State Election Board Report – Post-Election Executive Summary

January 12, 2021

Introduction

Seven Hills Strategies, LLC (SHS) has been contracted by the State Election Board (SEB) to serve as an
independent, non-partisan monitor for the pre-electoral processes in Fulton County leading up to the
November 3, 2020 general election and for any subsequent runoffs. SHS will observe absentee ballot
request processing procedures, absentee ballot processing/scanning, early voting procedures, and actual
ballot counting on Election Day and beyond. The goal of SHS is to ensure that Fulton County is
adequately prepared for the scrutiny that they will be subjected to due to the national implications of the
election results.

Fulton County’s Compliance with the Terms of Sec. 12 of the Consent Order

In addition to this report on compliance with the terms of Consent Order, SHS believes that is necessary
to share this observation: From October to January, I spent nearly 270 hours at various locations
observing every aspect of Fulton County’s election processes. At no time did I ever observe any conduct
by Fulton County election officials that involved dishonesty, fraud, or intentional malfeasance. During
my weeks of monitoring, I witnessed neither “ballot stuffing” nor “double-counting” nor any other
fraudulent conduct that would undermine the validity, fairness, and accuracy of the results published and
certified by Fulton County.

A) Absentee Ballot Procedures:

1) Leading up to the Nov. 3 general election, SHS had the opportunity to observe the signature
matching processes for absentee ballot applications being processed both at Darnell Senior
Center and at Fulton County headquarters. During the runoff, I was stationed at Georgia
World Congress Center (GWCC) and was able to monitor the vast majority of signature
matching for the weeks leading into the runoff.

SHS determined the signature matching processes to be in-line with the terms outlined in the
Consent Order, and generally erred on the side of “give it further research” when there was
any doubt about a signature’s authenticity.

However, although most applications were being processed within 48 hours of being
received, SHS found one ballot application at Darnell Senior Center that had been in
Fulton’s custody for more than two weeks. Given the massive influx of applications and
ballots, it is not surprising that a few ballots might be left behind, but Fulton must re-double
their efforts in future elections to speed up processing times.
 Additionally, SHS received multiple reports of absentee ballots being sent to the wrong
addresses, which seems to be the fault of sloppy data entry by staff. Future staff trainings
should underscore the importance of correctly entering the temporary/preferred addresses of
all ballot applicants.

2) Although Fulton County allocated ample resources for absentee ballot processing leading
into the general elections, the processes themselves were extremely sloppy and replete with
chain of custody issues as the massive tide of ballots bounced around the Fulton Gov’t HQ
building.

The system, created by Ralph Jones, Registration Chief for Fulton County, seemed to
function, but there were many processes that seemed to be ad hoc solutions to problems
caused by a lack of organization or permanent staff with the expertise to manage the system
in place. For example, the room which housed the team doing additional voter verification
was also a temporary housing location for ballots between the mail room (which receives,
opens, and records the numbers of ballots) and the ENET processing room. Staff in this
room seemed to not understand the process, and Jones had to intervene to stop a temporary
staffer from moving a pile of recently-accepted but unverified absentee ballots into the stack
to go straight to State Farm Arena for scanning/counting. Had Jones not been there with me
to catch this mistake, it is safe to assume that those ballots would’ve been counted as if they
had been verified.

I observed an additional security issue here, as one staff member told me that people had not
been signing out batches of ballots as they moved around the building in trays between
processing rooms, which is a clear failure in the chain of custody mandated by the O.C.G.A.

Given the inefficiencies of this system and the volume of absentee ballots received, there
was no way that Fulton could possibly comply with the mandate to “process all absentee
ballots by the close of business on the next business day after the ballot is received.”

Despite the aforementioned deficiencies during the general, Ralph and his team were able to
both streamline and improve processes for the Jan. 5 runoff. The Fulton team migrated the
entire signature verification process to the facility established at GWCC and for several days
even attempted to do the voter credit step on-site before resolving to handle that at Pryor St.
before bringing credited ballots to GWCC. Performing the entire process1 linearly and in
full view of the public was a tremendous improvement on the labyrinthine system concocted
for the general. In my opinion, Fulton clearly made available sufficient resources to handle
the influx of ballots for the Runoff.

