voting machines anyone?

[jimbo3123]

While developing a voting machine based on a PLC is an interesting exercise in thinking about the problems involved, I can see a couple of major disadvantages when compared against using a PC or other computer hardware.

One is that a PLC program does not run directly on the hardware supplied. There are complex propriety operating systems and translation hardware at work under the surface. Becuase PLCs are not specifically targeted at the voting machine market now, I would be inclined to trust that the user program instructions work as advertised, but that may not always be the case. I would perfer a system where all software is open to inspection and review.

The other problem with using PLCs is that they are not designed for this kind of problem. While encription and hash functions could be written on most PLCs, the hardware/software is definatly not optimized for this type of calculations.

I personally believe that an open source, verifyable system, running on common hardware (ie. PC w/open BIOS) is the only safe and secure way to run an electronic voting system. I can't believe that we have trusted electronic systems with our money for decades, but have yet to get the voting thing down.

 

[elevmike]

To start with I vote for using the AD DL-06 or Dl-240 for the voting machine. Small, inexpensive, Non Volitale (EEPROM) memory to hold the votes. Easy to program, and above all.. I have DirectSoft...

 

[Ron Beaufort]

personally, I’d like to see it all done pretty much the same way that Allen Nelson laid it out ... what’s wrong with just marking a paper ballot with a big magic marker and dropping it in a box? ... after the polls close, a lot of interested people take the paper ballots out and (keeping all of the cards on top of the table) count and tally the votes ...

the problem is that the “I-gotta-know-NOW” American public won’t go for that common-sense approach ...

after listening to many-many-many months of non-stop campaigning, America just can’t wait for an additional two or three days to get the votes counted ... heck, the news media won’t even let us wait until all of the polling places close ... they insist on giving us minute-by-minute “forecasts” about how the vote is going ... and note that these are the very same nitwits who can’t even reliably forecast what tomorrow’s weather is going to be like ...

in my opinion, I think that the reporting of ANY election results should be illegal until a nice civilized hour of say something like 9:00 o’clock the next morning ... personally I would have no problem waiting until then to see who really won ...

but as I said, since the American public (or at least the news media) seems to be hooked on the instant gratification of play-by-play updates, I’d say that the voting machine thing is here to stay ...

so ... can’t we come up with something better than what’s already out there? ...

and I liked the analogy using the banking system that we all know and trust ... but remember, we have ways of tracking the money that we put INTO and take OUT OF the bank ... with the voting machines in use today, we have no such way of tracking our voting transactions ...

think about it this way: if tomorrow the bank told you something like: “oh, we’re not going to be using receipts and statements anymore ... just hand us your money and we’ll take care of it for you ... you really don’t need a written record of any of the transactions that you make with us ... you can always just take our word for how much money you have left in your account” ...

that just wouldn’t fly, folks ... there has to be a system ... and the system has to be independently verifiable ... and the sad news is that we don’t have the ability to independently verify the results that today’s voting machines are giving us ... personally, I’ve got a bad feeling about that ...

so yes, I agree ... Dogpatch probably isn’t going to rush right out and buy a “Phil-O-Matic” voting machine ... but I still would like to see what we can come up with ... what will it hurt? ... except, of course, the companies who make the “secret code” models ...

and sorry, Rube, but it was Mammy Yokum who was the unofficial mayor of Dogpatch, not Li'l Abner ... you can read all about it here ... but you and I must be getting pretty old to remember anything about this particular topic of conversation ...

 

[elevmike]

The other problem with using PLCs is that they are not designed for this kind of problem. While encription and hash functions could be written on most PLCs, the hardware/software is definatly not optimized for this type of calculations.

Jimbo,

What type of calculations are we going to use that the PLC cant do?? For sure we arnt talking about floating point math?? I disagree a simple add instruction will be all that necessary to tabulate votes. PLCs as we all know are VERY stable platforms. In my book this is a much much simpler idea then the banking system.

