Georgia’s 127k Missing Votes, Disproportionate to African American Neighborhoods
coaltionforgoodgovernance.sharefile.com
coaltionforgoodgovernance.sharefile.com
- the missing votes are from the Lt. Governors race only, not the total ballot
- many people reported software glitches with that specific ballot item
- the study blames faulty programming and insists the state not use electronic voting since it apparently is not reliable
Their conclusions are strongly worded but sound reasonable to me.
Quoted from the executive summary:
• Forensic examinations of the machine programming must be promptly undertaken to obtain answers, locate the source of the errors, and, if appropriate, hold officials accountable.
• Electronic voting systems must be immediately abandoned and paper ballots adopted so that no future elections are conducted on Georgia’s unauditable machines.
• Governor Kemp, Secretary Raffensperger, and legislative leaders must abandon their plan to adopt a new ballot-marking-device electronic voting system that, like the current system, is unauditable and vulnerable to problems of the type experienced in November’s election.
Electronic ballots really shouldn't be a very difficult engineering challenge. We're already doing all of our financial transactions electronically and mistakes are relatively rare with an absolutely massive volume of transactions.
Of course it must be more complicated than that, but even the basic user interface always feels wrong. Just let me push a button that has the candidates name on it or directly above it.
Oh well.
I checked and your county still offers touchscreen voting devices. Probably 1 per poll site, and at the early voting center, to comply with HAVA.
A claim that the problem with electronic voting is bad UI could imply that fixing the UI would fix electronic voting.
With sufficient effort, a 100% reliable paper count can be produced. Of course in many cases it may not get that effort, but it's possible, and this attribute is particularly important in an audit.
The thing is, mitigating this for a physical system like paper ballots, requires fairly basic training to become a poll monitor. Whereas the auditing of electronic systems is substantially more complicated, not least of which when for-profit companies compel the county/state into NDA's baring 3rd party audits, which happens to be the case in Georgia.
We witnessed people transporting votes in their personal vehicles in Broward county so there's no real chain of custody along with no verification of identity.
If that story is true, that's a problem with Broward County, not with paper ballots.
Voting (election administration) is not double entry accounting.
What you want would require every single person being required to cast a ballot and every ballot be counted.
It'd also eliminate the possibility of the secret ballot.
Are you sure that's what you want?
When I moved to California I was really impressed by the way voting is handled here. Specifically two things:
(1) I was able to vote one block away at a neighbors garage. I don't know exactly how this works, but apparently people can volunteer to host a polling place, go through a training program to get certified to do it, and then the state will give them the equipment to do it. These polling places are just about everywhere, and it made voting exceptionally quick and easy.
The neighbors who ran it were also very courteous and professional. While there, I saw them handle several questions and problems that people had, and they did an excellent job. I was truly impressed.
(2) The ballot system was: you mark a paper ballot with a pen, and then you feed that into a scanner which electronically tallies your vote. It also keeps the paper ballot around as a record. So it would seem to offer the practical efficiency of an electronic system with the auditability of a paper system. I liked it.
Having previously voted in three other states, this was the best experience by far. I have my own complaints about California, but I'd love to see this way of handling elections become more widespread.
I’m glad you had a good experience, but I’m very skeptical.
Edit: it seems lots of people can disagree but no one can make an argument this is safer or more reliable in practice.
From my experiences in other states, I would say it’s very possible to systematically lower voter turnout by making the process laborious, in part by having only a small number of polling places that you need to drive across town to access, and then by having long lines to vote (because so many people have to use the same polling place).
Given a choice between the two I personally would favor the decentralized model.
https://www.miamiherald.com/news/politics-government/electio...
I think in a previous topic on HN, someone described his experience in one of the states way better. I'm describing what I remember, which may not be completely accurate:
You take the paper ballots, and you split them into lots. Each lot is passed through at-least two voting counter machines, and a recount is triggered if either machine produces a different result. Random lots are manually hand counted, to ensure they produce the same results as the voting machines.
To me, this makes the electronic counting extremely tamper resistant, as tampering with a subset of machines will very likely produce a discrepancy against other machines that is detectable. Or if the entire fleet of machines has been tampered with, this becomes apparent with the manual counts.
As for the machine at the poll, the initial scan can just be a best practice, that before the ballot is submitted can indicate to the voter if the ballot scans OK, leaving an opportunity to correct.