1
Voter credit à 1st pass signature verification with ENET à 2nd pass signature verification with RocketFile à
Return RocketFile rejects to Pryor St for curing
 3) SHS has not yet been able to conduct an audit to graphically represent the rate at which
absentee ballots were scanned for the general election; however, my research indicates that
the staff was able to scan fewer than 80,000 ballots in the period leading up to Nov. 3.
Judging by final absentee/UOCAVA numbers (approx. 147k), in the 72 hours from 11/3 to
11/5, the staff were able to scan nearly 80 percent (approx. 67k) of that which they had
scanned in the previous two weeks. Regardless of whether the bottleneck was in receiving
the ballots, verifying the signatures, opening the ballots or scanning them, this rapid
acceleration in scanning rate indicates that Fulton failed to adequately utilize the pre-
scanning period allowed by SEB Emergency Rule 183-1-14-0.9-.15.

The runoff, however, was a stark dichotomy and a comparative great success. With the eyes
of the world watching, Fulton was able to report 106,117 absentee votes (the vast majority)
on Election Day itself due to the diligent pre-scanning work by Fulton staff. By the time that
the operation was closed at 2 a.m., Fulton had fewer than 5,000 absentee ballots left to
process. This small remainder – all received from ballot drop boxes on the evening of Jan. 5
– is a testament to how hard the Fulton team worked to comply with this item in the Consent
Order.

4) Based upon a conversation with Captain M. McHugh, Fulton County Police Department,
regarding the security protocols installed to ensure the protection of ballot drop boxes, I am
confident that Fulton’s robust security architecture made it impossible to tamper with votes
at ballot drop boxes.

Given the daily influx of new ballots to the GWCC facility, I believe that ballots were, in
fact, collected each day as required by SEB Emergency Rule 183-1-14-0.8-.14. On Election
Day, multiple shipments of drop box ballots were received at GWCC (one at 4:38 p.m. and
another at 11:30 p.m.) after first being checked-in at the Pryor St. mail room. As far as I
witnessed, Fulton fully complied with this item of the Consent Order.

B) Poll Workers and Poll Worker Training:

1) Fulton greatly exceeded the target number (2,200) of poll workers required for both the
November and January elections. Fulton enlisted so many poll workers to account for any
potential emergencies, attrition, or no-shows on Election Day.2

Poll Workers Poll Workers

Indicator Target
Assigned (Nov.) Assigned (Jan.)
Dual Manager 81 81
Manager 174 174
Assistant Manager 510 510
Line Manager 558 525
Clerk 2,420 1,495
2
N.B. This point also covers Section 12.B.4 of the Consent Order
 Deputy Registrar Clerk 255 155
Provisional 30 17
Total 2,200 + 560 alts. 4,028 2,957

2) On October 28, 2020, SHS attended the four-hour Fulton County poll worker training at the
North Annex Service Center. This training accurately and concisely reviewed all voting
implementation procedures, how to use Poll Pads and other hardware, and the test at the end
ensured that workers had actually learned the content.

A particular importance was placed on securing election materials and ensuring that all zip
ties and numbered seal stickers are appropriately installed and recorded at the beginning and
end of each day. In accordance with O.C.G.A Code, verifying the zero-count in the morning
and recording the final count at the end of the day were also underscored, though there was
no emphasis placed on the need to dually-sign these count receipts. Additionally, the trainer
underscored processes for keeping voting open despite technological issues, stating that,
“you can open the polls with one poll pad, one BMD, and one scanner; if you are not able to
open at 7a.m., immediately contact Fulton County and see if you need to fall back to
provisional ballots.” The trainer also frequently repeated that “we do not turn voters away” to
encourage poll workers to find a workable solution to any problems that may arise.

The sole training deficit that I recognized was regarding the Senate District 39 Special
Election. While it was somewhat odd that a primary election would be taking place during a
general election, this lack of knowledge was a failure to adequately train the trainers
regarding this special election. This lack of knowledge was passed on to poll workers, which
resulted in numerous complaints to SHS about a failure to offer voters the opportunity to
participate in this special election.

