Random Draw Oversight Procedure
Citizens Oversight (2016-11-07) Ray Lutz
This Page: http://www.copswiki.org/Common/RandomDrawOversight
More Info: Election Integrity
, Election Team
, Snapshot Protocol
, Snapshot Protocol
San Diego random draw in 2014 election using ping-pong balls drawn from bins by a member of our oversight team.: (Click Image for full size)
The "Random Draw" is one of the most critical of oversight activities. If we do nothing else, we need to provide oversight of this one step and make sure we demand transparency. This step does vary from state to state. Unless otherwise indicated, the comments refer to the Random Draw as specified by the California Election Code, Section 15360.
The purpose of the Manual Tally in CA is specified by Section 336.5 of the election code:
“One percent manual tally” is the public process of manually tallying votes in 1 percent of the precincts, selected at random by the elections official, and in one precinct for each race not included in the randomly selected precincts. This procedure is conducted during the official canvass to verify the accuracy of the automated count.
Although not mentioned in the election code, the 1% manual tally can also thwart major central tabulator hacks which might flip an election. This is the sort of thing we are most urgently worried about.
If you can help us with this, please sign up here so we know you are on board: Sign Up
Step 1. Access Public Notice
Public notice, hopefully on the registrar's website is required 5 days in advance of the start of the manual tally procedure. Usually, this means prior to the "Random Draw" meeting. (However in AZ, they start the manual tally process with a proforma meeting and conduct the random draw later.)
(d) The manual tally shall be a public process, with the official
conducting the election providing at least a five-day public notice
of the time and place of the manual tally and of the time and place
of the selection of the precincts, batches, or direct recording
electronic voting machines subject to the public manual tally prior
to conducting the selection and tally.
For some reason, we find registrars are hiding their public notice, not posting it at the office, not on their website, etc. If they don't make a public notice on the website, then they are in violation and we want to know right away so we can make formal legal complaint.
Step 2. Ask for and get the Snap Shot file
In California, Citizens Oversight has sent a letter to the top 24 counties (comprising 92% of the vote) requesting the "snapshot data file." This is a computer file using the same format as is required by law for the final certified results, broken down by precinct, race, and ballot type. This file can be pretty big, so it is best to ask for it on a CD or just ask that they post to their website.
Please download the file from the website or ask for it BEFORE they do the random draw. We must confirm that the results are FROZEN, so they can't back-out any fixes to the data once the audited precincts or batches have been selected.
If they hand you a CD, document this in your video recording and check that the file is on the CD.
Note: The "Snapshot File" should contain the results broken down by precinct (or batch) and contain ALL the precincts or batches that will be drawn from. It should be possible to ADD UP the results in all the precincts or batches in this file and get to the grand totals they advertise as the preliminary or final results.
We have had trouble with counties that do not sort their VBM ballots by precinct and instead process them by batch, as they claim they can't produce a snapshot file that contains ALL the batches without a lot of work. This is a serious glitch in our oversight protocol, because it means their IT staff will know the selected precincts and only then produce reports. Assuming the employee is compromised, then they could back-out any fixes to the election in the selected batches, apply those fixes to other batches that are not chosen, and then provide pristine reports that will then match the manual tally.
So if we do not get the complete report or a dump of the central tabulator database, then our oversight protocol is broken and they could still fix the election without any notice. Given the BIG DIFFERENCE between VBM and polls ballots in recent elections, we suspect such fraud may be occurring but it is undetectable unless we get the complete Snapshot at that point in the election.
MULTIPHASE AUDITS (CA ONLY)
Registrar's have the option of conducting the 1% manual tally audit in PHASES (in CA only). For example, phase 1 might be all POLLS ballots and phase 2 might be all VBM ballots. If they do it this way, they may have to have MULTIPLE RANDOM DRAWS. You will need to go through all parts of this procedure multiple times. Please try to record each phase completely!
Florida conducts their manual tally audits AFTER the certification of the election and final results are posted. Thus, in Florida, the Snap Shot
file should be easily available as the certified results but it must be broken down by precinct to be valuable to us. In Florida, they choose only ONE race to tally. The selection of the RACE is the most critical. Make sure they do not use "slips of paper" or some crude method that can be easily fixed, see below in the section about the ten-sided dice.
Step 3. Document the Random Draw meeting
Please use a video camera or smartphone video camera to record the entirety of the random draw meeting. Make sure you get the audio. Do not let them conceal anything they are doing. This is a public meeting and you must understand everything. They should provide a procedure document they will use. If they deviate, then cry foul and let us know. Please also record all the names of the elections officials participating in the meeting (write these down).
If they attempt to disallow your camera in the meeting, cite Cal. Gov't. Code §54953.5(a)
Cal. Gov't. Code §54953.5(a)
(a)Any person attending an open and public meeting of a legislative body of a local agency shall have the right to record the proceedings with an audio or video recorder or a still or motion picture camera in the absence of a reasonable finding by the legislative body of the local agency that the recording cannot continue without noise, illumination, or obstruction of view that constitutes, or would constitute, a persistent disruption of the proceedings.
And Election Code §15360(d)
(d) The manual tally shall be a public process, with the official conducting the election providing at least a five-day public notice of the time and place of the manual tally and of the time and place of the selection of the precincts, batches, or direct recording electronic voting machines subject to the public manual tally prior to conducting the selection and tally.
So if they try to say that this is just like any other part of their operation, that is incorrect. If they try to bring out their own rule that apply to all parts of their procedure, those do not apply here because this is a "public process" and involves a noticed public meeting.
