Hacker News Reader

Comment by spanhandler | Hacker News Reader

Looks like

1) about 1/2 of them are due to plain ol' User Error (not that there isn't room to improve the "UI" so that happens less) like not signing, not including required parts, simply leaving the ballot out of the envelope entirely(!), and so on,

2) 1/4 are "signature didn't match" (most of those are probably false negatives and should have counted, I'd bet, but given other issues I wouldn't be surprised if people in the same household accidentally signing one another's ballots or something else silly like that is some measurable part of the problem) and,

3) 1/4 are other :-/

[EDIT] looks like most of the "other" category is probably a couple towns in one county having a vote so screwed-up that 3,200 ballots were rejected en masse. That's about 20% of the rejected ballots, or about 2% of the total vote, right there.

View on HN

>2) 1/4 are "signature didn't match"

I've always felt that matching signatures was a terrible way of authenticating voters. The person doing the matching is not a trained handwriting expert. Also, like many people, I have subpar handwriting and I highly doubt my signature today looks anything like what it did when I first registered to vote 25+ years ago.

When I lived in King County Washington half the time my vote was sent back due to mismatching signature. Heard similar stories from others. Maybe it stops fraud, but the high false negative rate is egregious.

maybe it is fraud. who audits the people who compare it?

Plus we're talking about "signature experts" as if there's a solid science there. From what I understand, it's a bit of a cargo cult where you just package up "do they look similar" in the trimmings of hard science.

Plus, it has a way of working out that people with names like Lakisha or Jamal might find themselves stuggling to match their signatures more than people with names like Chris or Alice.

It works well enough because all of the real authentication happens after you register and before you ever get a ballot. So there isn't really much fraud risk from you actually sending in the ballot, the signature is just the cherry on top.

The on site signature books are set up for right handed use. The helper has to move their hand shielding the printed signature so lefties can just copy what they wrote :)

So I guess we can say that at most 5% of the ballots were rejected for a "good" reason; that is, for a reason that might be related to actual fraud. But in reality it's probably much lower, since for #2 it's pretty much a given that a lot of those are false positives (as you suggest), and presumably not all of the #3 cases are fraud.

Is that good? Bad? At the very least, if there are fraud attempts, it's good that they're being caught?

It's just lame that at least 5% of ballots, and probably more, are being rejected due to ballot UX issues or a broken verification process.

Right: false positive from the perspective of catching fraud, false negative from the perspective of accepting good ballots, is what I mean, exactly. Same thing, different way of phrasing it.

Anyone know what the rejection rate for in-person paper ballots is, in the US, typically? It's hard to judge how much worse this is at successful voting than in-person with paper. My gut tells me increased participation would dwarf even a hard 10% rejection rate, assuming no improvement could be made, and if the in-person rejection rate's not near-zero then it's really a big win, but maybe not.

Meanwhile stories like this are making me really nervous about the public perception of November's vote, regardless how well they're actually run.

> not that there isn't room to improve the "UI" so that happens less

Definitely this. I live in NJ and found the mail-in ballot instructions to be somewhat confusing: multiple layers of folding and envelopes, a perforated piece which you were explicitly NOT supposed to detach, an additional section that that only applied if you were mailing it on behalf of another person, etc.

Pretty sure I filled it out correctly, but I mean, I had to re-read the instructions three times and I'm a former FAANG engineer.

>2) 1/4 are "signature didn't match" (most of those are probably false negatives and should have counted, I'd bet, but given other issues I wouldn't be surprised if people in the same household accidentally signing one another's ballots or something else silly like that is some measurable part of the problem) and,

How does this work afterwards? Are the voters notified? Seems like bad design to silently drop votes just because some guy arbitrarily decided your signature "didn't match".

They don’t drop. You can view your vote online to see if it got rejected. Also if your registered to a party, the party will track you down to fix the issue in a close race.

> 1/4 are "signature didn't match"

If this was a factor in allowing someone to vote, I'd never be allowed. The only time i use cursive is the once or twice a year I have to sign something (new card, paperwork, etc) so my signature never looks the same as other signatures.

>simply leaving the ballot out of the envelope entirely(!)

Some / all of those could've been intentional no-votes, just like leaving your ballot blank would be in an in-person vote.

Reply on news.ycombinator.com