Complete In Duplicate And Forward The Original To The County Health: Fill & Download for Free

TL:DR: No and Sometimes.No. I’m not confident that American voting systems accurately reflect the vote of the people. The root cause of this is that it’s rare that the ballots cast by voters are counted by actual human beings.Let’s look at a few voting systems (in the sense of the mechanics of voting and counting):Electronic Voting Machines with no Paper Trail:This is the worst case. You make your choices on a computer, it presents a summary of your selections, you commit to the ballot. Sounds good so far, but how do you know that your vote was counted as it was cast? County clerks and Secretaries of State will claim that there is a rigorous process the ensures that no malware can possibly interfere with the counts, but the software is proprietary, not subject to third party review, and not checked after the election to insure that it matches what was checked before the election. Further, if a question arises, there is no way for a human (or a group of humans consisting of disinterested third parties and representatives of the candidates) to count because there is no permanent record of each “ballot”!Electronic Voting Machines with a Voter Verifiable Paper Trail (VVPAT):Not quite as bad as the first case, but this too has problems:Software determines how the VVPAT is printed and how the “ballot” is counted. How do we know these are the same? It would be easy to program the machine to print a VVPAT that is accurate, but to record the vote differently. This could be caught - if we looked for it! But it’s rare that the VVPATs are used to cross check the machine “counts”.Most voters don’t check the VVPAT! So much for the “VV”. From a hacker’s point of view, this is an opportunity to mess with things - especially down ticket races. Voters are more inclined to check the “top of the ticket” races than they are to check all races.The remainder of the systems separate the casting of ballots from the counting. All share one thing:Optical Scans machines that tally marked ballots:Someone (in most jurisdictions the voter themselves - at least for in precinct voting) feeds the ballot into the machine, which tallies the votes and deposits the ballots in a sealed ballot box.This too has its problems:Again, it is software based and rarely counter checked. So a hacker can get a program onto a machine that (for instance) rolls the dice with each ballot and, if the proper random number appears, doesn’t count ballots for candidates they oppose, considering it to be an undervote (no vote) or overvote (voting for more than one candidate in a single winner race).Poorly marked ballots, or ballots with stray marks (even dust that interferes with scanning!) don’t get counted.These machines rely on precisely preprinted ballot - so the markings are expected to appear in precise locations. If the ballots are printed poorly, they can be difficult to read - to the point of counting your vote for candidate A as if you voted for the adjacent candidate on the ballot!I’ll comment about saving “ballot images” in the “accuracy testing” section of my answer.Let’s now turn our attention to the ballot marking process. Where paper ballots are used, there are two ways in general use: Machine marking and hand marking.Machine Marked Ballots:Here the voter makes their choices on a machine and the machine prints a ballot which is then fed into the aforementioned optical scan devices. They come in two variants:Machines that print a ballot that appears as if it was hand marked by an extremely precise voter. Other than the precision, they appear as if they were marked by hand.Machines that print a ballot that has a human readable summary of the votes cast and a 2D barcode that supposedly encodes the same information.Both of these have problems.In case #1, they’re similar to the problems with electronic machines with a VVPAT - at least in terms of the tendency of voters to check only “top line” races.In case #2, how do you - the voter - know that what is encoded in the 2D barcode matches the human readable summary? Voting machine manufacturers also sell “scanners” that can read a single 2D barcode and “tell” you who you voted for, but these scanners are expensive so counties only buy one or two and keep them centrally - i.e. they’re not available to the voter in the precincts! Moreover, the handheld scanners aren’t what actually “count” the votes - that’s done by the optical scan machines. It’s separate machines with separate software optimized for different purposes. I had a county official admit to me that he couldn’t prove that the two “counts” would be identical!Hand Marked Ballots:Finally, we’ve come to what is, in my opinion, the only choice we can trust: ballots hand marked by the voter. There is no need for the voter to “check” what the machine has recorded from their screen input because there are not screens!If they are paired with better optical scan machines, the scan machines can also detect over and undervotes and offer the voter a chance to redo their ballot before it is placed in the sealed box.If the hand marked ballots are counted by actual people, with representatives of all the candidates participating able to be a part of the group that determines how each ballot is counted, done in full