I don't think any scheme that requires people to be willing to commit felony fraud en masse is very likely. They would need to open other people's mail, forge their signature (on my ballot it clearly states that you will go to jail for this) and send it back in. They would need to risk all of that for one extra vote since they are only accidentally receiving an extra ballot.

We also don't need to sit around speculating about this because the data we have on voter fraud already tells us that it's incredibly rare.

What we should focus on is the 1 in 10 voters here who had a ballot rejected for primarily UX reasons that could be fixed. That's a problem which is supported by facts and could be fixed. Instead the current administration is gutting the USPS and making voting even more difficult in the middle of a pandemic. My ballot application is already taking twice as long to get in as it did last year. How is that OK?

You don't know how common voter fraud is. When it's easy to do and you can't be caught, as in other people's ballots coming to your mailbox, many will do it. They do it. I know someone who voted on behalf of another in California because ID is not necessary. There must be many more others doing the same.

If that happened regularly and widely it would be quite obvious because 40-50% of eligible voters vote. So there could be as much as a 40% chance that the person whose ballot you forged also tried to vote and is either turned away or submits a provisional ballot or complaint.

This would create a record that could then be used to examine the signature on the ballot and charge the person with voter fraud.

Note: not many states have changed their laws on vote by mail this year, so theoretical VBM fraud could have been done by submitting fraudulent ballot applications in most past elections too. The main difference this year is more voters are expected to use the no-excuse absentee ballot that was already available to them in other years.

This being a hacker forum, would such security measures as you believe in and propose be acceptable in any computing application? In finance? Medicine? Anything really?

This is how the system has worked, oftentimes for decades past.

No, they would not be considered acceptable in a computing application. But in computing, it’s considered acceptable to discriminate against users for the sake of security.

I’m unaware of any major democracy that uses computerized voting with a username, password, and 2FA to replace the traditional ballot box, for obvious UX reasons. (Never mind that computerized results might actually be easier to forge than paper ballots).

Regardless of what is “acceptable” it is a fact that widespread voter fraud using the mechanisms proposed would be detectable.

That's what the ID cards are used for, and have been for decades... Including for voting all over the world.

That’s never how voting has worked in the USA though. The entire idea of even having an ID card is relatively modern.

We’ve had voting for centuries, and done it without ID requirements.

Not sure how ID requirements would work in the several states that have done all-mail-in ballots for decades successfully.

Fraction of an percentage is all that is needed much of the time.

A few hindered ballots from a [left/right] leaning district going missing can be enough.

Don’t even need to have individual fraud, just some stats on how likely one area is to vote then “wrong” way.

Hell nursing homes or others could easily swing lots of votes by a single staff.

I’ve dealt with plenty of high officials that have negative documents go missing. It’s just routine for many.

If the person living there is so inclined and successfully forges the signatures on the ballots.

At least in WA state, a signature is required on the exterior of the envelope, and it is validated against the signature on the voter registration form.

Signatures are a lousy form of validation for voting in my opinion:

I'm a Washington state resident and because I sign a lot of checks I have two forms of my signature: My 'fancy' signature that has every letter of my name and my check signing signature that is a scrawl of simple squiggles.

I used my fancy signature for voter registration and accidentally used my squiggle signature for my fist mail in ballot and did't get any kickback. I've used it for six years and King County elections still has never questioned it.

As a counterpoint, I also live in Washington.

About five years ago I decided to change my signature. The first ballot I mailed in after changing my signature was rejected for having a non-matching signature, and I had to prove my identity and update my signature on file with the King County elections in order for my ballot to be counted.

My signature ended up changing over time and I had to update my signature with the county elections office in Washington to stop my ballots from being rejected. So really not sure how you're slipping by.

In Ohio its signature and ID number on your driver's license or last 4 of your social security number.

So ballots go to the wrong place, fine. The probability of random person there being interested and then finding any of this information is low.

Election officials could assign "strict" signature validators to certain targeted districts, so approx 10% of ballots get thrown out, and "loose" validators elsewhere, so only 2% of votes get tossed.

That would be illegal. See the “Bush v Gore” decision which was about a similar idea but applied to counting hanging chads, etc. rather than signatures.

Of course it is illegal. The discussion is about how anti-fraud measures (like signature verification) could potentially be flipped to actually assist in cheating/fraud.

Let's keep in mind, depending on where power lies post-election, that doing something illegal, even if caught, doesn't necessarily means one would be punished for it.

