The Instrument That Breeds Its Own Specimens

One notes, in the accounts of the Supreme Court’s decision to revive a controversial federal data system for checking citizenship, a figure that does not appear. It is the number of non-citizens the system has found voting. It is not that the number is small or large. It is that it is absent, and the machine is present, and one learns to pay attention when the apparatus arrives before the thing it was built to catch.

That is the first specimen. The others collect easily.

The system exists to filter alleged non-citizen voting. The word “alleged” does the work in that sentence, and it survives every retelling, including the ones sympathetic to the system. The allegation is disputed. The database is not disputed; it is merely controversial, which is a different thing. A controversial database continues to exist. A disputed phenomenon has to be demonstrated, and demonstration, one gathers, is what the database is for. So the evidence for the need is to be produced by the thing whose need was in question. This is a common arrangement. Churches were once built on the sites of miracles, and later the miracles were reported at the churches.

The case for the system deserves its strongest form. A citizen’s vote is diluted by any ballot that should not have been cast. A government that never checks cannot say it has nothing to find. Verification is not accusation; banks check signatures without believing everyone is a forger. The Court did not rule that fraud exists, only that the tool may operate. All of this is reasonable, and Fort has no quarrel with checking. His quarrel is with what the check returns.

Consider what such a system actually produces. It does not produce non-citizens. It produces mismatches: a record in one federal file that disagrees with a record in another. Citizenship is not a fixed property of a person, as far as the files are concerned. It is a status that changes, most often when someone naturalizes. Every naturalized American was once, correctly, a non-citizen in the government’s records. Whether a database knows that the status changed depends on whether one office updated and another noticed. The people most likely to show up as anomalies are those who did exactly what the law asked them to do.

So the instrument has a peculiar property. It will reliably return results. Some of those results will be errors of timing and clerical lag, and they will be counted, in the first press release, as “potential non-citizens identified.” The word “potential” will do the same quiet work “alleged” did. By the second release it will have fallen off, the way scaffolding comes down once the building is thought to stand. A naturalist would recognize this. It is the behaviour of a creature that secretes its own habitat.

It is worth watching the institutions involved as one watches a colony of social insects. The Court rules on whether the tool may be used, not on what it finds. The agencies that run it report how many records were processed, a number that grows impressively and means very little. State officials receive lists and must act on them, and an official who ignores a list has done something that can be reported later. Every member of the colony behaves sensibly. No member is responsible for asking whether the flags are true, because each has been told that another member checked.

One pictures a county clerk’s counter on some ordinary Tuesday. A voter has come in with a naturalization certificate, the heavy paper kind with an embossed seal, because a letter said the voter’s eligibility could not be confirmed. The clerk examines the seal. The certificate is real. The database is also real. The clerk has no procedure for deciding between two real things, only one for recording that the voter “resolved the discrepancy,” which will appear in the next report as a success of the system rather than an error in it.

A hypothesis follows, offered with no more credulity than the official account requires. Perhaps non-citizen voting is not a phenomenon the system detects but one it generates. It is continually produced from the friction between federal files, the way static appears on a radio that is not quite tuned. Such a phenomenon would never be confirmed and never go away, which is the ideal condition for any program that wants funding.

Fort does not claim this is so. He notes only that the record will soon contain a count of flags, and it will not contain a count of the people who stood at a counter holding a certificate, waiting for a clerk to decide which piece of paper the United States believed.