Oversight for 2020 General Election
Citizens Oversight (2020-10-29) Ray Lutz
This Page: https://copswiki.org/Common/OversightFor2020GeneralElection
More Info: Audit Engine
On this page, we will be collecting links to relevant information to assist in the audits of the 2020 General Election using Audit Engine or other methods.
- 2020 General Election Oversight Plan -- This document identifies the plan as established prior to the election by identifying battleground states and priorities based on the Pareto principle, based on the most populous counties.
Post Election Strategy
The strategy for oversight differs from the pre-election plan, because the outcome is now known, and priorities can be set by the actual margin of victory in the state. Counties involved in a ballot-image audit initiative can be selected based on weighted prioritizing methodology. The approach is similar to what might be done in a batch-comparison audit within a single county.
Most counties are heavily reliant on vendors to provide services to process the canvass, and with there being only two primary vendors, Dominion and ES&S, rural area that would not be able to modify the result on their own, can be potentially affected as a group. Thus, smaller counties should not be completely ignored.
- GA - extremely close result, where Biden led by a little over 14,000 votes out of almost 5 million cast.
- GA has a large number of counties (159) with many extremely small.
- The smallest 81 counties (78-159) have fewer voters than the margin of victory (14,000). But the most votes that could be swapped is half the number of voters.
- The next 36 counties (42-77) in size have fewer voters than twice the margin of victory.
- We can consider the top 42 counties which comprise 80% of the electorate, and leave out the other 117 counties.
- The other counties comprise about 1.2 million voters. If a few of these were compromised during the recount, the 14,000 vote margin could dissolve.
- We don't actually have to do the audit until AFTER they do the hand count, and then add more scrutiny to any counties that flip.
- Sampling counties, assuming it would take just 7 to cause the lead to dissolve, would require 48 of the remaining 117 counties for 95% confidence.
- As a compromise, we can deal with the top 100 counties comprising 94% of the electorate, in terms of requesting ballot images in advance.
- After thinking this over, it may be more reasonable to just send a letter askign them to preserve and secure the ballot images in case their county somehow magically flips the other way due to a whole bunch of marks that seem to have been added or new ballots discovered or some missing. This can be sent to all counties.
- FL - Miami-Dade and Osceola Counties had a surprising large shift from prior year toward Trump.
- 372,707 vote margin of victory out of 10.96 million by Trump would need a very large trauch of counties to cause.
- However, the state was thought to be a win for Biden by many pollsters, and Miami-Dade and Osceola counties are big changes in favor of Trump.
- This article documents this: https://www.nytimes.com/interactive/2020/11/04/us/elections/florida-election-results-by-county.html
- Broward county has a very large number of write-ins (nearly 45K) and so we can probably access the ballot images from that county.
- Scrutineers has already submitted records requests to all 67 Counties.
- PA - Extremely close result in favor of Biden by 54,000 votes.
- We suspect that tweaking may have taken place in the rural areas that were heavily tilted to Trump, more than expected.
- If these were fraudulently modified, they would have had to estimate in advance how much to move them, and they just did not go enough, in their view.
- PA has 67 counties. The top 29 counties have 86% of the voters, with the smallest county having 72K registered voters.
- However, the smallest 38 counties comprise 1.1 M voters
- These counties could have been involved in an attempted hack that just did not go far enough.
- Only 14K votes on the average spread out. Checking just a few of these randomly is prudent.
- MI - This state, along with GA, has no audits at all.
- Michigan has 83 Counties, with the largest 22 comprising 80% of the electorate.
- The margin of victory was 148K out of 5.4M.
- The smallest 20% of the electorate in 61 counties comprise 1.5 M voters.
- It would be feasible to hide that margin in those 61 counties, so we will need to randomly sample them.
- Will work closely with MERA and Jan Ben Dor on this.
Proposed letter to GA
The following text is proposed to be sent to the counties in GA in lieu of asking for all the ballot images. If the recount finds any discrepancies, then we will have recourse.
Public Records Requests
Public records results for information in the critical states, or due to concern that outcome differed from expectations.
- Please see Audit Engine for a description of the files required.
Broward County, FL
- Some of the specific settings we will need:
- vendor: ES&S
- style_num_low_limit: 1
- style_num_high_limit: 116
- writein_str: 'Write-in/Escribir/A lekri'
- page_length_inches: 17
- fuzzy_compare_mode: left_only <-- this will allow us to disregard the multilingual translations and work only with the first english portion.
- permute_option_list: False <-- Florida does not shuffle option names.
- layout: fully_joined <-- options box does not have any lines separating the options.
- pstyle_pattern: ^Ballot Style (\d+)$
- lookup_style_by_pstyle: True <-- This is a guess because the style masters do not show the usual left-edge barcode.
- yes_option: 'Yes/Si/Wi' <-- we may have to add this option, not sure.
- no_option: 'No/No/Non' <-- we may have to add this option, not sure.
Sarasota County, FL
Coller County, FL