Letting the Voter Count

[glatt]

I'd have to disagree. All the things that make computers great are also what make them easy to mess with. It's very easy to change data in an electronic format versus a paper format. A couple keystrokes and you can change a thousand ballots. With paper it would take all day to do the same thing.

Tampering isn't even the thing I'm most concerned about. It's computer glitches. What happens if the computers crash? Typically, districts lean pretty heavily toward one party or another, so if a power outage, line surge, or system failure causes all the votes from one district to be lost, that could turn the election. With paper ballots, you would have a record of the votes.

[Undertoad]

Well I'd have to disagree right back. Add a system of electronic auditing and it becomes much much harder to mess with that data. Mark the vote with encoded timestamps, put incoming data in several different places, etc. Smart people have come up with a lot of ways to ensure data integrity. They've thought of that stuff.

[Happy Monkey]

It would, indeed, be a good thing if they added a system of electronic auditing, encoded timestamps, and redundant storage. They haven't. In fact, they haven't added a system of human auditing - they say that the code is a trade secret, and nobody can see it, and object to any testing of the system that they don't conduct themselves - with good reason. Independent testing regularly fails.

[Troubleshooter]

So, in essence it falls back on the same old problem.

The technology is more mature than the operators.

Sad really.

[glatt]

I'm a big fan of technology, and far from a luddite, but technology can fail. When it fails, it's almost always because it was implemented poorly by humans. As Happy Monkey mentioned, this stuff hasn't really been tested yet. And the big test is an election that will impact the world.

Sure, in theory, an electronic system can be set up to allow secure and accurate voting. However, in reality, that system will be run by temporary employees who have received just an hour or two of training in its use. It will be created by companies that hide their work, and it will be implemented by election officials who don't necessarily have technology backgrounds.

A hybrid solution, like scanners that read paper ballots is really the best of both worlds. You can use the machines to quickly tabulate the results, but the paper is still there in case there are screwy results or questions.

[tw]

I am also a major fan of technology having even soldered together the ICs to make a computer long before PCs even existed. But I am also an engineer who must see major advantages to justify the more technically complex systems. The electronic voting only complicates the system without addressing or solving the original problems.

Yes paper only ballots can also be 'gamed' (as Undertoad notes) because (in part) too many hands are on the system. Paper only ballots, once sufficient, are not longer appropriate for large voting numbers and where voting error in hundreds (and maybe thousands) of votes was quite acceptable. We are also talking about storing and processing tens of tons of paper.

Posted up front were two necessary standards for making a poll work properly:

First, the vote must be written to (stored on) a 'write once' media. Paper does this. Paper votes cannot be modified without detection. Most electronic systems violate this principle.

Second, the voter must be able to confirm his vote is as he intended. Again paper does this.

Arguments should have started with these standards. Then have expanded to include the other standards I intentionally did not provide.

For example, (third point) voting must be anonymous. One must be able to vote without anyone knowing how that unique vote was entered. For example, if you walked away with a copy of how you voted, then others could intimidate your vote. The wife should be able to vote as she pleases without the husband having any say. Wife should not be able to prove the the husband that she voted as he demanded. One should not be able to 'sell' his vote. Also a vote must be completely untraceable meaning that time stamps on a voting record violate that anonymous voting principle.

Furthermore and fourth, a voting system must have the ability to confirm the entire voting record. An audit must be possible to both verify the count and verify the security. Punch cards locked in a ballot box do this. Should the punch cards be read by a machine, then a recount can be preformed on those punch cards. Currently the only voting method that appears to meet these voting criteria is optical voting machines as well as the older mechanical voting machines still being used in NY, CT, LA, most of eastern and Pittsburgh area PA, and western VA.

Fifth, the entire voting system must be secure. In Scientific American (if I remember the source correctly), authors literally walked into voting machine storage areas and opened doors to voting machines. No one challenged them. There was no log of everyone who entered and left the building.

Yes, I suspect so many are so technically naive (deficient) as to be enthralled by electronic voting rather than first learn the standards (principles) on which good voting practices are based. Above are some basic standards that all voting systems must be dependent on. Moreso, we have a serious problem with the requirements of a system. Even HAVA money was not even spent? Where has the administration been for almost four years (besides spending all this time and massive monies not even authorized to invade a nation that was a threat to no one)?

If voting is done electronically, then voting machines must be literally hardwired. The computer is a dedicated hardwired circuit, it features in-circuit confirmation that each 'hardwired' computer responds to (a unique code input results in a unique code output so that any machine without the hardwired chips would not verify - and I have not explained this sufficient to be understood), and the votes are written to a write once memory. A memory complete with security codes and the only place where the candidates names are provided to the machine. For example, a write once memory chip could then replace massive paper, contain code that traces that chip directly to a unique machine and that unique machine configuration, is handled minimally (not be gamed like paper ballots), and is treated with all the security required of ballot boxes - complete with security handled by multiple people simultaneously.