3) Fulton was to provide the SEB with weekly updates on total poll officers and alternates,
training, and allocation plan of poll officers to polling places, including contingency plan for
alternate poll officers for the November election, as well as any runoff election in this
election cycle. As these reports did not come to SHS, I cannot comment on this item.

C) Advance Voting Locations:

1) Fulton was required to have 24 early voting locations, but greatly exceeded this requirement
in both the general and runoff elections.

Indicator Target General Runoff

30 + 2 30 + 2
Early Voting Locations 24 mobile + 7 mobile + 2
outreach sites outreach sites
 Both Fulton staff and poll workers could have done a better job ensuring that ENET records
were kept up to date. Failure to keep accurate records of whether a voter had voted yet led to
a great deal of confusion at the polls during both the general and the runoff as well as
concerns of widespread voter fraud. Some human error is to be expected, but Fulton must
strive to reduce the number of these instances.

D) Election Day Logistics and Polling Locations:

1&2) Fulton was required to have 255 voting locations for Election Day, and met this
requirement in both the general and runoff elections. It is also worth noting that Fulton
established 91 new polling locations for this election cycle to meet this goal.

Indicator Target General Runoff

Election Day Voting
255 255 254
Locations

3) Fulton was to provide the SOS with their plan for Election Day distribution of election
equipment and poll officers no later than October 2, 2020. As these plans did not come to
SHS, I cannot comment on this item.

4) On October 29, Rick Barron shared early voting turnout data with the Gabriel Sterling, Chris
Harvey, and Blake Evans from the SOS’ office. Sterling ran the modeling through MIT’s
Election Data Lab allocation tool, and shared the results with Barron. Complying with this
term of the Consent Order, Barron then re-programmed Poll Pads and redirected election
materials to buttress any weaknesses revealed by the data model.

5) At no point during either the general or runoff did any polling unit run out of
emergency/provisional paper ballots, paper backup pollbooks, or required forms. In January,
three polling units (all served as both early voting and Election Day locations) received re-
supply from headquarters but never ran out of materials.

6) During the general election, Barron negotiated with the ACLU to provide 255 deputy
registrars to use ENET to cancel absentee ballots. During the runoff, this task was performed
mainly by a smaller number of non-ACLU deputy registrars. SHS received no complaints
during the runoff about unnecessary wait times related to not having additional dedicated
deputy registrars.

7) Fulton established three call centers with a combined staff of more than 100 people to answer
questions from poll managers during Nov. 3 and Jan. 5. My poll worker training encouraged
me to call the hotline if any problems arose while voters were casting their ballots.

8) After 9:30 a.m. on Nov. 3, no polling precincts in Fulton County had a wait time greater than
30 minutes. The same was true for the entirety of voting on Jan. 5. Both of these should be
seen as tremendous victories for the Fulton team, as they had allocated sufficient staff,
 resources, and procedures to ensure that all voters were able to cast their ballots quickly
regardless of where they lived in the county.

E) Technical Support:

1) Fulton trained 255 technicians for the general election, and additionally ensured that each
early voting site also had a dedicated tech aside from State Farm Arena, which had five techs
on-hand to manage their large number of BMDs. For January, Fulton trained 254 technical
support experts, but 22 did not report for work on Election Day for one reason or another.

F) Audit Preparation:

1) Fulton’s document retention processes at State Farm were adequate for protecting ballots
from tampering and the system of marking boxes with scanner number, batch number, and
date made it much easier to process during the forthcoming audit and recount.

2) Risk-Limiting Audit (RLA)

§ The scale to which Fulton prepared for the RLA was staggering. With a maximum of 174
teams of two processing ballots by-hand, Fulton completed the RLA more quickly and
accurately than anyone had anticipated. It is a testament to the team’s leadership that they
were able to keep feeding the processors while keeping accurate records.

