Publicly Observed Verification Counts

[Deborah Sumner]

Selected Sample

Choose from 1) elections expected to be close 2) highly contested, high stakes contests most vulnerable to tampering 3 ) importance of race. 

In November, US President, US Senator, US Representative and Governor fit all the above criteria. Could hand count one or more and randomly select WHICH one.

Random Sample:

In communities with more than one computer, random selection (publicly observed) of computers to be checked (at least two preferred)

Random selection among all contested races (publicly observed). At least two checks preferred.

See attached file for description of "sort and stack" double hand count procedure, authorized by state and used in most state-conducted recounts.

Hybrid model for the Nov. 8, 2016 election:

Conduct TWO hand counts. 

1) Randomly choose among the four contests named above. (publicly observed)

2) Randomly choose among all other contested races on the ballot (publicly observed)

Moderators can add "intent of the voter" differences to those totals and report. 

If there is an inexplicable discrepancy between hand/computer totals of .5% or more, report the computer count total (with legitimate votes added) to the Secretary of State. Report the discrepancy to Secretary of State, State parties and candidates, (so they can decide whether to request recounts of their races)

Election officials or volunteers, who take the oath of office, can conduct the hand counts. Once the hand count totals have been verified (with the second count), the moderator checks to see if it matches the computer total (or does when the added "intent of the voter" votes are added.