Sounds great! I'm sure nobody would take advantage of that.

Sounds like a violation of equal protection.

Do you think they have handwriting experts validating every vote?

And even if they did...most of the time their opinion would be "yeah, looks OK". You'd like that to just be stage 1 and call for further scrutiny if it turns out it would affect the result - but you can't, if you keep the evidence longer to revisit it during a recount, you lose the secrecy of the ballot.

You can do it the other way around. If the signature seems fishy, you set it aside unopened, and only do further validation before opening it if the election is close enough for it to matter.

For example, in an election with 100 votes cast. If 10 signatures are suspicious, you only do extended validation and open them if the election was won by a margin smaller than 10.

I’m confused.. do you think it’s likely that someone forges the signature of a person they’ve never met within a believable margin of error? Seems the burden of proof is on the people claiming that happens a lot

A few years back me and the wife went on a cruise that disembarked in NOLA. We had a lovely time and at one point went to buy some new luggage cases: one of ours had fallen to bits and the other had suffered somewhat (my fault.) We took the tram out of town to a mall on the outskirts and hit a decent sized department store.

I was offered a tablet to sign on when I offered up my card. So I took back my card and signed it on the back and then loosely repeated that same signature on the tablet thing.

I'd had that card for around two years. I never sign them these days because it doesn't really mean anything around here any more in the UK. Bear in mind I am old enough to remember signing cheques for "To cash".

How much validation is actually done in practice? In the UK, I sign something about once at year _at most_ - I can't remember the last time I signed something without going back a few years, and I've no idea how I signed it.

But how could somebody actually KNOW the signatures of the previous tenants? In order to try forging them you need first some models...

How is the person living there going to figure out how to correctly forge the signatures of all those previous tenants?

There is no such thing as large-scale signature verification.

If a close match were demanded, people would complain of disenfranchisement. People don't sign the exact same way every time.

Realistically, the workers are tired and they don't care. If they do reject anybody, it's probably because the vote comes from an area that tends to vote in the undesired way. In other words, this human element is a problem. It would be especially easy to bias the election by rejecting based on name.

An imperfect signature of the desired ethnicity (more likely to be a "correct" vote) gets a pass. An imperfect signature of the undesired ethnicity is a fail. That's all it takes to quietly flip an election, with no evidence trail at all.

There doesn't need to be. I feel like we're failing to recognize that this is a soft solution, not a tech solution. The signature is there so that you go to jail for felony fraud if you forge it. Then with simple detection schemes the cost of trying compared to the benefit of like one extra vote doesn't make sense.

That argument is weak, but not insane, for in-person voting.

For vote-by-mail though, how could anybody go to jail? The ballot shows up. It is accepted or rejected. If rejected, then what? Are we going to pick fingerprint DNA out of the paper fibers for this? There is no reasonable way to ever determine who attempted to cast the vote.

It'll be fraud, but never part of the official statistics. It can't be proven without using the level of resources normally reserved for famous murder cases.

There just isn't a deterrent. Severe punishment means nothing if the criminal thinks he has no chance of being caught. The fraud can be done in bulk, and it is, via ballot harvesting.

Also, what about the corrupt verifiers? Being prone to rejecting some names more than others can change an election, and there would be no way to prove it. If the signature verifier has a preference between Cohen and Abdul, or between Garcia and Washington, what can anybody do?

What’s different about in-person voting? You show up, you vote, you leave. Are they going to take fingerprints and DNA samples of the polling center?

In theory, a bad signature at in-person voting could be cause for arrest, and we could actually do it.

With the mail-in voting, the person isn't present. You can't possibly find the person. No matter how obvious the fraud, there is nothing to be done about it. The person is certain to avoid arrest. No charges will ever be filed, and some people will then conclude that fraud doesn't happen.

In the one case, the criminal faces a small risk that is easy to dismiss. In the other case, there is no risk at all.

There's no police officer hanging out at the polling location to arrest someone for signing their name incorrectly on the voter roll. At least in California, signatures on the rolls are only verified after the fact. At least with mail in ballots you have the recourse of not counting the vote if the signature does not match. For the in-person roll you can't even do that!

Not meant to sway the argument either way, simply an anecdote - in Pennsylvania when you vote you sign your name next to the signature on file. I've seen people be given a hard time when the signatures are wildly different.

Seems like you could just copy the signature from the box next door then?

Signature validation is not enforced 100%. Go asked those who worked with it in the pasts elections informally. You will know how bad it is.