Specifically, high-volume signature verification is a farce.
=============
Yesterday, we walked thru my linked experiences: getting a Maryland driver’s license, registering to vote, requesting an absentee ballot and casting an absentee ballot.
The key takeaways:
- As part of the D/L process, the state captured an electronic facsimile of my signature (that I etch-a-sketched on an electronic keypad)
- Also as part of the D/L process, I registered to vote … and, my electronic facsimile signature was posted to my voting registration.
- Later, I requested an absentee ballot online. When I did, I provided basic ID info but no signature — just a checked box indicating that I agreed to an electronic non-signing
- When I sent in my absentee ballot, I manually signed the over-wrapping secrecy envelope — not the ballot.
Now, the million dollar question: Given the above takeaways, how did my ballot get verified?
=============
Disclaimer: While I have some experience with signature capture & verification … and have researched the topic … much of this analysis is based on reasonable conjecture, not verified facts and testimony.
My absentee ballot was enclosed in a secrecy envelope which had a window exposing a unique bar code that identified me for process tracking.
So when my ballot packet was received, the visible bar code was scanned and I got an email saying that my ballot was received.
Then, I imagine, the processor opened the envelope flap to reveal my manual signature.
So far, so good.
=============
Now for the tricky part … verifying that I was really the person who sent in the ballot.
Note that the pivotal piece of evidence (i.e. the only piece) to work with is my signature.
Let’s imagine a best case: a well-trained, conscientious, unbiased processor searches the digital voter registration files and gets my electronic facsimile signature displayed on a computer screen.
The processor then eyeballs the 2 signatures — one is my”real”, manual signature and one is an electronic facsimile (the one I etched in at the DMV) — and decides whether or not they match.
THINK ABOUN THAT FOR A MOMENT !
That’s a best case …
So, how much confidence do you have in that process?
Color me skeptical.
Since I’ve watched practically every episode of Forensic Files (some multiple times), I know that signature verification is as much an art, as it is a science … it requires an expert eye and meticulous, time-consuming inspection.
Signature verification is not an activity well-executed by amateurs in a high-volume, time-constrained, politically-charged processing environment
Note: Some questionable cases can be quickly and easily flagged for exception processing — e.g. the signature is missing on the ballot envelope, there’s a gross name mismatch (Joe Smith instead of Digger Brown), or the signature syntax is wrong (e.g. John Doe or John Quincy Doe instead of what’s on file: John Q. Doe)
So, the error rate can be sky-high.
How high?
A Las Vegas reporter conducted a test and found an 89% failure rate identifying mismatched signatures.
For details of the test, see the source article Evidence that signature verification is a flawed security measure.
That’s not surprising … it’s well known and common-sensical that there are problematic differences between manual and electronic facsimile signatures (which are used for ballot verification). People write differently pen-on-paper than they do stylus-on-keypad.
Given the expected level of manual-to-electronic variations, it’s easy to imagine processors either (1) disqualifying practically all ballots, or (2) lowering their guard and ok’ing practically all ballots that come through.
Note: Some states & counties use signature matching software. Keep in mind a that — unlike a fingerprint — a person’s signature is subject to natural variance (e.g. depending on the signing media) … and, that a particularly challenging comparison is a scanned picture of a manual signature done on paper to an electronically etched facsimile signature. Given the wide “natural” variations, matching software tends to have a low correlation threshold — say, passing a signature as a valid match if it’s 40% like the signature on file. Accordingly, no professional election integrity body has certified signature matching software for use. Of course, that hasn’t stopped states from using it (e.g. Nevada). Source
=============
Widening the “tolerance levels” and ok’ing practically all ballots — is a particularly thorny problem, since there’s no way to retroactively reconstruct the ballot to signature linkage.
Important: Keep in mind that mail-in ballots aren’t signed … it’s the over-wrap secrecy envelope that’s signed.
Once the signature is deemed “verified”, the signature-bearing envelope is set aside or discarded … and, even if it isn’t tossed, there’s no practical way to reunite it with its ballot.
So, even if a signature is later found to be invalid, there’s no way to ID and toss the suspect ballot.
============
Bottom line: The only way to verify a ballot is via signature verification … which is, at best problematic … and, at worse, a complete sham.
Heaven help us.
November 23, 2020 at 8:12 am |
Prof. Homa,
I appreciate you highlighting the concerns and loopholes in the system. I have no evidence to believe that this happened.
I can argue similarly with data, likelihood of identity theft is much higher ( close to 100%) and it does happen, that doesn’t mean it happens everyday on mass scale.
November 25, 2020 at 11:40 am |
This is very very informational. Thank you, Pro Homa!
Since it’s hard to image that fingerprints or other biometrics will be used for verification mail-ins in US, either people should be requested to submit 10 signatures on file (5 manual 5 electronical) so that a software can help with better pattern recognition; Or, just get rid of electoral collage, so that small ballots in a certain place won’t have an outsized effect on the results.
December 7, 2020 at 10:00 am |
[…] Part 2: Mail-in ballot verification is a sham! […]