UPDATE: Georgia law is likely to create confusion as the State moves forward with a statewide hand recount

NOTE: This blog includes discussions and explorations of legal issues. The content is intended for educational purposes only, and is not intended to be legal advice. You should always consult with a licensed attorney before taking any action that could affect or impair your legal rights.

Updated to include insights from Philip B. Stark and to place a greater focus on the somewhat murky treatment of “audits” and “recounts.”

As a lawyer, I believe it is important to use my privilege, as a lawyer, to help provide — where and when I can — answers and clarity to current events. Setting aside political considerations, the announcement made earlier today by Brad Raffensperger, the Georgia Secretary of State, is likely to cause confusion because of the murky distinction between the result of an “audit” and the result of a “recount.”

As of October 14, 2020, “[f]ollowing November general elections in even-numbered years, each county shall participate in a statewide risk-limiting audit with a risk limit of not greater than 10 percent as set forth in this rule prior to the certification by the Secretary of State.”[1] Ga. Comp. R. & Regs. r. 183-1-15-.04(1)1. [Effective 10/14/2020].  What is a risk-limiting audit (“RLA”)? As the Brennan Center for Justice explains:

in plain terms, more ballots are counted in a close race, while a race with a larger margin would require fewer ballots to be counted. If testing of the sample is consistent with the original vote total, it is almost certain that the initially declared winner won the race. If, on the other hand, the sample has substantial discrepancies with the original tally, the audit continues until there is “sufficiently strong statistical evidence that the apparent outcome is right, or until all the ballots have been manually counted.”

The result is that RLAs “can provide a high level of confidence in the results while generally requiring fewer ballots to be hand counted than what is already required in many states using traditional audits.”

While risk-limiting audits are usually implemented in the hope that a full “re-tallying” of the ballots can be avoided (assuming sample batches align with the reported outcome of the election), the risk limit for an RLA can be set to zero (if the original results can be superseded by the count produced by the RLA) — which simply means that every single ballot will be re-tallied. Under Georgia law, there is a ceiling on the risk limit (i.e., the likelihood of inaccuracy that would be tolerated by the audit) of “10 percent,” but there does not appear to be a floor on the risk limit (i.e., the likelihood of accuracy required by the audit could be infinite). Applying those confines means that there is a minimum number of ballots that have to be re-tallied in an audit, but there is no maximum number of ballots that can be re-tallied in an audit — which means that all of the ballots could be re-tallied in an audit seemingly without offending the restrictions placed doing a manual hand recount.[2] See Ga. Comp. R. & Regs. r. 183-1-15-.03(2)(a).

A recently amended State Election Board rule limits the use of a hand recount to two situations.[3] Ga. Comp. R. & Regs. r. 183-1-15-.03(2)(a). The first situation is if after doing tests to ensure the accuracy of the ballot scanners that were to be used for a recount, a “discrepancy” between a hand count of the “test deck” and the tabulation count of the test deck from the ballot scanners, “cannot be resolved. . . . If, after testing all available ballot scanners, there are no ballot scanners authorized to be used in the recount, the recount shall be conducted by manual hand count.”[4] Ga. Comp. R. & Regs. r. 183-1-15-.03(1)(c) (emphasis added). The second situation is if a court orders a manual hand recount.[5] Ga. Comp. R. & Regs. r. 183-1-15-.03(2)(a)2. The latter is pretty straightforward: there is either a court order or there isn’t. The former could be viewed as less so because whether the “all available” portion refers to the entire pool of ballot scanners authorized in the state, or it refers to a pool of machines that the Secretary of State has chosen, is somewhat ambiguous, but the consequences of how “all available” is interpreted could make a manual recount less likely or more likely, respectively by those interpretations.  

Now, all of this assumes that the audit and the recount occur separately. If that is the case, it seems unlikely that a re-tally all of the ballots in Georgia for the purposes of a recount could be done by a manual hand count without a court order or the “unresolvable discrepancies” as explained above. In which case, going ahead with a manual recount anyway could potentially jeopardize the validity of the recount, and potentially expose the Secretary of State to suits from counties not wishing to expend the time and money for a “recount” that could arguably be invalid.

Nevertheless, if, the audit and the recount are occurring simultaneously,[6] Presumably, deciding to do a full, manual recount at the same time as an RLA (i.e., before knowing the results of the RLA), is at least somewhat counter-intuitive if the purpose of the RLA was to reduce the number of ballots that must be counted (and resources spent doing so) and if the audit could be operated to a satisfactory risk limit without having to manually count every ballot. The results of the RLA could then be used as grounds for undertaking a full manual recount, if necessary, and hopefully, if unnecessary, avoid expenditures on full manual recounts that are not statistically necessary. On the other hand, if setting the risk limit at zero is the result of first-time jitters, the (perhaps) overly cautious approach of having a risk limit of zero is understandable. as was reported, it is somewhat unclear as to whether this situation would just be a loophole to the restrictions on a manual re-tallying, or whether a court might find greater fidelity to the law by determining that the two processes must be distinct in order to not defeat the plain language of Ga. Comp. R. & Regs. r. 183-1-15-.03(2)(a).

My hunch is that a judge is probably unlikely to make a decision on the merits of the case before the process is over, but a truly intransigent county (assuming they have standing to challenge Raffensperger’s call for a statewide manual recount) challenging the zero risk-limit as a pretextual manual recount in violation of Ga. Comp. R. & Regs. r. 183-1-15-.03(2)(a), could change that. The Rules on manual recounts and conducting an RLA appear to be somewhat incongruous. At first glance, because there does not appear to be a floor on the risk limit (i.e., the degree of precision required by the audit), doing an RLA with a risk limit of zero by hand — which means every legal ballot cast is hand-counted — might operate as a workaround to the regulation that requires a recount be performed manually only under a court order or a discrepancy between the hand count and the scanner tabulation of the “test decks” that cannot be resolved. The outcome of how a court might respond to a lawsuit challenging a recount mandated to be done by hand is probably more predictable than the reason that a court would use to arrive at such a conclusion. Whatever happens, it is highly likely that there will be a lot of eyes on the process.


References

1 Ga. Comp. R. & Regs. r. 183-1-15-.04(1)1. [Effective 10/14/2020].
2 See Ga. Comp. R. & Regs. r. 183-1-15-.03(2)(a).
3 Ga. Comp. R. & Regs. r. 183-1-15-.03(2)(a).
4 Ga. Comp. R. & Regs. r. 183-1-15-.03(1)(c) (emphasis added).
5 Ga. Comp. R. & Regs. r. 183-1-15-.03(2)(a)2.
6 Presumably, deciding to do a full, manual recount at the same time as an RLA (i.e., before knowing the results of the RLA), is at least somewhat counter-intuitive if the purpose of the RLA was to reduce the number of ballots that must be counted (and resources spent doing so) and if the audit could be operated to a satisfactory risk limit without having to manually count every ballot. The results of the RLA could then be used as grounds for undertaking a full manual recount, if necessary, and hopefully, if unnecessary, avoid expenditures on full manual recounts that are not statistically necessary. On the other hand, if setting the risk limit at zero is the result of first-time jitters, the (perhaps) overly cautious approach of having a risk limit of zero is understandable.