the framework was written about, which is the usual pattern for a national minimum.

The reform that would help is narrow and cheap: publish the match criteria alongside the removal counts, so a rate can be compared against a method.

A jurisdiction removing twice the share of registrations as a neighbour may have a cleaner list or a looser match, and at present nothing in the public record distinguishes the two.

Maintenance is necessary and the argument is not about whether to do it. It is about whether anyone outside the office can tell how it was done.