3) Recount

§ As with the RLA, Fulton aggressively tackled the Recount and initially seemed as if they
would complete their recount more quickly than estimated. However, failure to comply
with approved technological procedures led to a server crash and significant, costly
delays that required the Fulton team to completely rescan all ballots once again.
Additionally, during the fourth count (the second lap of the recount), sloppy document
storage procedures led to confusion as box labels no longer had precinct names and batch
numbers on them but instead all said “ELECTION DAY.” This mistake therefore made it
difficult to ascertain which ballots had been missed while trying to solve the second
technical issue that resulted from accidentally naming two scanners “ICC 16” during the
fourth count. Until this point, proper ballot handling, storage, and manifest procedures
had been observed.

Appendices
• Appendix A – Challenges and Recommendations from the Entire 2020 Election Cycle
 State Election Board Report

Appendix A - Challenges and Recommendations from the Entire

2020 Election Cycle

I. The Pre-Election Period

• COVID-19 preparedness was obviously on the forefront of Barron’s mind. He and his
team had taken a multitude of steps to ensure that everyone was safely fulfilling all
required duties in the lead-up to Election Day, but the virus had taken a heavy toll on the
permanent staff leading the warehouse team. This caused several pivots and logistical
changes to protect the staff, but there was still concern that a team of new players would
be able to handle the tremendous workload as seamlessly as the high stakes of this
election required. SHS learned that the SOS office offered vendor support to mitigate the
breadth of the COVID outbreak, but this was offer was declined by Fulton.
• SHS received multiple reports that Fulton was slow to update MVP and give voters credit
for having voted by absentee ballot (both mailed in and deposited in a drop box). It was
imperative that - as the Consent Order mandates - the BRE keep accurate and up-to-date
records about who has voted in the publicly-visible portals lest they face double voting
problems. Reports have shown that this problem has affected both absentee and early
voters, so the problem was bordering on systemic.
• Additional training should be done regarding O.C.G.A. § 21-2-381(a)(1)(A-G) pertaining
to relatives or helpers filling out the absentee ballot for their temporarily out of state,
disabled, or elderly voters. SHS witnessed multiple staff having difficulty deciding how
best to handle family members and helpers requesting absentee ballots for others.
• SHS has received a multitude of reports of absentee ballots being sent to wrong addresses
even though alternate/secondary addresses were provided or already on file. One notable
case being from a servicemember currently serving out-of-state who felt disenfranchised
by Fulton’s inability to properly process his absentee ballot request. As witnessed at the
Darnell Senior Center, the data entry for processing absentee ballot requests can be
burdensome, but each entry much be triple-checked for accuracy to avoid careless
mistakes like this.
• On 10/23, SHS saw one absentee ballot request dated 10/07. While this was a lone outlier
and the vast majority of the ballot requests seen were dated 10/21, Fulton must ensure
that all absentee ballot requests are processed in a timely manner.
• In his press conference on 10/22, Barron stated that there was no wait time difference
between the early voting locations in the north and south parts of the county; however,
anecdotal accounts have said that there have been long wait times in the Alpharetta/Johns
Creek parts of the county.
• The Senate District 39 special election was a persistent problem. As witnessed during
SHS’ poll worker training, there was a failure in the protocol for the training of trainers
 that should be corrected in the event that this occurs in the future. This failure to
adequately prepare trainers regarding this special election led to a countless number of
voters not being able to participate in this election. SHS suggests that to fix this, Fulton
consider pivoting to an “opt out” instead of an “opt in” policy for these types of elections
so that all voters may participate regardless of whether or not they are aware of the race.
• There were myriad problems with the absentee processing system at Fulton Government
Headquarters, including:
o Failure of staff to understand the process of moving ballots around the office
o No chain of custody forms being used as ballots move from room to room
o Mask-optional policy putting essential staff at unnecessary risk for COVID
o Failure to sufficiently protect spoiled and rejected ballots in the mail room

