A Micro Audit Case Study
In some jurisdictions the ballot images from the Voting scanners are available via FOIA requests. Jeff O’Donnell, a very talented individual has acquired several sets of those and he's written a program that makes it possible to crowd source a recount of the election in those jurisdictions. The people doing this are average Americans that have volunteered their time. The program is written so that more than one person has to agree on the results of a ballot before the count is accepted as accurate. It is a well designed program by a competent programmer with lots of experience in critical industries and across the integrated technology spectrum.
The first run through for his program was Harding County NM. This is a small county and was a logical place for the first run of a project of this nature. The left columns in the chart below are what the volunteers counted. The right column is what is officially reported on the NM State web site. The difference is small and easily written off as some kind of minor mistake - an honest mistake perhaps? In this case the five ballots with no selection were turned into votes, and another one magically appeared. It only changed the margin by a tiny fraction and likely would be unnoticed at all, and if it was, it would likely be ignored as insignificant. The point is that it never should have happened.
This is how the perception that things are turning blue can slowly be created and also add a few votes to the chosen column without raising much suspicion. This was a county with 500 votes. What could you hide in a county with 5K or maybe even 500K votes? I do not believe this was an accident. Either the logic and accuracy testing on the tabulators wasn't done properly, or the programming was designed so such situations wouldn't be tripped during such a test. I'd like to know the answer, unfortunately that software is proprietary and we aren’t allowed to look at it. The way we find out what happened there is to audit the system. I only selected two races for brevity, but more were affected.
Presidential Election |
Volunteer Recount |
Official State Total |
|
Senatorial Election |
Volunteer Recount |
Official State Total |
|
|
|
|
|
|
|
Trump / Pence |
316 |
319 |
|
Mark V. Ronchetti |
338 |
341 |
Biden / Harris |
176 |
179 |
|
Ben R. Lujan |
153 |
157 |
Jorgensen / Cohen |
3 |
3 |
|
Bob Walsh |
5 |
5 |
La Riva / Freeman |
2 |
2 |
|
No selection |
8 |
|
Tittle / Sandige |
2 |
2 |
|
Mismatch |
|
|
Hawkins / Walker |
0 |
0 |
|
|
|
|
No Selection |
5 |
0 |
|
|
|
|
Mismatch |
0 |
0 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
504 |
505 |
|
|
504 |
503 |