The Snapshot Protocol is designed to detect election fraud, in jurisdictions that use durable paper ballots (that can be recounted) and a post-election audit, a manual recount a fraction of the precincts and all races. This includes all counties in the state of California and a few other states.
California implements post-election manual tally procedures. These are intended to make sure the electronic devices are working correctly, and nothing more. Specifically, this process is not intended to detect election fraud due to compromised election officials or hackers. However, if we collect an election "snapshot," (a data file describing the election to the precinct level) at the right time, we can use these procedures as a check for central-tabulator election fraud which can flip elections. The protocol is easy to implement and is not unduly time consuming. We hope this protocol can be deployed to other election districts throughout California and other states where post-election audits are required.
Conducting this oversight protocol is important in itself to stop election fraud. If any potentially compromised employee or hacker knows an oversight group is conducting this protocol, they will be less likely to attempt such activities when there is a threat of being caught.
The steps in the oversight protocol are:
Send a letter to the Registrar of Voters informing them of our requirements
In San Diego, For June, 2014 election we sent this on May 15, 2014, and we received confirmation that they will be cooperative with our oversight requirements on May 28, 2014.
In November of 2014, we sent letters to all 58 counties in California to enlist their support of the protocol, and many cooperated.
Meet at the Registrar of Voters office the day after the election OR have the election office upload the snapshot file to our FTP Server (or access it on their website).
On June 4, 2014 we:
Picked up a CD with a full snapshot of the election as of the counting on election night.
Assisted with and ensure the random selection of precincts for the 1% manual tally.
We also fully video recorded the actions we take at this step.
We also downloaded the (XML) report of the current results of the election
(The snapshot file (173MB) was reduced to a spreadsheet (2MB).
For Nov, 2014 we:
Repeated the steps above for San Diego plus interacted with many Counties to get their cooperation remotely.
On June 9th, and Nov 10th, they started the 1% manual tally.
We visited the Registrar's office again and witnessed and video recorded the start of this process and some of the actual tallying.
Registrar Michael Vu resisted video recording but he finally did allow it in June. In November, he agreed to videorecording without any resistance.
Access the 1% Manual tally report, and create our oversight report.
Completed for June 2014.
Pending for Nov 2014
If we detect manipulation of the election, will pursue it at this point, probably in the courts.
We did not detect any election fraud in this election.
This protocol can be implemented in any jurisdiction that includes a post-election audit similar to the 1% manual tally required in California. The following map shows the current state. Please note that the light gray color in the legend relates to the lighter gray colors in the map! Primarily, we can focus on: Alaska, Connecticut, Montana, Minnesota, Missouri, New Mexico, and West Virginia. Also, the states of Nevada, New York, Oregon, and Wisconsin are also probably compatible. We will need to review the actual procedures used in each state to confirm that they are indeed compatible.
Snapshot Results Ca San Diego 2012 Nov - Although we attempted to implement the protocol in this election, the Registrar Of Voters did not provide the snapshot CD until AFTER the random selection, so the protocol was broken at that point. We accepted the CD but did not process it any further due to that critical error.
California counties are now required to use durable paper ballots and perform a one-percent manual tally. The manual tally was required by the Secretary of State to check that the computers are all working correctly; the manual tally was not designed to detect fraud. This Snapshot Protocol will provide a means to detect election fraud if it is occurring to a significant level. We know that this method does not eliminate all chances for fraud, but it is important that the Registrar Of Voters is aware that we intend to review it in this manner will put a check on some possible abuses. Hopefully, they will realize that fraud may be detected and they will not even try it.
The one-percent manual tally is based on a preliminary snapshot of the election, created within days of election night but before the election is certified. In the past, the Registrar Of Voters would not include in this snapshot ballots that included any write-ins, Vote-by-mail (VBM) ballots or provisional ballots, and the sample size is known to be insufficient to be used as a check on the election, but it still usually will mirror the election quite well.
