The Oklahoma State Capital

Oklahoma Libertarians had much to celebrate last month: Oklahoma's newest political party had concluded its first gubernatorial-primary runoff election with nearly 1,000 voters taking part.

There's just one problem: approximately 66% of their votes are easily reverse-engineered.

Though rapidly growing, the still-small size of the Oklahoma Libertarian Party means that ballot secrecy cannot be afforded to most of the voters who participate in its internal elections.

That's because, though Oklahoma Libertarians cast votes in 555 of the state's precincts, in 458 instances, or 82% of the precincts, the votes cast were unanimous in favor of one of the two Libertarian candidates for Governor.

This allows the reverse-engineering of the election and near-certain identification of approximately 611 votes out of the total 926 cast.

Oklahoma's election policies require the state to publish the names of voters who vote in each election. The state also publishes a precinct-by-precinct list of results. These two reports, once matched up, show that 365 Libertarian voters in 259 precincts cast their votes for winning candidate Chris Powell, with no votes cast for losing candidate Rex Lawhorn; Lawhorn won the votes of 246 voters in 199 precincts where no votes were cast for Powell.

The 315 remaining Libertarians who cast votes in non-unanimous precincts aren't guaranteed that their votes won't be similarly reversed-engineered. The state further stratifies its reporting by type of vote, e.g., those who early vote, cast their votes on election day or absentee through the mail. This stratification allows additional extrapolation from the extremely small sample size of each precinct; in only two precincts were more than 10 votes were cast.

Additionally, the candidates are well-positioned to further reverse-engineer the votes by comparing their campaign supporter lists to the state's report. If a candidate lost a precinct by a 2-1 vote, and he knows that a known-supporter voted in the precinct, then he's able to discover the identity of the two voters who voted against him.

Candidates Voted for Themselves?

Should there be any doubt, both Powell and Lawhorn appear to have voted for themselves. Both of the candidates' home precincts are amongst the unanimous precincts. Powell's home precinct went for Powell by a vote of 2-0 while Lawhorn won his home precinct in a 3-0 unanimous vote.

Republicans Also Impacted

This anomaly isn't just a factor for Libertarians. Republicans who voted in small or low-turnout precincts also experienced the same loss of secrecy.

This allows the reverse-engineering of the election and near-certain identification of approximately 611 votes out of the total 926 cast.

85 voters voting in 20 precincts cast a unanimous vote for winning candidate Kevin Stitt while nine voters in four precincts unanimously supported losing candidate Mick Cornett.

Additionally, even though they voted in non-unanimous precincts, the votes of many other Republican voters were vulnerable to reverse-engineering with minimal analysis.

For example: on election day, 16 Republican voters at the Johnston County EMS Building in Tishomingo cast their votes—all for Kevin Stitt. Their votes were augmented by six other voters from the same precinct who cast their ballots before election day: five of these voted for Stitt and one for Cornett. This brought the precinct's total to 21 for Stitt versus only 1 for Cornett.

So who was that lone Mick Cornett voter in a sea of Stitt?

The one Cornett vote was cast by one of the three voters from the precinct who voted early and in person at the county election board.

Of those three voters, two share the same last name.

Did the couple split their vote?

Or was it the third individual who would be destined to become the only Cornett voter in the precinct?

Either way, just this brief exercise has allowed the curious to unlock the "secret" ballot of nearly every gubernatorial voter in the precinct.

Anomalies

Those who seek to reverse engineer the vote must qualify their analysis with the realization that voters do behave unpredictably at times. This may including "undervoting"—not selecting the name of any candidate, or "overvoting"—spoiling a ballot by selecting more than one candidate in the same race.

While it's likely to be a small number, there were several potential undervotes in the Libertarian runoff from Libertarians who did not want to support either of the two candidates.

As of the time of this report, county election officials have not "credite" all voters; e.g., Oklahoma County does not appear to have finished documenting the identity of the voters who participated in the election. Until the "crediting" is complete, it's difficult to analyze the impact that undervotes may have played.

Proof in Court

The large number of unanimous-voting precincts creates potential recount-related complications for Libertarian campaigns.

In 2006 a Hughes County court reversed the outcome of a state House election after the losing candidate proved that multiple voters were both ineligible to vote in the election and had voted for the opposition candidate.

The losing candidate was able to prove this because the ineligible voters had voted in a unanimous precinct where all votes were cast for the winning candidate.

The court agreed, threw out the votes and flipped the election.