Of course, this is just on the counting side, all sorts of additional best practices are needed to ensure the paper ballots are transported securely, aren't lost or tampered with, are stored securely if a recount is needed, etc. And each step has scrutineers, even if the polling station is in someone's garage.
Just count the damn votes by hand.
At the end of the election period, poll workers would secure the scanned paper ballots and the magnetic cartridges, and mail one set of mag. carts. to the state elections officer, and carry the paper ballots and the other set of mag carts to the local (city or county) elections office.
The second set of mag carts provide the “election night” results. A local protocol & set of procedures cross-check the carts and the number of paper ballots and the number of voters who appeared to vote at each location/precinct. The first set of mag carts provide a cross-check and a backup copy. (A small number of exhausted elections officials will incorrectly put both sets of carts in a mail box, and hilarity will ensue.)
Election operations in the US have a long and complicated history. Reinventing elections as a greenfield _de novo_ exercise without considering most of the complexities that have grown up around them is appealing, like most big rewrites. Poorly-implemented electronic voting should be viewed with honest suspicion, IMHO.
> minor candidates' vote shares almost double when their names are adjacent to the names of major candidates
The nice things about the mark & scan systems are that the voter gets immediate notification of an unscannable ballot: stray marks, mis-marks, overvotes, and similar issues are correctable in real time by the voter, perhaps by exchanging the spoiled ballot for a new one, and destroying the spoiled ballot. The scanned paper ballots are also available to be rescanned for a recount or audit.
Creating a voting system for a many-race election that is intrinsically resistant to manipulation is harder that it first appears. I used to (circa 2000) think that electronic voting would supersede paper ballots in a few years. Now I think that paper ballots have a simplicity and durability that outweighs the apparent convenience of electronic approaches.
source: elections official in two states, voter registrar in one.
Use a touchscreen, fill out a ballot, hit the print button, receive a paper printed ballot, inspect it for errors, drop it into a scanning system that scans and retains it. Importantly, the optical scan is done not by reading any sort of barcode printed on the paper but by reading the same English letters as the voter is inspecting.
All the convenience of electronic systems, hand recount is available and easy. Hand audit some of the machines after every election to ensure accuracy. Doesn't matter how messed up the touchscreens are, since the paper ballot can be inspected by the voter before being cast. No such thing as hanging chad or butterfly ballot problems. Zero bad ballots. Voters who skip races will be doing so intentionally. "LIEUTENANT GOVERNOR: NO VOTE". If they turn that in to be scanned, that's what they wanted.
Having a paper audit trail has become a fairly big cause, so I'd be surprised if the legislature recommends solutions that don't satisfy this.
Isn’t that somewhat contradictory to the rest of your comment? The mark & scan machines work well and have for decades, and not just in voting but in standardized testing in various contexts (school, drivers license tests, etc).
It seems the voting machine element of the equation “intrinsically resistant to manipulation” is pretty much solved, if election officials would just keep what’s worked well for decades rather than transitioning to shiny new digital toys with no paper trails and poor security.
In the worse cases it means dealing with voters who want to read the ballot aloud to their partner, or camp out in a voting booth for 30-40 minutes, or who are disruptive in some way. To be fair, the last general election here (Bay Area) had 21 races with 43 total significant candidates, plus 12 referendums/measures, for 67 items to count per ballot - this is not a good "count by hand" situation.
In the long run what people really want and need is an election system (including voting machines, ballots, voter rolls, polling places, staff) that gives them a sense that their votes count.
In general the voting system can guard against false positives or false negatives. By that it can either bar legitimate voters or allow illegitimate voters. Since the USA has a historic tendency of barring the poor, minorities and women from voting; elections officials should take that into account when considering trade offs.
From an election integrity standpoint, the modern implementations of vote by mail is little different from electronic voting.
--
The jurisdictions that I'm aware of scan AND tabulate ballots as they are received. This permits peeking at early results. Citizens expect that votes are not tabulated until polls close on election day. Administrators argue that the pre-scanning and tabulating isn't really a count and that only the final report is a count.
--
Until very recently, vote by mail lead to a novelty participation bump followed by a long decline. Explained by losing the culture of voting. Postage is now prepaid in some jurisdictions, which may have lead to a +4% boost in participation, matching prior poll site participation rates. But it's too early to separate the prepaid postage boost from the overall boost in 2018. Time will time.