BTW we all know nowbodys going to buy it because it will be too practicle, reliable, and cost effective, and that's not how the current system works.

 

[jimbo3123]

Paper Trail

For a long time I have resisted the notation that a paper trail is necessary to verify a voting system.
However, I'm starting to see that I may be wrong.

I do not think that voters should be allowed to leave a polling place with a reciept that shows who they voted for. This would make it too easy for people to sell their votes (or at least people could make this claim) However, they could be given a slip with a barcode that could be used (with the right access to the information) to verify their vote in cases of suspected vote tampering within the system.

I think that the individual voting machines should be networked to central database servers that immediatly recieve those choices and report back to the machine with a CRC or other hash of the vote choices that were recorded. That vote hash (basically a glorified parity check) of the vote would also be assigned a unique ID number. The ID number, combined with the hash of the vote card would then be printed on a paper tape (or even hard-drive/flash memory/etc) within the vote machine. This tape could be used to audit the machines/system. The ID could also be printed (possibly endoded as a barcode) and given to the voter as a recipt.

There should be no need for a recount, as a computer system should always return the correct answer with identical data inputs.
However, this system should allow for adequeate auditing of every step of the process.

 

[jimbo3123]

personally, I’d like to see it all done pretty much the same way that Allen Nelson laid it out ... what’s wrong with just marking a paper ballot with a big magic marker and dropping it in a box? ... after the polls close, a lot of interested people take the paper ballots out and (keeping all of the cards on top of the table) count and tally the votes ...

the problem is that the “I-gotta-know-NOW” American public won’t go for that common-sense approach ...

In this day, when digital electronics are present in just about every aspect of our lives, I don't think that there is wrong with expecting that voting be done the same way.

Electronic voting should be faster and also more accurate than using a paper box. Remember, we're talking about a country with about 300 million people.

 

[jimbo3123]

Jimbo,

What type of calculations are we going to use that the PLC cant do?? For sure we arnt talking about floating point math?? I disagree a simple add instruction will be all that necessary to tabulate votes. PLCs as we all know are VERY stable platforms. In my book this is a much much simpler idea then the banking system.

BTW we all know nowbodys going to buy it because it will be too practicle, reliable, and cost effective, and that's not how the current system works.

If you read my next post, you'll get an idea of the type of system I'm talking about. The functions would be required for secure (encripted) communications with the central server, as well as hash coding/decoding from/to a completed vote card. This card would include all candidates as well as proposals, etc. present on the ballot.

 

[jimbo3123]

More on hashes

As an interesting side-note,

There exist hashes that could be used to verify a completed ballot, but could not be used to re-generate that ballot. Such one-way hashes generally require a bit of computing power. They could provide a good way to verify a voter's choices.

The one example that I'm familar with are ZKT's (Zero knowledge tables). I do know that other schemes exist for doing the same thing though.

 

[akreel]

For sure we arnt talking about floating point math??

We don't need that anymore. Everyone is now entitled to one full vote, each.

It's interesting to see how many ideas are already being offered here. I agree with the nonvolitile memory option. In fact, if you could use a $99 DL-05 you might as well pull the CPU after the election and file it away. Use a new one next time. You still need a PAPER copy, though.

In my mind, the program should be VERY simple: up counters and you're done. The REAL programming is all the stuff that interfaces with the real world: printing receipts, HMI screens, etc. I think gizmos like encryption would add complexity to a program that we've "decided" should be simple enough to explain to a jury of 12.

AK

 

[TimothyMoulder]

The devil and the deep Red Hat

The reason these voting computers are so secretive is simple to understand: easy to inspect for fraud is easy to tamper with for fraud purposes. If everybody can understand it, then the hackers will have a field day with it. For the programmers, disclosure is exposure.

Personally, I'd rather see those boxes locked. I'm not worried about home-grown fraud, because politicos can't keep their mouths shut - look at what happened to Nixon. They'd be caught eventually, and that impeachment trial would make Clinton's look like the joke it was.