Traditional practice has been to select a set of precincts for manual tally by using a "random" selection process. Then, the operators of the central tabulator would produce reports of these precincts and provides those to the staff that will be performing the manual tally. There are usually about 17 precincts chosen where all the races are tallied, plus more are selected to make sure that all races have at least one precinct tallied.
Although we are not accusing anyone of any impropriety, we must assume certain possible manipulations of the vote to design a means to detect it. One manipulation can occur by compromised Ro V staff in the central tabulator by shifting votes from one candidate to the other. Without a full recount, this is very difficult to detect. A full recount is not automatic unless the vote count is close. Therefore, if this manipulation is used, it will be a significant modification such that the election will not undergo any form of full recount.
But the election is still subjected to the 1% manual tally, which could potentially detect a significant modification. If the election was fraudulently manipulated by Ro V staff using the central tabulator as mentioned, then it would be possible to cover that up in the 1% manual tally by producing "good" reports (undoing the manipulation) for the precincts in the manual tally while allowing many other precincts to remain compromised.
The Snapshot Protocol requests that the Ro V supply our team a CD with a preliminary report of all the precincts PRIOR TO SELECTION for the manual tally. Thus, any precinct may be chosen for the manual tally and the election staff cannot "fix up" the data for the precinct (removing any election fraud) prior to the manual tally. The data on the CD should match the manual tally. If the manipulation occurred prior to the manual tally, then the snapshot review will discover errors in the 1% manual tally. If the manipulation occurred after the 1% manual tally, then the tally will not show an error, but the results in the snapshot data will be different from the final election results.
Assuming a worker at the Registrar Of Voters is compromised, for example, with this snapshot comparison in place, it will be much more difficult to perform such electronic election fraud. If widespread fraud is occurring, it will likely also occur in the precincts chosen for the manual tally, and thereby detected. Therefore, compromised actors will not be able to perform their fraud.
With the snapshot protocol in place, the only other opportunity for fraud would occur in ballots added to the total after the manual tally occurred. It will be instructive to compare the preliminary snapshot with the ultimate results of the election. If any races flip due to those final ballots included, fraud may exist.
The following protocol uses the terms "Registrar of Voters" (or "ROV") and "1% Manual Tally" or "Manual Tally." These are specific to San Diego County and the state of California. Other states may use other terms, such as "Post-Election Audit," "Manual Count," etc.
Send a letter to the Registrar Of Voters informing them of the oversight protocol we will be using and make specific requests, (see sample) as follows:
Request a snapshot of election data on a CD. This must be made available for pickup prior to the random selection of manual tally precincts. This snap shot must be of the stage of the election that is also used for the manual tally. In other words, no additional ballots can be added to the election results until after the manual tally is complete. This CD MUST be received prior to doing the random selection. If it is not received prior to the random selection, then the protocol is broken.
Request to observe (and possibly participate in) the random selection of the precincts to be used in the manual tally (probably about 15 precincts in San Diego).
Request to observe the manual tally process. (You don't need to observe the whole thing, but you should understand what is going on).
Request the report of the 1% manual tally including any discrepancies with computer counts. We are particularly interested if any reruns of computer counts are performed with differing results. If the final results differ from the snap shot, that information should be included and why it occurred. Also request checklists, logs, and work-papers created in the manual tally process.
Download the official report of the election after election night from the Registrar's website. (In San Diego, we "view source", copy all, and save to a text file.)
A group of team members should pick up the CD just prior to the random selection of manual tally precincts. A video camera should be used to document this.
NOTE: In a prior attempt to perform this oversight procedure, they did not provide the CD until just after the random selection of manual tally precincts as they were "still working on it." After reviewing that procedure, we realized it was absolutely necessary that we receive the snapshot CD prior to the random selection of the manual tally precincts, or the protocol is broken and cannot be used to detect fraud.
Team members should observe (and participate in) the selection of the manual tally precincts. It is okay if you are the outsider who rolls the dice, pulls ping pong balls, or whatever they use. Also video record this process.
Team members should observe part of the manual tally process to understand the process. Video record your visits for documentation.