• While touring the mail room at Fulton County Government Headquarters, SHS saw many
ballots set to be cancelled because they were returned to drop boxes without the yellow
exterior oath envelope. It should be stressed more clearly to voters that they must
precisely follow all the instructions on the absentee ballots and return both envelopes if
they want their vote to be counted. Since it is impossible to update the MVP of “naked”
ballots when they are processed, it is likely that some of the complaints that the SOS
office received about a failure to record ballots deposited at drop boxes were due to the
fact that voters failed to correctly follow the necessary protocol.
• While there was a large focus on the “Know Before You Go” Campaign and encouraging
voters to use the FultonVotes App to notify voters that their precincts may have changed,
it is concerning that SHS has received a report that Fulton waited until 5:51pm on
October 26 to mail 169,714 postcards notifying voters of changed precincts. SHS
received complaints that Fulton was “suddenly changing polling locations without
notice.” It would have been prudent to send these notifications earlier so that the news
did not surprise people already making plans for in-person voting on Election Day.
• On October 29-30, widespread power outages resulting from Tropical Storm Zeta forced
seven polling precincts to close on 10/29 and two to stay closed on 10/30. This
unanticipated closure surely had a negative impact on turnout numbers as early voting
came to an end on 10/30, but there was very little that the BRE could have done to avoid
this. In fact, it seems that they handled the crisis well by deploying the two mobile
voting centers to the downed precincts to help manage the flow of voters.
• On October 26, the ACLU raised concern that the Fulton office was sitting on 1,500 voter
registrations that for all intents and purposes seemed to voters to have gone missing. It
took two days before SHS was able to get an update from Ralph Jones, who said that
there were indeed 1,500 remaining voter registrations awaiting processing and that they
would be finished by the end of 10/28. This is cutting it far too close to actual Election
Day for new voters who are likely unsure of the process. These voter registrations should
have been processed weeks ago.
 • It was brought to the attention of SHS that Fulton has been using an outdated version of
Easy Vote to check in voters and keep ENET records up to date. All software used must
be updated to benefit from the latest bug and security patches.
• Fulton has leaned very heavily upon an army of temporary workers to fulfill the litany of
tasks that must be completed from logistics to processing ballots to scanning final results.
It would perhaps be best to offset this number of workers with stakeholders from the
local community who would like to get involved in the electoral process. By conducting
multiple interviews with temporary staff, it was made clear that some have no keen
interest in participating in this immensely-important process, which is perhaps to blame
for some of the sloppy clerical errors and logistical shortcomings that have plagued the
complicated electoral process. However, others (particularly those scanning at State
Farm) are the glue that holds the entire process together. It is the opinion of SHS that
several of these leaders should be hired full-time if the budget allows.