I'd be more concerned with our election being decided by a consortium of freelance hackers hired by the Saudis. Face it people, 17 of the 20 hijackers were Saudi - they are NOT our friends, and they have enough money and sophistication to hire the technical know-how.

Our little PLC world is secure mainly because there is no interest in hacking a mixing machine or stoplight. Most of our favorite products trumpet the fact that they are network accessible, and can be modified from anywhere. Not the kind of product I'd want securing my vote.

Personally, though, I'm still waiting for Flori-duh to suffer another debacle when all the voting machines across the state flash to a BSOD at 11 AM Tuesday morning. At least we PLC'ers don't have THAT problem...

TM

 

[Ron Beaufort]

for a basic “how are we going to handle that?” question ... how will the candidates’ names be loaded into this thing? ... is this a touch-screen set up and then download operation? ... or something as uncomplicated as a cardboard template laid over a row of dirt-simple pushbuttons? ...

 

[MikeW]

Ron,

think about it this way: if tomorrow the bank told you something like: “oh, we’re not going to be using receipts and statements anymore ... just hand us your money and we’ll take care of it for you ... you really don’t need a written record of any of the transactions that you make with us ... you can always just take our word for how much money you have left in your account” ...

It's already here. My credit union went almost completely on-line last year. If you want a printed statement, they will send you one but also charge you for it, not sure what the fee is as I decided a secured connection to the CU was good enough for me. Between on-line payment, direct deposit, and cash machines, about the only time I go to the actual building is to access my safe deposit box or deposit a random check from a non-standard money source.

I'm not sure the rural township I currently reside in is ready for this "high tech voting" thing. I moved from a suburban location three years ago with it's optical reader and now use a real paper ballot. Your choices are filled out in plywood booths with a privacy curtain that looks likes it's somebody's old living room or kitchen curtains held up with shower rings. Fold your ballot when done and put in the wooden box. And don't forget to give back the #2 pencil for the next voter! No complaints from me as it should keep my taxes down a little bit.

The Diebold code was looked at by John Hopkins University and they found some flaws in it that makes one wonder what will happen on Tuesday.

 

[elevmike]

for a basic “how are we going to handle that?” question ... how will the candidates’ names be loaded into this thing? ... is this a touch-screen set up and then download operation? ... or something as uncomplicated as a cardboard template laid over a row of dirt-simple pushbuttons? ...

I like the touch screen. By the time you wire all thoes buttons it'll cost a lot more. The PLC code would be simple. The touch screen a little more complex. The network, well I'll leave that to somebody else....

 

[Tom Jenkins]

Use a DL-06 or ML-1200 for each. A touchscreen with names for each booth, tied to a dedicated PLC. Discrete outputs to run each punch for the tape. I prefer punches to priner because they are going to be more reliable, and less prone to tampering. Discrete inputs for the "I'm done" lever and "I'm closing the booth" lever. Simple counters for totals moving the count to separate registers. (You might need that floating point math because you might have more voters per PLC than the integer registers need.)

A PC would be set up to read totals from each PLC every 15 minutes, with a printout on hard copy of running totals at these 15 minute increments. That would make tampering obvious, since a sudden jumnp in one candidate would be obvious. It wouldn't be hard to write a "crippled" comm driver that only had read commands, no write commands. The PC could be networked, and a firewall used to keep hackers from getting to the PLC comm link.

At the end of the day you could have a separate auditor come in with a PC and read the PLCs for a back-up record and verification of the polling place PC.

For a recount you could run the tape through a reader, using photocells to trip discrete inputs on a separate display and PLC unit. If you really like inspectors looking for hanging chads, you could also manually read each ballot section of the tape.

 

[elevmike]

Tom,

I think you've got it! I vote for you as the chief resident project manager..

p.s. I'm still trying to stuff the ballot box for the DL-06.