After a controversial campaign season in the U.S., the voting process will finally conclude on the evening of November 3rd. The pandemic has forced many citizens to avoid voting in person and many are using vote-by-mail for the first time. With over 50 million mail ballots expected to be cast, election officials across the U.S. are bracing for the avalanche of mail ballots.
Some politicians have cast doubt on the security of the entire vote-by-mail process. To make the job of counting mail ballots even more difficult, there have been a myriad of lawsuits filed and late changes to the rules on how mail ballots are validated.
This has made us all a bit nervous and losing sleep over several possible scenarios. What if the US Postal Service cannot deliver ballots on-time due to no fault of the voter; will those ballots still be counted? Will so-called naked ballots (those received without the security envelope because of simple human error) be disqualified? What if a signature is judged not to match the voter record and the ballot is disqualified; will the voter be given an opportunity to correct it or will their ballot simply be discarded?
No one knows exactly how this will all turn out or when the election results will be official. But there is one thing in this process that is certain. To prevent voter fraud, no mail ballot will be counted unless the voter is first verified.
To ensure each ballot is legitimate, the most common verification method – one employed by 33 states – is the use of signature match or review. Each signature on the outer ballot envelope is compared with the signatures from official voter records to determine authenticity.
Election officials and workers across the nation will soon hunker down in secure rooms, and they will painstakingly begin the arduous process of analyzing every signature on every mail ballot. When a signature is deemed not to match, that ballot is not counted.
The workload at many election centers will be overwhelming and could take days if not weeks to complete. Both political parties have lawyered-up and plan to challenge the process. The pressure is higher than ever, and signature decisions made will be more crucial than ever before.
In most elections that use vote-by-mail (also known as absentee) ballots, the signatures are all reviewed by humans. This causes problems, not only in the time and resources required, but also with the ability of humans to reliably compare signatures and do so consistently over time while possibly being exhausted.
In a recent Reuters article, Michael Herron, government professor at Dartmouth College was quoted as saying, “An error here means someone is losing their chance to vote.”
Make Every Mail Vote Count
Automated Signature Verification (ASV) provides a solution to this problem. Thankfully, ASV is used today in the vote-by-mail processes of many states to provide solid assurances that each vote is treated fairly and thoroughly reviewed. For this election, some of the larger cities and counties in America are deploying ASV software from Parascript to assist their verification teams to produce accurate results. Parascript has a history of success at preventing election fraud, which was highlighted in the recent NY Times article, “Two of These Mail Ballot Signatures Are by the Same Person. Which Ones?”
Parascript ASV uses AI and machine learning to read and match signatures. We are aware this might concern some voters who are wary of AI horror stories about programmed bias or bots out of control. Voters deserve to know “what’s under the hood”. Let’s demystify the Parascript engine by reviewing how it works.
What’s Under the Hood?
Automated Signature Verification is machine learning software trained on literally hundreds of thousands of signatures, both genuine and fraudulent, to identify the key characteristics of each and enable high performance analysis. The main advantage of automated signature verification stems from replacing or augmenting the error-prone human review. Gone are the risks that stem from exhausted humans, inconsistent training and different capability levels.
Instead, the same algorithm applies to all signature verification. The most advanced signature verification software takes advantage of artificial intelligence (AI) systems to imitate the type of analysis that humans perform. It combines this approach with the systems to imitate the type of analysis that humans perform.
Automated Signature Verification
SignatureXpert is Parascript’s award winning automated signature verification software that uses several different machine learning techniques in order to provide the industry’s highest level of accuracy at signature verification for elections. It can authenticate a signature even when there is significant variation between the signatures on a ballot and the voter registration. Conversely, it can also identify forged signatures even if two signatures look similar to the naked eye.
Automated verification is completed using a powerful combination of signature analysis algorithms called verifiers.
These verifiers employ multiple methods and principles to confirm authenticity. This include a human-like holistic approach to signature interpreting, feature extraction and comparison (using several neural network-based learning and interpretation agents), fuzzy logic and other advanced techniques. This process allows for the most comprehensive and intelligent analysis of signature characteristics, while taking into consideration the random variations that occur in signatures.
Parascript has also developed a unique and revolutionary approach that closely approximates the way humans read and write. This approach, inherent to SignatureXpert and all other Parascript products, uses a special descriptive language based on a set of formative characters, or “XR elements.” XR elements are a set of 64-character motions that can be used in any combination to form all the letters used in handwriting.
After the verification process, SignatureXpert issues a confidence value that indicates how confident the software is about the match between the signature presented for verification and the authentic reference signature. A high confidence value indicates a high probability that the signature presented for verification was written by the same person as the reference signature.
Is This Proven Technology?
In the case of Parascript, the answer is yes. The AI that powers SignatureXpert has been field-proven in the banking industry for over a decade and is trusted to produce reliable voting results. States such as Oregon, Colorado, Washington and Utah use it to ensure that every election is efficient, legitimate and secure. Voters can also be assured knowing SignatureXpert is always used to assist, not replace, election officials so there is no worry that the ASV bots will secretly throw the election!
For instance, after an incident involving forged signatures on petitions the state of Colorado overhauled signature verification and now has a process that includes Parascript ASV to ferret out the fraud attempts. Denver County, the largest in Colorado, now processes every ballot through ASV software, which provides automated verification on a large percentage of ballots, leaving only a smaller segment to be reviewed manually. This process has been successfully emulated by several other municipalities in Colorado and other states.
No matter which candidate wins the 2020 election, the voting process will have changed forever. AI such as that provided by Parascript will become more commonplace in the never-ending battle to keep our elections safe and secure from fraud. If you voted by mail this year in a district using Parascript ASV software, then you can rest easier knowing that your ballot has received the best attention possible to ensure that your vote is counted.
About the Author
Dan Lucarini is an AIIM Fellow with over 25 years’ experience in capture and content management technologies. He voted absentee from the UK this year by email and will check the North Carolina online voter portal to ensure his vote was counted. Learn more at www.linkedin.com/in/danlucarini.