II. The General Election

• The 4-BMD unit transporters are not ADA-compliant if used for duplicating ballots.
People must stand for hours to duplicate and the screens are too tall to sit and operate.
One of the Fulton staff has a bad knee and uses a cane. She was saying that her knee was
hurting but she needed to keep working.
• The truth about what happened on the night of November 3rd between 10:30PM and
11:52PM continues to be elusive. GOP party poll watchers say that Fulton staff told
them and the media to go home (implying that they did so in order to count without
supervision). Fulton staff tell me that the poll watchers and the media just left when
Moss sent home everyone but the scanner team. A SOS investigator is involved, so the
truth will come out, but if the party poll watchers are correct, then there is a serious
problem.
• There were persistent chain of custody issues throughout the entire absentee ballot
processing system. Aside from the problems with the system at Pryor St (see executive
summary report), the fact that ballots were being delivered to State Farm Arena in
unsecured mail carts is very concerning. Protocol for securing ballots exists not only to
protect the ballots themselves but also to ensure that no ballot box stuffing occurred.
This problem was exacerbated by poor managerial processes by Ralph Jones, who failed
to do intake counts for the provisional ballots. Similar problems seem to exist at the
warehouse as well (e.g. poll pads for SC11). Fulton must bolster these processes to retain
faith in their process.
• The entirety of events on Saturday, Nov. 7 was plagued by the mismanagement issues. If
there had been a clear process on Friday, then perhaps that mess may have been avoided,
but the fact that no one verified the number of provisional ballots either at intake at State
Farm or at adjudication is concerning. Therefore, there was a possibility that 1) not all
provisional ballots made it to State Farm or that 2) some were missing because they never
did an intake count. It turned out that both were true. If Santé had not gone back into the
 office to look up her file on provisional ballots, what would have happened to the 17
ballots that remained at Pryor St?
• The process for equipment delivery at the warehouse is in desperate need of an overhaul.
SHS concurs with Barron that a digital check-in/out system would make the logistical job
much smoother. Monday evening was far too chaotic for an operation of that size, and in
the disorder, many mistakes were made that just caused more trouble for a team that was
already underwater. As a result, SHS has received multiple complaints about a lack of
sufficient numbers of ballot bags making it to precincts, which led to a chain of custody
issue before tabulation. Additionally, SHS caught wind of missing CFs (e.g. Palmetto)
after Election Day that had likely been misplaced due to inadequate check-in processes.
o Furthermore, if Fulton implements a new digital system, it must be used by
both the poll managers and the Fulton staff. The fact that a poll tech was able
to show me that 157 polls were still “open” in Fulton’s backend demonstrates
that they were simply not utilizing a tool that they either developed or
purchased. Working partially from two systems is a fantastic way to forget
mission critical materials.
• Staff not using correct terminology caused confusion on multiple instances, including for
this monitor attempting to audit Fulton’s data. In pre-election reports, Fulton reported that
they had “processed and scanned” 127k ballots. The term “processed” was used multiple
times and by different teams, which indicates organizational silos and led to confusion
because SHS thought that “scanned” meant literally scanned instead of having the
barcode read and processed through MVP. In actuality, few ballots had actually been
scanned in the pre-election period.
o This same problem was evident when a staffer told SHS that ballots had been
“found” instead of “cured.” It is a distinction with dire consequences.
• The entire Fulton team must be more aware of the optics of their actions in such a high-
scrutiny environment. It was a judgment call, but I still think that bringing ballots in
through the back door on 11/5 was the wrong call for transparency purposes. It would
have ignited a media firestorm if the Fulton team had not immediately held a press
conference afterward. By far the worst maneuver for optics occurred on Saturday in
using the OPEX cutters to count ballots. Aside from being slower than counting by hand,
this gave the impression to everyone (myself included) that they had found more ballots
after the deadline. I personally had to talk to the media and the party poll watchers, who
were all understandably concerned by what was appearing to happen, to tell them that
those were empty ballots being counted.
• In the “Provisional Ballot Recap Notice,” Fulton stated that 1,205 people were “Not
found in Express Poll, researched and found to be registered in Fulton County, U.S.
Citizens, and ballot not challenged.” Why were that many people not in the Fulton system
and required to vote provisional?
• The OPEX scanners require constant re-calibration. The machines being out of
calibration and failing to operate properly generated more work for workers that were
 already exhausted and stretched thin. Fulton should either insist that OPEX techs remain
available for service calls during election crunch time or dispatch a large number of letter
openers to vote processing centers as a backup plan for the inevitable failure of the
technology.

III. The Risk-Limiting Audit

• There were persistent chain of custody issues throughout the entire RLA process. From
ballots being left unattended in front of party audit monitors to unsealed bags being
transported for storage to zip tie seals being left unattended to not recording the seal
numbers placed on the ballot bags, Fulton’s system is plagued with these procedural
issues. They must strengthen their chain of custody systems to follow the strict guidance
in the O.C.G.A. code given the (inter)national significance of the processes happening
here.
• Additionally, regarding proper seals, Fulton staff complained that the stickers provided
by the Secretary of State’s office for sealing cardboard boxes do not stick to tape and
cardboard. I even noticed a few that had just fallen off boxes of absentee ballots. Would it
be possible to change vendors for these stickers and provide counties with something
more robust?
• Transparency is of utmost importance, and the party audit monitors are completely
necessary, but the parties must strengthen their vetting procedures for their monitors,
train them on the process they are observing, and brief them on their roles. Furthermore,
it is my suggestion that repeat offenders who show a frequent disregard for the rules
should be barred from serving as monitors again.
• Fulton was initially slow to report their numbers into Arlo because they only had one
login. Then, to catch up they overcompensated and assigned too many staff to work on
data entry. Is it possible to split the difference and provide Fulton (and the other large
population counties) with more Arlo logins from the beginning? Fulton leaders were
complaining that they should have more than the one they were initially assigned so that
they could better manage the workload.
• There was a clear training deficit for auditors working through the new audit process. For
future RLAs, additional guidance should be provided about how/when to use the manila
envelopes, what constitutes clear voter intent (checkmarks, bubbles, or x’s), and large
number batch counting best practices to remove as much confusion as possible from the
audit process.
• Following the procedure detailed in the training video, audit teams quickly ran out of
envelopes for write-ins, under-votes, undecideds, etc. It is imperative that the Fulton team
have a back-up supply of these envelopes for the next RLA so that their team does not
have to scramble to help those working according to official procedure.
 • Some of the precinct batches (particularly for early voting) were massive (3,500+), which
increase human error due to fatigue as well as call into question the policies regarding
leaving the audit table for necessary bathroom and food breaks. Is it possible to split
batches larger than 1,500 to mitigate these issues if proper ballot manifests are kept?

