Summary
In three Antrim County precincts, the 1st Amended results of 5th November contained incorrect values for number of ballots cast. In two of them, the reported value was exactly double the assumed true value, 1552 instead of 776 in Custer Township, 1204 rather than 602 in Echo Township.
In the Wayne County Primaries of August 2018, Election Source’s ElectionReporting web-site crashed on the night of the election. Election Source’s CEO explained that processing delays caused some results to be doubled.
I ask whether the same phenomenon was at work on both occasions.
The 1st Amended Results
The Unofficial Results that were issued at 4:09am on 4th November gained their extraordinary character from a right→left vote transfer in 9 of the 16 precincts, resulting in Biden receiving Trump’s votes, and Trump receiving Jorgensen’s. In addition, there were major errors of a different kind in 2 further precincts, namely Elk and Milton Townships.
These errors appear to have been corrected in the 1st Amended Results released at 4:42pm on 5th November, but at the same time new errors were introduced. Whereas the values for number of cast ballots appear to have been correct in all precincts in the Unofficial Results of 4th November, this was no longer case in the 1st Amended Results for three precincts, namely Custer, Echo and Elk Townships. In all three the value was greater than the number of registered voters and, remarkably, in Custer and Echo the number of cast ballots was exactly twice the value reported in the 2nd Amended results, which may be treated for the sake of analysis as the true values, as a working assumption. The suffices 1, 2, and 3 in the table below refer to the Unofficial, 1st Amended, and 2nd Amended results respectively:
For a single number of the magnitude of 776 or 602 to be exactly doubled would be noteworthy. For two such numbers to be exactly doubled is highly suggestive of a common cause in some malfunction that resulted in a multiplication by 2.
The 2018 Wayne County malfunction
As I described in my last post, Election Source is one of two subcontractors named in Michigan’s contract with Dominion Voting Systems. Election Source operates an election reporting web-site, ElectionReporting.com, with interactive map which allows viewing of results as they come in for those Michigan counties which elect to subscribe.
This web-site reporting service broke down:
in the Wayne County primary election of 7 August 2018, with no results available on the morning of the 8th:
Jeffrey DeLongchamp, CEO of Election Source, said that the large size of the Wayne County results file caused ‘a 15 second delay in processing old and new data’, and led in turn to results sometimes being doubled:
DeLongchamp reportedly told the Wayne County Board of Canvassers a few days after the primary that the technical problems experienced were ‘totally separate’ from the vote-counting process:
If that was an entirely accurate statement then I think it would probably follow that the doubling of results in 2018 had a different causation to the doubling of results in 2020. But it seems to me inherently improbable that such a distinct and noteworthy phenomenon would be produced independently by two different routes. So I think it is a least worth investigating whether DeLongchamp may possibly have been mistaken.
Results Tally & Reporting (RTR)
Wayne County and Antrim County both chose to use Dominion Voting Systems (rather than Hart or ES&S) as their vendor. Both therefore employ Dominion’s Democracy Suite (D-Suite) 5.5 as their Election Management System (EMS), and therefore, as I explained in my last post, use D-Suite’s ‘Results Tally & Reporting’ (RTR) application for their post-election tasks, this being one of D-Suite’s two main pillars, along with ‘Election Event Designer’ (EED) for pre-election tasks.
RTR takes results from individual tabulators, aggregates them to produce precinct or county level tallies, and issues reports with totals for county or precinct as required. I showed in my last post that the incorrect 2020 Antrim County Unofficial and 1st Amended results were produced with RTR.
‘Election Night Reporting’
Dominion offers an optional extra feature called ‘Election Night Reporting’ to create a graphical display of the results. As is clear from the diagram below (Michigan contract, p. 116), the Election Night Reporting Module is additional to and separate from RTR, importing data which has been exported from RTR, as shown by the arrow from the leftmost cog wheel to the wheel on its right:
Election Night Reporting outputs the election results in EML (Election Markup Language):
thus enabling it to be used by news media to power their own displays. Alternatively, they can use the ‘Election Night Reporting’ display:
‘Election Night Reporting’ and ‘ElectionReporting’
A comparison of Dominion’s ‘Election Night Reporting’ web display (contract, p. 117):
with Election Source’s ‘ElectionReporting’ web display:
shows that Election Source, despite being a Dominion subcontractor, are not using the Dominion display.
Source of the data used by ‘ElectionReporting’
It is noteworthy that all 17 counties 1 that use or have used ElectionReporting:
also use Dominion (navy blue in the the Secretary of State’s map below) rather than Hart (sky blue) or ES&S (pale blue):
It follows that in every county for which Election Source provides its ElectionReporting service, the results displayed have been tallied with Dominion’s RTR. I doubt whether this necessarily has to be the case, since RTR exports results in standard formats including Excel and html (RTR User Guide, p. 9):
as presumably do comparable applications in the EMS of other vendors. It may simply be that counties are much more likely to use ElectionReporting if they are already in relationship with Election Source through using Dominion as their vendor.
What happened in Wayne County in 2018?
DeLongchamp said in 2018 that the problems were caused by the size of the file containing the Wayne County results:
He also said that data was being extracted from the site at such a rate as to prevent new data loading correctly:
He makes no suggestion that there could have been errors in the data file itself.
Now that we know from the Antrim County 1st Amended results that D-Suite’s RTR is capable of producing values exactly twice their their true values, is it worth considering the alternative hypothesis that it was Wayne County’s RTR that was producing the doubled values, which would then be contained in the exported data file, and that Election Source’s ElectionReporting web-site was merely reproducing these?
What happened in Antrim County in 2020?
Despite the attention given to the Antrim County 2020 election, there has to my knowledge been no recognition of the exact doubling of values in the 1st Amended Results, and accordingly no discussion or explanation of possible causes for it.
Does Jeffrey DeLongchamp’s statement that results were doubled in Wayne County because of a ‘delay in processing old and new data’ shed any light on the possible cause of the doubling of numbers of ballots cast in 3 precincts in Antrim County?
Gross errors in election results should be a matter of the gravest concern and every effort should be made to understand their causes, so as to prevent their recurrence while this type of electronic voting continues to be employed in democratic nations.
Andrew Chapman