Indiana Is Purging Voters Using Software That's 99 Percent Inaccurate, Lawsuit Alleges (thedailybeast.com)

An anonymous reader quotes a report from The Daily Beast: More than 99 percent of voter fraud identified by a GOP-backed program is false, a study by Harvard, Yale, and Microsoft researchers found. Now Indiana is using the faulty program to de-register voters without warning. In July, Indiana rolled out a new law allowing county officials to purge voter registrations on the spot, based on information from a dubious database aimed at preventing voter fraud. That database, the Interstate Voter Registration Crosscheck Program, identifies people in different states who share the same name and birthdate. Crosscheck has long been criticized as using vague criteria that disproportionately target people of color. Now Indiana voters who share a name and birthdate with another American can have their registrations removed without warning -- a system ripe for abuse, a new lawsuit claims. Crosscheck's premise is simple. The program aims to crack down on people "double voting" in multiple states, by listing people who share a first name, last name, and birthdate.

Indiana has used Crosscheck for years. But until July, the state had a series of checks on the program. If Crosscheck found that an Indiana resident's name and birthdate matched that of a person in another state, Indiana law used to require officials to ask that person to confirm their address, or wait until that person went two general election cycles without voting, before the person's name was purged from Indiana voter rolls. Under the state's new law, officials can scrub a voter from the rolls immediately. That's a problem for Indiana residents, particularly people of color, a Friday lawsuit from Common Cause and the American Civil Liberties Union argues.

Comments

Re: Erm

[guruevi]

Itâ(TM)s closer to 1 in 23 for a birth date but that doesnâ(TM)t work if you include birth year.

So statistically speaking, if you have 100 people with the exact same First, Middle and Last Name born in the same year, on average 4-5 people per common name pair across the US will share the same birthday.

Not sure how this pans out across 300M people with a somewhat uneven birth year distribution but I highly doubt a few hundred votes are going to matter.

Re:Not a bug but a feature.

[Mr D from 63]

One number that might be good to know is the number of legal voters who have actually been de-registered under the existing system. From what I can read, that number is zero, and so the system may actually be working properly. If they have enough years of data to determine that every one that checked with the same address was a proper deletion, then that would support not waiting two years.

This lawsuit is about potential abuse, there is no evidence of actual abuse, and its not clearly stated exactly how it would be abused. More info needed.

Re:Still not a problem

[dwillden]

"Actual fraud is rare. This is solving a problem that does not exist."
No, Actual fraud convictions are rare. This because voter fraud is extremely difficult to prove at the individual voter level. You basically have to be caught by the Poll workers somehow recognizing that you are not the Darinbob that is their neighbor and then them managing to hold you there until the police arrive to arrest you.

If they don't hold you there and have you arrested on the spot, you disappear forever and they have no way to prosecute and thus no record of your attempt at voter fraud. If they don't catch you out at the poll, they might realize later that your voter signature does not match that on record and they disqualify the vote, but that's not counted as voter fraud, just a disqualified ballot. Or they may never realize that you committed voter fraud.

That it is rarely prosecuted does not mean it does not occur far more often than you are willing to admit.