III. The Recount

• Cardboard seems to be an insufficient storage method for document retention. Glancing

at the boxes, it is clear to see that many of them have been crushed by the weight of the
other boxes on the pallets upon which they were loaded. Additionally, the leak at State
Farm Arena – though certainly anomalous – revealed the necessity for a more robust and
potentially waterproof system for document retention. SHS recommends using plastic
storage bins instead of cardboard for future election cycles.
• Generally poor records keeping led to a multitude of procedural problems for Fulton
throughout the recount process. The poor managerial decision at the Fulton warehouse to
reclaim ballot bags for then-upcoming December runoff and mix ballots of different types
(e.g. early voting and Election Day) together for “deep storage” required additional
rounds of scanning during the recount because the wrong ballots were scanned on two
separate occasions.
o In contravention to what they had done on Count 3 (during which they
labelled all Election Day boxes with the precinct numbers on outer labels), all
of Count 4’s Election Day ballots were simply being placed in boxes marked
“ELECTION DAY” but with no precinct information visible on the outside.
This became a problem later when they had to retrieve particular batches
because they had been overlooked during scanning. This may have produced
a chain of custody issue at the end as two Fulton leaders were sending out
individual ballot batches instead of full boxes to make sure that each batch
contained only Election Day ballots as expected. They were careful to
correctly complete the coversheets for each batch, but it would not be difficult
for a batch to be forgotten or fall to the wayside as it changed hands.
• Transparency is of utmost importance, and the party monitors are completely necessary,
but the parties must strengthen their vetting procedures for their monitors, train them on
the process they are observing, and brief them on their roles. Furthermore, it is my
suggestion that repeat offenders who show a frequent disregard for the rules should be
barred from serving as monitors again. Throughout the time at GWCC, the party
monitors flagrantly disobeyed guidance from Fulton staff and GWCC police regarding
the mask policy and taking photos/videos of the procedure. One monitor even yelled,
“THIS IS TREASON UNDER PENALTY OF DEATH!” in the face of a Fulton manager
who was simply trying to check his party monitor credential – which he turned out to not
have – for sign-in.
• SHS had received reports of several unsealed ballot bags, and hunted down the bag
numbers to investigate. SHS found four unsealed ballot bags that were clearly marked
 with zero counts on the exterior labels. For future best practices, it is encouraged that
staff seal every ballot bag regardless if it’s empty to mitigate accusations of “magic
ballots” appearing from thin air. Additionally, if ballot bins are empty, it is a good idea
to place the tops in them so that it is clear to monitors that the box is empty and is not an
unsealed ballot bin.
• Technological issues abounded during the recount. The server crash on November 29 was
a costly error caused by a failure to properly follow protocols for backing up and
uploading data to the servers. This mistake cost Fulton taxpayers several days’ worth of
staff time as the entirety of the ballots had to be rescanned for a fourth time.
Additionally, the small typographical mistake of accidentally naming two scanners
“ICC16” on the fourth count led to a great deal of confusion and another full day of staff
time for solving the problem. Fulton technological team must work more slowly,
carefully, and in accordance with all protocol to ensure that these mistakes do not happen
in the future.