Most every electronic machine I have read simply violates (and grossly) the principles of secure voting. Using disk drives and software program is a gross violation of secure voting principles. You have no way of knowing how your ballot is being written into the write many times and easy to manipulate memory. If every machine has the same hardwired voting machine - a computer that is literally only programmed unique only by the voting names on that write only IC (voting machine does not even have firmware), then we know your ballot is being written to the memory chips by the exact same hardware also operating in all adjacent counties. Either all voting machines are fraudulent - or none - and the same secure hardware (just like lever voting machines) is used every time.

Recount literally involves reading every vote from memory chip and the corresponding error detection codes. Recount instead becomes a confirmation process where all hardware and the data in that memory chip is verified by secure and independently random codes. IOW recount becomes a security audit as well as a count of votes permanently stored in a write only memory device.

Optical voting also has great advantages. First you vote. Then the vote is physcially taken to a machine that reads your vote media. Furthermore, your voting media (punch card, smart card, etc) is locked into a ballot box where a recount can be performed completely separate from the machine that original accepted and tallied your vote. Of course, the voter leaves 'anonymously' with no record of how they voted so their vote cannot be sold. You literally see what your voting media says how you voted when you submit that voting media to the separate hardwired counting machine. Two separated methods of tallying the vote so that a recount is secure and separate from the original counting machine.

Just some of the many ideas for voting - using principles that all electronic voting machines violate.

Diebold sells the AccuVote-TSX which provides no paper (or other) record to make security auditing possible nor any method where the voter can confirm his vote is being properly registered. AccuVote-TSX meets the paper record requirement by simply printing the final totals on paper - violates the reasons why that paper record is required. Voter is given a voter card that permits access to the voting machine just once. Diebold uses the Windows Operating System rather than a secure and dedicated embedded software program. Diebold is also the machines that created massive failures in a SanDiego county electronic last April when the hardware drained batteries of power before voting could even start. Probably good that the batteries did not die during voting. CA decertified all Diebold AccuVote machines in four counties for numerous problems with security, auditability, etc. UT believes Montgomery County is using Diebold machines.

Second large manufacturer is Sequoia Voting Systems Inc. Diebold and Sequoia claim to own 80% of he market. But ES&S claims to be the world's largest maker of electronic voting machines. Ironically, ES&S was founding by Todd and Bob Urosevitch. But Bob is now president of Diebold.

Provided above are but some five standards that a voting system must meet. Does your poll? Unfortunately, even good standards for polls - four years after the FL fiasco, still don't exist nationwide. Voting remains a hodge podge of some juridictions with people who learned the standards and logic; and other jurisdictions where officials are impressed (emotion replaces logic) only because the machines are computerized.

[Happy Monkey]

Excellent post. It's appalling that the current electronic machines weren't laughed out of the room when first shown.

[Troubleshooter]

Originally Posted by Happy Monkey

Excellent post. It's appalling that the current electronic machines weren't laughed out of the room when first shown.

That's because it wasn't technicians looking at them.

[Undertoad]

UT believes Montgomery County is using Diebold machines.

I said precisely the opposite: I only knew they were NOT Diebold.

Now having checked, they are Sequoia.

[Griff]

Nice points tw. Every system can be gamed, requiring a certain amount of trust or at least mutual distrust between parties. With paper ballots and optical readers at least gaming is detectable. A couple election cycles back Bridgewater twp was the last to get their ballots into Montrose. It took something like 8 hours to travel maybe 5 miles...

[tw]

Originally Posted by Undertoad

I don't remember who produced the machines (not Diebold), but all election officials in Pennsylvania are elected, not appointed.

I don't remember voting for election officials when I lived in Montgomery County. Do every county (in and outside of PA) elect their voting officials?

[OnyxCougar]

Well, I'm in NC, and don't know if election officials are voted in or appointed, but I have never, in the three states I've voted in, seen an election of electors. I mean, not even on the ballot.

Dumb luck? Let's see... Nevada, Arizona, Oregon, (and now) NC.

[Undertoad]

Originally Posted by tw

I don't remember voting for election officials when I lived in Montgomery County. Do every county (in and outside of PA) elect their voting officials?

Yes. There are three officials at every polling location: a judge of elections, a majority inspector and a minority inspector. Often there isn't enough interest in the job and the parties fail to run someone, and in that case the position is nominated. My ex won minority inspector in that way - when the Ds failed to run anyone, she asked a few people to write her in, and won with 5 votes.

This election happens in odd-numbered years which are the "less important" elections and so, less memorable and people often don't vote in them.

[xoxoxoBruce]

and the votes are written to a write once memory. A memory complete with security codes and the only place where the candidates names are provided to the machine. For example, a write once memory chip could then replace massive paper, contain code that traces that chip directly to a unique machine

Thanks, TW. I've read a bunch of articles on electronic voting, over the last 4 years and none of them put it all together.
Question - When the votes are written into a write once memory, is it every choice I make goes into a permanent field after the guy before me and before the guy after me? Or does a vote for Joe Blow get routed to Joe Blows field?
The reason I ask is because my poll (Delaware County, PA) keeps a record of names (signatures) and the booth used, in the order they vote. Hence, sequential data fields could be easily matched to the voter list.

[Happy Monkey]

TW's standards are a theoretical machine, not a real one, so we might as well add some way of randomizing the order of votes on the media.