Step 4. Observe the random draw itself
We prefer the use of ten-sided dice in the process. For example, if a county needs to choose one of the precincts numbered sequentially from 1 to 1654, roll four ten-sided dice, one each for the ones, tens, hundreds and thousands digit. But for the thousands digit, if the number is EVEN then it will choose 0 and if ODD, it chooses 1 to save rolls. Roll ALL FOUR dice each time, and re-roll if the number is not less than or equal to 1654. DO NOT incrementally decide if the thousands digit die should be rolled. But if you start with the thousands digit, and they roll a 1 (odd), then if the second die is rolled and it is greater than 6, then they can stop right there and re-roll, and so forth. The only action you can decide to do based on the value of the die is to re-roll, and for every accepted roll, all four dice must be rolled.
If the (precinct or other) numbers are not sequential and have gaps, it is better to have a list of those and their assignment to sequential numbers that can be drawn IN ADVANCE of the selection. Please get a copy of this file or document that makes this correspondence. If the district uses SEQUENCE NUMBERS the correspondence with the actual precinct or batch numbers must be set (frozen) prior to the selection of sequence numbers. We need to stay and watch them convert the drawn numbers to the precinct or batch numbers (which are not sequential).
The various methods:
| ten-sided dice
|| Roll ten-sided dice, one for each digit. If the rolled number is out of range, reroll.
|| Must be a list of valid numbers set in advance which can be verified
| digits pulled using 0-9 balls
|| One container for each digit, with ten balls with digits 0-9 represented. volunteers pull balls from containers after being shaken
|| This method is fine as long as if they are redrawn if the number is out of range.
| Raffle using tickets
|| Raffle tickets are purchased printed with sequential numbers for each precinct or batch to be chosen, with a fixed conversion table. Once verified, tickets are ripped apart and placed in a raffle drum, spun, and then one ticket is drawn for each precinct to be tallied.
|| Used in LA and this is acceptable because the tickets are not separated until after they are verified.
| Raffle using numbered poker chips, balls, or slips of papper.
|| Same as above but they are difficult verify because they must be set in order prior to being mixed up, and may lend themselves to sleight of hand.
|| Not recommended.
| "program on the internet to select random numbers"
|| we have no way to know who makes that program, and such programs can be predicted as they do not use a truly random number generation method.
|| No! Ask them if he program they are using has been approved by the Secretary of State and that usually stops them.
| Pseudo-Random Generator (PRG) using random seed (software)
|| If the method is documented, uses a standard PRG algorithm which can be reproduced, then this may be acceptable. The Random Seed should be generated using ten-sided dice, for example.
|| Requires that we duplicate the PRG algorithm and test that with the same seed, we get the same numbers.
| File of numbers
|| Worker shows you a file and says this file contains the numbers selected
|| This is not a random process and must not be used!
This is supposed to be a transparent process. You must understand everything and if you do not, ask questions until you do. If they refer to any documents, you have a right to see them.
Make sure you understand how the selected precincts/batches corresponds to the ENTIRETY of the election. In CA, all ballots should have a chance to be randomly chosen for review. Our lawsuit in San Diego was about the approx. 39% of ballots they were excluding from the manual tally process. Take the total number of ballots in the election, divide by the number of ballots in a typical (precinct or batch) and this should be the number of total items to be drawn, then divide by 100 and round up.
Ask them these questions:
- Are all ballots included in the scope of the random draw?
- Are later-arriving VBM ballots included?
- Are provisional ballots included?
ADDITIONAL RACES: In CA, all races must have at least one precinct included in the random tally. Demand that you be allowed to observe those selection processes. WHAT THEY SHOULD DO: they will have number of precincts that are included in this race. Make sure you understand that these are the complete list and that they are choosing RANDOMLY and PUBLICLY. Say no to the idea that "We had a county employee do it because they are also a member of the public." We need you to volunteer to be that member of the public that has no ties to government.
Step 5. Document the numbers chosen.
Photograph the numbers chosen as they choose them.
Step 6. Submit.
Upload your video to You Tube
and tell us where it is and send us any photos and notes you may have taken.
A picture is worth 1000 words and a video is worth 10,000 words. Check out the videos below to see some examples. As these meetings can be repetitive, you can skip through accordingly.
Number of topics: 9
- (M1859) Random Draw in SF, 2018-06 -- , SFElections, 14 Jun 2018
- (M1858) San Luis Obispo Bars camera from Public Random Draw -- Ray Lutz, Citizens Oversight, 11 Jun 2018
- (M1853) 2018-06-06 Random Selection for San Diego County, 2018 Primary -- Ray Lutz, Citizens Oversight, 6 Jun 2018
- (M1721) CA_SantaClara Manual Tally Nov 2016 -- Lise Rehbock, Citizens Oversight, 2016-11-23
- (M1719) FL_PalmBeach Manual Audit -- Ray Lutz, Citizens Oversight, 2016-11-18
- (M1715) CA_San Diego Nov 2016 Manual Tally -- Ray Lutz, Citizens Oversight, 2016-11-15
- (M1696) Citrus County Random Selection and Audit Process -- Margaret Carlock, Citizens Oversight, 2016-09-09 FL_Citrus
- (M1689) Marion County, Florida choosing random race and precincts for tally of 2016-08-30 primary -- Margaret Carlock, Citizens Oversight, 2016-09-01 FL_Marion
- (M1662) San Diego County Manual Tally selection meeting, June 8, 2016 -- Ray Lutz, Citizens Oversight, 2016-06-08 Save time and skip through most of the drawing part!
THANK YOU FOR YOUR HELP!