After days of posting the number of invalid signatures for anti-gay Referendum 71, elections officials are retracting their counts, declaring that hundreds of signatures previously disqualified are actually valid.

Last night, the secretary of state’s office office had reported that the cumulative error rate was over 13.5 percent. "The maximum error rate that they can withstand is 12.43 percent, so they are currently exceeding that," spokesman David Ammons said. So it looked like R-71 was on a trajectory to fail to make the ballot.

But now the secretary of state's office is reporting that 11.63 percent of the signatures are invalid. At this rate, it could make the November ballot. So what happened?

Shane Hanlin, an assistant director of election for the secretary of state’s office, says that so-called "master checkers" have been reviewing signatures over the past week. Even though daily counts have been announced (and widely reported by media), these checkers may not make a final decision on the validity of a signature until days later. They are authorized to consider the reason a signature was initially disqualified, check the state database, and move an "invalid" signature into the "valid" category. Hanlin says that the state's five master checkers have taken this action on least 409 signatures.

There are three reasons a signature can be initially disqualified: (1) No matching voter is found in the state database; (2) the signature is a duplicate; or (3) the signature on the petition doesn't match the one in the state's voter database.

It's unclear why the signatures are being reversed. No computer-generated list of all the signatures that have been reversed exists, says Hanlin; however, crews are keeping paper records of the reversed signatures, should a dispute arise over the methods for qualifying the signatures. Anecdotally, he says, "Two master checkers say their best guess is that most of the signatures they are reversing are signatures that don’t match [the signature on file]."

But it can take a day or more for master checkers to review all disqualified signatures. "They take more time to do this and have more experience doing searches on middle names and partial addresses," says Hanlin. "They use more sophisticated techniques to finds a match. Those 409 signatures were overturned based on that process."

"We are learning as we go. We made a decision to report a daily total even though some of those signatures might be accepted the next day when the master checker reviewed them," Hanlin says.

Ammons adds, "The daily snapshots are accurate for that day, but subsequent review can change the error rate. Sorry this is confusing — it’s the first time they’ve been releasing daily tallies and reasons for tentative rejection of signatures."

The secretary of state's office launched a new page yesterday that is routinely updated with the latest signature counts. Although, as of this moment, the page is essentially meaningless: the daily totals don't even add up to the cumulative totals.

Election officials need to pull this act together—soon. While their aims for transparency are respectable, the numbers don't even add up anymore. They cite 409 signatures that have been reversed, but none of the figures on their website show how they reached that figure. The final count is slated to be complete by the end of next week; however, if this slapdash reporting continues, the dispute will be tied up in court far longer.