The final report should be accessed of the manual tally results, and other checklists, logs and work-papers accessed from the ROV.
Note: In the past, the report was vastly incomplete because it did not provide the actual election results for that precinct. But since we have the snapshot prior to the start of the manual tally process, that snapshot is also the partial result of the election that corresponds to the manual tally.
Data analysis. After the election is complete the following steps should be completed.
Vote counts for the precincts used in the manual tally should be extracted from the rest of the election data on the CD.
Citizens Oversight has developed some computer programs to parse the election data and create concise spreadsheets, and to pull out the precincts utilized in the manual tally process.
Review the Manual Tally report, taking special note if any reruns had to be performed.
If reruns were required (i.e. re-scanning the ballots) to gain agreement with the manual tally, then election fraud may have been attempted.
In this case, you should request that you inspect the tapes from the actual scans of those precincts. Take snap shots of the tapes to document them.
If the tapes differ from the snap shot, then election fraud (central tabulator results manipulation) has likely occurred prior to the manual tally.
Total up the results of the election using only the manual tally precincts from CD data. Compare with the actual results of the election.
If the manual tally matches the snapshot and no problems were found in the review of the manual tally (no reruns) BUT if the sample of the election represented by the 1% manual tally DOES NOT match the results of the election (such as if the results were flipped significantly), then this implies election fraud may have occurred.
If there are no discrepancies reported in the manual tally, and if the snapshot results are roughly the same as the election, then they probably did not attempt this type of election fraud, or simply lucked out and selected precincts that were not included in the sample.
Manual Tally Procedure as used by the San Diego Registrar of Voters
Manual tally procedure in a nutshell (COPS has several issues with this procedure and the Ro V has been notified).
Select dates for the random draw of precincts and the actual tally, notify the public by placing a notice at the front counter (and perhaps the website).
Random Selection of Precincts
Calculate one percent and round up.
Have scheduled observers attend the draw
The procedure uses three sets of balls from 0-9 and one set of balls numbered 0 and 1, representing the four digits, (ones, tens, hundreds, and thousands, respectively).
Shake the containers and draw balls manually.
Draw ones, tens, hundreds digits.
If the three numbers in step six is 650 or less, the thousands container is then shaken, otherwise if the number is 651 or greater, the selection is complete.
Probability of selecting any number 650 or less or from 1000 to 1650 = (0.1)*(0.1)*(0.1)*(.65)*(0.5) = 0.0325%
Probability of selecting any number 651 to 999 = (0.1)*(0.1)*(0.1)*(0.35) = 0.0350%
The only time this would be correct is when "Cut Off" (i.e. 650 in this case) is 666 so that 666/2 = 1000-666.
Probability of selecting any number 666 or less or from 1000 to 1666 = (0.1)*(0.1)*(0.1)*(.666)*(0.5) = 0.0333%
Probability of selecting any number 667 to 999 = (0.1)*(0.1)*(0.1)*(0.333) = 0.0333%
Therefore, this procedure does not produce exact results and is not generally applicable, however, results are approximately correct because 650 is almost the same as 666.
Selecting precincts randomly will not necessarily result in even sampling of the precincts. In the November 2008 election, no precincts were selected in the range from 1 to 453 inclusive while nine precincts were chosen in the highest numbered 450 precincts. The likelihood of that happening is about 1 chance out of 10,000 tries, which is rare indeed.
Record contests as selected by the random draw.
Select additional precincts for the manual tally to cover all remaining contests on the ballot.
Prepare for the manual tally, including:
Order summary reports for election night and the related ballots. (COPs note: the staff creating these reports will know which ones are being requested, and therefore, could conceivably prepare known good reports for those precincts while the balance of the election may be substantially modified.)
Conduct the manual tally
Each precinct as one three-person team, one calls out the vote and two tally on separate sheets.
Supervisor compares with summary report.
If there is a variance, they re-tally until they agree, but no more than three times.
Supervisor records results, variances, and causes (if knoswn) on consolidation log.
When the tally is complete, return ballots for secure archival