--
The transition to vote by mail is driven solely by appropriations (pork) and administrator's desire for centralization.
With vote by mail comes new business models. Whereas with poll sites we'd only pay per ballot to accommodate projected turnout plus 10%, we now pay for every voter every election. Vote by mail ballot packets are ~$2 whereas poll ballots cost ~10c. Additionally, all new tasks like signature verification and ballot tracking are new opportunities for rent seeking. Again, per voter per election, vs time & materials.
There's all new gear to buy, of course. One could argue this is no different than any other IT lifecycle.
--
Another way vote by mail is like electronic voting is the complete loss of voter privacy (the secret ballot). Elections are administered per precinct. Received ballots are binned. So very likely that your ballot will be the only one from your precinct in the bin. To protect the secret ballot, ballots must be sorted into precincts before processing (opening). This adds considerable effort and expense to the entire process (logistic nightmare). Last time I checked, my jurisdiction still was not doing the legally required precinct presort.
--
Because voters are no longer able to fix their own errors (like with the mark sense poll based systems mentioned elsewhere), administrators have to adjudication voter intent. Now that ballots are optically scanned as they are received, like a fax machine vs the mark sense systems, voter intent is adjudicated electronically, meaning there's no paper trail. Makes the manual recounts and audits kind of tricky, if not completely suspect.
You state many other things in your comment that are incorrect. For example, if I make an error on my mail-in ballot, I get a new one... just like the instructions say. That's never happened to me, because it's pretty straightforward to not spoil my ballot.
And if the ballot is marked oddly, there is a paper trail, it's... the original ballot. As you're probably fully aware, some recounts look only at ballots that show overvoting or otherwise are judged less confident by the counting system, and more precise recounts look at all ballots. It's not that tricky, and it's certainly not completely suspect.
My county uses the same system for in-person and mail-in votes.
I believe ballots in CA must be postmarked by election day to be counted. Whereas OR's rule is received by election day. So the Santa Clara number will be updated as more ballots arrive. But there is a count on election day.
FWIW, IIRC, my jurisdiction's elections are certified, the final count with all the bookkeeping, two weeks after election day. Santa Clara will do something similar.
"My county uses the same system for in-person and mail-in votes."
That can't possibly be true. The ballot may look the same. But the handling, administration, bookkeeping, and tabulation is completely different.
For a long while, jurisdictions did try to use the same mark sense devices at both poll sites and central count. But they way too slow to be practical for mail ballot processing. I'd be very surprised indeed if a county your size didn't use the newer high speed imaging scanners.
"You state many other things in your comment that are incorrect."
Each jurisdiction has it's own chaotic mutant rules. And it keeps changing. So no one person can know everything about all election administration in the USA.
But your first "refutation" wasn't. Your last wasn't even wrong.
And I don't have the heart to correct the rest.
--
Update: I was curious what gear Santa Clara is using, when I spotted this:
"Returned Vote by Mail Ballot Processing
In accordance with California Elections Code, the ROV will begin to open, process and count returned Vote by Mail Ballot Envelopes approximately 11 days before Election Day. Vote by Mail processing will continue after Election Day until all ballots received prior to 8:00 pm on Election Day are processed."
https://www.sccgov.org/sites/rov/Voting/observers/Pages/defa...
So I was wrong about the postmarks.
I strongly encourage you to observe your elections. And attend the certification (canvas report) meetings. Seeing how the sausage is made will illuminate you.
Santa Clara County's publicly available information is actually quite good. I spent years clawing information out of my county and didn't get anything nearly as detailed: https://www.sccgov.org/sites/rov/Voting/VotingNovember2016/D...
Oops. Santa Clara is still doing "ballot enhancement" and "ballot duplication". So now I believe it's likely they're still using the mark sense (opscan) devices for mail ballots. My county switched to image scanners over a decade ago, so I assumed everyone would have switched by now. My bad.
But wait, then why does it take so long to get the final mail ballot tallies? I'd bet it's because "duplication" and "enhancement" are manual tasks, and they're manually feeding the mail ballots into the mark sense devices, which takes days.