V. The Runoff Election

• When it comes to communicating with monitors, it is encouraged to keep comments short

and to the point without much editorialization. Early in runoff proceedings, Ralph Jones
had a good faith conversation with several GOP monitors, one of which had declined to
sign the sheet that said he would not record in the processing center. It turned out that
that gentleman refused to sign because he was, in fact, wearing a recording device, and
recorded Jones’ answers to his question without Jones’ knowledge. These monitors then
submitted an eight page complaint to the SOS quoting long passages from their nearly 45
minute conversation.
• Monitors were very concerned about compact flash memory cards being left in scanners
in the L&A side of the warehouse. Additional training regarding election security
protocol is required to mitigate alarmist fears that these memory cards are arriving at
precincts pre-loaded with votes.
• Parties must fully brief monitors on their role and the appropriate limit of their duties.
Multiple monitors told me that they had been recording the license plates of the staff that
parked in the deck as well as on the L&A side of the warehouse as “evidence.” This
seems like a massive invasion of the privacy of the election workers. It is recommended
that Fulton County put pressure on the county wings of political parties to have greater
accountability for the actions of the people to whom they provide monitoring credentials.
• Fulton was having an accuracy problem due to the data entry required to verify the
signatures on received ballot envelopes. In order to improve both speed and accuracy it is
recommended that Fulton provide barcode scanners to all signature verifiers in the future.
These scanners allow workers to go directly to the correct voter page in ENET without
worrying about typographical errors. This system was deployed with great success
during the second half of the runoff.
• Fulton staff must be careful to accurately enter data into ENET. SHS received several
reports of voters receiving multiple absentee ballots (N.B. not ballot applications) during
the runoff. Additionally, there were widespread stories of voters showing up at the polls
and being told that they had already voted. Taking voters’ claims of not having voted at
face-value and as the entire system is built to catch double voting, the only logical
explanation for this problem is that an election worker incorrectly pulled voter
information in ENET at some point. Extra training on ENET accuracy must be conducted
in future elections.
• While I vehemently disagree with the assertion that proximity is tantamount to
transparency, it would have alleviated a great deal of stress on Election Day if Fulton had
initially provided more access to the party monitors. The floor was set up to allow more
access, but the potential of the “cattle calls” was not utilized until it became a necessity.
Additionally, SHS suggested that the blue barriers be removed from the UOCAVA
duplication station on 12/30, but my suggestion was not followed for the worthy cause of
ballot security. Unfortunately, the perceived lack of transparency led to a court order that
immensely disrupted Election Day processes. If Fulton had more actively allowed
monitors to approach election processes, then it would have been easier for them to see
that Fulton had absolutely nothing to hide. The resultant overcompensating backlash left
many staff fearing for their personal safety due to monitors violating the photography
rules, staff receiving threats on social media, and astoundingly poor mask hygiene by
monitors. Furthermore, the increased access to the ballot cage generated a considerable
ballot security concern due to the proximity of partisan monitors to ballots being
processed.
• A persistent impediment to continued processing was the rate at which ballots were
transported from Pryor St to GWCC. While most days that can be attributed to sending
all ballots that they had received, on Election Day there must be a faster turnaround.
Though three ballot bins had been delivered at 7:04PM, it was not until 11:30PM on Jan.
5 that five bins arrived at GWCC from the 7pm collection of ballot drop boxes. At that
point most of the election staff had already gone home due to a lack work, but the
massive tide of ballots to be processed made it impossible to finish processing in its
entirety on election night. If the Fulton team had dispatched the ballots sooner – even in
smaller batches – then perhaps everything could have been finalized on Election Day.
This same problem was repeated on Friday, Jan. 8. The tremendous GWCC team had
waited all day for provisional ballots to arrive from Pryor St, but it was not until 3:47PM
that four ballot bins were delivered. A large portion of the staff clocked out at 4:30PM,
but the remaining team was left working until 8:15PM to handle the workload while
shorthanded. This could have been mitigated by sending smaller batches of ballots as
they became available.