public view immediately after polls close, then we’ve finally arrived at a method I would trust.Now on to what can be done to improved both accuracy and trust in these systems:The basic approach I take is that recounts should be possible going back to the original ballot.On that basis, electronic machines with no VVPAT fail completely - there’s no ballot to recount. Nothing can improve the accuracy or trustability of these systems.Electronic machines with a VVPAT fare only a little better. Yes, it’s possible to hand count the VVPAT record, but it’s rarely done. Even most “recounts” on these systems don’t actually do a recount of the VVPAT, but simply repeat the consolidation process of taking the individual machines reported totals and producing a single total for the election. Rarely are the VVPATs from a single machine tallied by hand then compared against the machine reported results. And even if this was routinely done, there’s still the problem that hackers can mess with down ticket races that are rarely “verified” by the voter. How can we know - after the election - that the VVPAT accurately reflected the voter’s intent? Answer: We can’t. Because the voter didn’t actually mark the “ballot” (aka VVPAT) that we are later trying to “recount”. Lastly, what do we do if the hand tally of the VVPAT doesn’t match the machine tally? It’s a strong suggestion that there’s been chicanery somewhere, but the printed record is now in question - we can’t go back and ask the voter! Should we toss all the votes from that machine? Should we toss the votes from ALL machines?In short, any system that doesn’t provide the ability to know exactly how each ballot was cast can’t be relied on. And yet we do. And even if we were able to prove that votes weren’t counted as cast, we have no recourse to know how the voters intended to vote.The only system I will fully trust is one that uses hand marked paper ballots, counted publicly in precinct immediately after the polls close.But there are measures that would lessen my distrust. There are 2 measures:Routinely do “risk limiting” audits. Take a small random selection of actual ballots and hand count them. If the results match the machine tallied results closely, accept the machine counts. If not, take a larger random sample and try again. Repeat up to the point of a full hand count. It’s still possible that the vote counts were hacked - especially if the first random selection of just a few ballots happens to match the reported outcome - but it’s far less likely to be hacked.Where optical scanners are used - save the image of all ballots and publish them on the web. At this point, there’s no way of tying any ballot to a particular voter, so ballot secrecy is still preserved. And now anybody who wants to can do their own recount! And raise objections if a discrepancy is found. One thing is required here: Proof that the ballot pictures match the actual ballots. This could be done by printing a unique identifying code on each ballot and checking a small random sample of actual ballots against their image - in precinct in full public view. This number would appear nowhere else, and a process to ensure that there are no duplicates record would be necessary.Note that I haven’t addressed a few issues yet, but they are issues with ballot custody, not with accuracy:Mail in ballots:So you’ve been mailed a ballot. You’re either in a jurisdiction that does only mail in ballots, or you’re voting absentee. How do you know that your ballot isn’t “lost” in the mail? Recent revelations regarding social media micro targeting suggest that it’s possible to identify how you’re likely to vote. How much harder will it be to “lose” the ballots of people likely to vote in a certain way?Some mail in ballots will always be necessary - for people who can’t vote in precinct due to health/mobility issues or the need to be away from home on election day. Requiring in precinct voting would disenfranchise these people. But why take the ballot custody risk for people who can vote in precinct?Chain of custody after the election - to preserve the original ballots for a possible recount. “Sealed” ballot boxes can fail or be lost.In summary, the only results I’ll completely trust are from elections done on hand marked paper ballots with few people voting absentee. All ballots are counted in public view by groups of people including representatives of the candidates. Those cast in precinct are counted before the ballots are sealed for possible recount and forwarded to a central repository. Those cast by mail close to the election are opened and counted in public similarly - but at a central location. Those that arrive early enough should be routed to the precinct and counted with in precinct cast ballots.One final note on accuracy: The more ballots that are counted by hand, the more trustable and accurate count we’re likely to get. So why not count all ballots by hand? If the networks are so eager for quick results, let them pay for optical scanners that can count quickly - and require them to always report the results and “unofficial and non-binding pending in precinct hand counts.”!Original Question:Are you confident that American voting systems accurately reflect the vote of the people? Do we have any way of independently testing the accuracy of the results?