--
For the record, per your own county's documentation, the only thing I was wrong about is the lack of a paper trail (enhancement, duplication). Because your county hasn't upgraded to the ballot image scanners. Yet.
Why do you keep on referring to "mail ballots"? We use the same ballots for mail and in-person, as I clearly said before, you know, in the part that you said "That can't possibly be true" about?
> And I don't have the heart to correct the rest.
If your goal is to have a polite discussion, you're not getting there.
Mail ballot processing is different from casting ballots at poll sites.
What should I call a ballot that's cast via the mail? "Postal ballot"?
"If your goal is to have a polite discussion, you're not getting there."
And you could have researched how your county's election administration works before responding.
I didn't say that mail ballot processing was the same.
I said: "We use the same ballots for mail and in-person"
The same machines are used to count them.
So your comment "So now I believe it's likely they're still using the mark sense (opscan) devices for mail ballots." is odd because we use the same hardware to count in-person ballots that we use to count mail ballots, because they're the same physical ballots.
--
How a cast ballot's votes are counted is just one part of ballot processing.
To citizen's, marks on paper seems like a pretty straight forward thing. Like everyone else, I just marked my ballot and dropped it in the mail. Voila! How hard could it be?
To the election admins, poll and mail ballots are very different things.
Closest analog I can think of is "autonym".
Citizens think about the "what" and the admins think about the "how".
Poll ballots and mail ballots will never be comingled.
Poll ballots will never be "enhanced" or "duplicated".
Central count opscan machines are programmed differently. So they generate different reports. Which must be collated differently.
Signature verification is done differently.
There are many more failure modes with mail ballots. IIRC, my jurisdiction records 30+ different kinds of errors.
Mail ballot processing is a sausage factory. You gotta see to believe it.
But why am I telling you all this? I've had this same convo with 100s (1000s?) of people over the last 15 years. And I'm the conspiracy nut for daring to explain to people how their elections work.
That's almost entirely theoretical, we hope. But I'm sure there are parents or adult children or spouses or whatever that force results that they want. The North Carolina election where people were picking up ballots and marking them doesn't seem to have been repeated in Oregon, as far as we know.
I tend to be a last-minute voter and I see other people who are scurrying into the library to deposit their ballot. Or the long line of cars at the voting office.
One thing about Multnomah County, not sure of other counties in Oregon, is that you're notified (by email, maybe text is also an option) when ballots go out, when they've received your ballot back, and when they've accepted your ballot.
a) eligible to vote b) voted only once c) it is the person that voted and not his/her proxy.
It would be much better to use some biometric, verified by 3 independently developed databases. As those would be hard hard to fake
They make access easier(seniors), errors are correctable, reduce physical human errors(mismark/badmark), and provide a paper and electronic ballot trail.
The race was decided by a margin of about 55,000 votes in Kemp's favor.
Kemp, as Secretary of State, was in charge of Georgia state elections, which presented a conflict of interest for him as he sought higher office. Kemp implemented a voter suppression policy called "exact match," which helped to severely complicate the voting process for about 53,000 Georgians, 80% of whom were people of color.
More info: https://www.washingtonpost.com/opinions/georgias-voter-suppr...
Also, voter fraud basically does not exist in the United States as per this article from the Brennan Center: https://www.brennancenter.org/analysis/debunking-voter-fraud...
In fact, the most notable example of election fraud in the United States in recent memory is in NC's 9th Congressional District, where it looks like the Republican in the race cheated his way to victory: https://en.wikipedia.org/wiki/2018_North_Carolina%27s_9th_co...
edit: as burlesona rightly points out below, the article to which this comment is attached is about the Lt. Gov.'s race. My comment is about the general lack of integrity in Georgia's elections, and the deeply undemocratic (small-D) results that emerge from this state of affairs.
The article says many people reported software glitches where the Lt. Governor ballot item did not display or work correctly, which matches their findings that there were a lot less vost cast than expected for that specific race.
Having said that, the explanation by OP does add significantly to the general case of the integrity of the whole process in Georgia being suspect and need review to ensure the rights of all voters are guaranteed. Otherwise, the process is more obviously a farce than even before and will sow discontent even further.
An explainer: https://www.vox.com/policy-and-politics/2018/12/11/18134636/...
Also, legally, neither one is treason.
Haha ok man but no for real that's actual treason.
I think you need to look up the legal definition of treason as specified by Article 3, Section 3 of the US Constitution.
Have you read it? If you don't think overthrowing our democracy is an act of war against the USA or giving our enemies aid and confort....
https://ih0.redbubble.net/image.16046137.2447/ap,550x550,16x...
Note that the Republican obsession with voter fraud is actually an attempt to justify election fraud.
CNN’s Jim Acosta tweeted “voter fraud in this country is actually very rare.”
Considered to be a justification for to continue voter fraud (because getting ineligible people to vote (and multiple times) ) -- is likely helping democrats).
[1] https://www.foxnews.com/opinion/voter-fraud-exists-even-thou...
"Securing" our elections is about suppressing the vote of the poor, not about stopping fraudulent voting.
So fraud doesn’t exist but then fraud does exist? Which is it?
The best option to me seems to be scanned paper ballots, which offers the ability to perform analog recounts when there is suspicion of problems with the digital count.
(Disclaimer: only a distant observer of this, there may be better options.)
Or does it mean it's so logistically unpalatable that you can bet on it not happening? Same thing with recount as a remedy to detected anamoly; it's theoretically possible. But is it so unpalatable that a gambler may gamble well on an event?
The fundamental difference between computer systems and paper systems is that the effort required to corrupt a computer system is mostly unrelated to the size of the corruption. Paper systems require a lot more effort for a large effect.
What do you do when you detect anomaly? Have an unpalatably expensive recount? That can be expensive enough to use as a reliable weapon. How do you know if the final count is complete? Do you track down every voter? These things sound expensive enough to simply not happen.
Instead one side detects enough anomaly for a court to intervene, and the court considers blunt options or leaving the results as-is. For things like Bush v Gore the court is under enormous pressure to act quickly and perfectly, and they voted along party lines.
In my jurisdiction, trust is built on mutual distrust. Every receipt, seal, report, lock had to be handled and signed off by members of opposing parties.
Sure, it's possible to corrupt poll site based elections. But you have to corrupt many sites. You'd also have to corrupt upstream and downstream actors as well. Super unlikely to go unnoticed.
Typically, there are ballot accountability reports. eg 100 ballots printed for election day. 50 ballots issued, 49 ballots cast, 1 ballot spoiled, 50 unused ballots. 100 ballots returned to election HQ (central count).
Adding or removing ballots from poll site based elections would be noticed.
I can't speak to (all) other jurisdictions, so YMMV.
No amount (or lack thereof) of technology matters if you get wrongly purged off the voter rolls.
Incorrectly purging people is way worse than incorrectly not purging people.
In some states, the standard for purging has been intentionally made overly strict - "Hyphenated-Name" doesn't match the "Hyphenated Name" the state has on file, so they get purged. It helps when the Secretary of State making the rules is also running for Governor:
https://www.politifact.com/georgia/article/2018/oct/19/georg...
> Minority voters are more likely statistically to have names with hyphens or suffixes or other punctuation that can make it more difficult to match their name in databases, experts noted. That makes them more likely to get caught up in the "exact match" law.
If you register to vote in Georgia, then move to Ohio and register there, you might have the right to vote in Ohio. You don't have the right to vote in Georgia, nor does anybody else have the right to vote in Georgia with your identity.
If you register with "Hyphenated-Name" after having previously registered with "Hyphenated Name", that doesn't mean you get to vote twice! You don't have that right.
Mistakes are a big deal. An improper vote is a denial of the rights of the legitimate voters to have their votes properly counted.
Fixing all the issues is difficult, but there is one simple improvement: have the federal government, perhaps the social security administration, publish a mapping from human to voting district.
> People don't have the right to vote, at least not unconditionally.
None of our rights are unconditional. You can give up free speech rights by joining the military, and we've got things like noise ordinances. You can lose your right to bear arms by being a felon. Et cetera, et cetera.
> If you register to vote in Georgia, then move to Ohio and register there, you might have the right to vote in Ohio. You don't have the right to vote in Georgia, nor does anybody else have the right to vote in Georgia with your identity.
No one has argued otherwise.
> If you register with "Hyphenated-Name" after having previously registered with "Hyphenated Name", that doesn't mean you get to vote twice! You don't have that right.
That's not what's being discussed, at all.
People register to vote as "Hyphenated-Name". The state checks their databases and doesn't find a matching name because they got their driver's license years ago as "Hyphenated Name" (or because the DMV doesn't support a hyphen, or any number of other reasons). They get purged off the rolls.
> Mistakes are a big deal. An improper vote is a denial of the rights of the legitimate voters to have their votes properly counted.
Most people improperly on the rolls aren't improperly voting. If they moved to another state, they're almost certainly not coming back to double their vote. If they died, the chances of someone else pretending to be them is vanishingly small given the severe repercussions of getting caught (which has to be balanced against the small chance of a single vote making a big difference).
The state checks their databases and doesn't find a matching name because they got their driver's license years ago
Having a driver's license isn't even a prerequisite to vote, nor does it confer eligibility to vote.> Under a 2017 Georgia law, a voter registration application is complete if information on that form exactly matches records kept by Georgia’s Department of Driver Services or the Social Security Administration.
Driver's license isn't required to vote, but if you have one, your voter registration form has to exactly match it. If you don't have one, you've got the same issue with the Social Security Administration with hyphenated versus non-hyphenated, misspellings, character limits (NYS DMV knows me as "C, M" for example), etc.
Driver's license isn't required to vote
... which is exactly what I wrote.I said that if they have one, the State of Georgia checks that your voter registration matches your name exactly on your driver's license, and rejects your registration if it doesn't.
Seems like a good kind of work to spread, thanks for sharing.
Just spitballing here, because it seems like an easy problem to solve, the US is just notoriously stubborn when it comes to mixing up the election process.
Votes must not be attributable to any given person, or else the potential for coercion exists.
Many developed democracies either never stopped using a pure paper ballot, or have gone back to it over the last decade or so. For all its flaws, it's easier to make safer than electronic voting.
That’s public info and should be audited.
The part that remains secret is who/what I voted for.
Edit:clarity
That doesn't cover every failure mode, but it's pretty good compared to nothing.
(And do you really think calling any other answer dishonest is going to help the discussion?)
Shouldn't there be a convention of writing Georgia (US) or (state)?
1. the voting machines used were known well in advance to be badly vulnerable [0]:
He said hackers could infect election computers by first gaining access to a state employee’s computer, possibly by tricking him or her into clicking on a dangerous link in an email. Once the malware is on one machine, it could reach central election systems through internal networks, USB devices or memory cards.
J. Alex Halderman prepares to demonstrate Monday at Georgia Tech how easy it is to hack Georgia’s electronic voting machines. In a hypothetical election, Halderman changed a 2-2 vote between George Washington and Benedict Arnold to make Arnold the winner by 3-1.
Election computers could also be subverted in person, by someone like a janitor or a temporary worker, Halderman said. Individual voting machines could be tampered with if someone unlocked the latch that protects the memory card port.
2. He reported security researchers (hired by state Democrats) to the FBI for trying to 'hack the system' [1], whereas...
In 2015, Kemp’s office inadvertently released the Social Security numbers and other identifying information of millions of Georgia voters. His office blamed a clerical error.
[...]His office made headlines again last year after security experts disclosed a gaping security hole that wasn’t fixed until six months after it was first reported to election authorities. Personal data was again exposed for Georgia voters — 6.7 million at the time — as were passwords used by county officials to access files.
<eyeroll/>
3. Georgia implemented a use-it-or-lose-it policy for voting. People found out that because they didn't vote in the past that they had been purged from the register of voters. That's insane to me, but apparently the Supreme Court determined it was constitutional. Georgia also implemented extremely strict name-matching (i.e. make sure that you did not use a hyphen when writing your name one place and a space when writing it in another, because then they won't match). [2]
But voting rights advocates fear that “use it or lose it” purges could be used as a voter suppression tactic — along with voter ID requirements, gerrymandering, polling place changes or closures, and registration obstacles — that often help conservative candidates, because infrequent voters tend to be younger, poorer and people of color who are more likely to favor Democrats. For instance, the APM Reports investigation found that such purges in Ohio disproportionately affected urban, Democratic-leaning counties.
[0] https://www.ajc.com/news/state--regional-govt--politics/how-...
[1] https://www.pbs.org/newshour/politics/a-look-at-the-election...
[2] https://www.wabe.org/georgia-purged-about-107000-people-from...