BBV - Voter Fraud will NOT happen in the TS code but in the Operating Sys.

[Florida_Geek]

IMHO.

A little background on me, I have been a "systems level" programmer since the late 60s. I have worked on everything from embedded computers in ticket scanners to large IBM mainframe computer. I have wrote software that ran at the lowest level in an IBM mainframe.

A little background on funny things operating systems do. I have made this as simple for the non-computer types as I can. I will pick IBM as an example. From the early 70s to the mid 90s IBMs main operating system for their mainframes was called MVS (it is still used but IBM has changed it's name). All computers run in one of two modes, supervisor state, or application state. In supervisor state your program can do things that programs running in application state can not do. (Windows blue screens, if a program in application state gets a program fault you get a nice pop-up, BUT when you see a Windows blue screen, that means the program was, probably, running in supervisor state).

IBM programmers wanted to pop into supervisor state on the fly in their normal application state programs. So they had a stealth method for doing this. There are many systems calls a program can call, for now lets talk about the time-of-day system call. So IBM did some funny things with the time-of-day system call. IF you loaded a secret value into one of the computer registers, not only would you get the time of day return to you but your program would be running in supervisor state. If you looked at the application program, nothing looked strange, just a "normal" call to time-of-day. And because most people could not see the code for the Operating System from IBM, this method was really used for many many years.

What does this have to with Touchscreen Fraud. Well there is a major push to get the code for Touchscreen software to be "Open Source". This is a RED HERRING Even if the TS code is release and reviewed, IMHO nobody will find the fraud because it will be done in the non-reviewed operating system.

Let's pick the famous voting machine company Acme (beep-beep) Voting Machines. Well Acme picked for their operating system for their Touchscreen product the Linux Operating System. You can get Linux source code and compile it yourselves but 99.99% of the people get a pre-compiled version from somebody like Red Hat. But if you go to the Red Hat site, you can get the source for most of their Linux code. And if you wished to, you could re-compile a module yourself.

How TS Fraud could happen at the Operating System Level

OK some very simple code examples.

First a voter record with three fields
1) One character for a vote for Bush.
2) One character for a vote for Kerry.
3) One 36 character field to hold the time of day.

Second in the TS software we have some code that looks like

move -1 to hold_time_of_day.

{10-20 statements }

Remark - go get the time of day for this voter.
call time_of_day with hold_time_of_day.
.
.
record vote......

If a outside company reviewed the TS code, there would not be anything to point to voter fraud. Why because the votes were changed in the time_of_day function.

Acme had change the Red Hat Linux time_of_day function to do something funny if the hold_time_of_day field was -1. It would change every 10th vote for Kerry to a vote for Bush.

I have made this as simple as I could. In the real world this would be a lot more code in the application program and the voter record have a ton more fields, but you get the idea.

The "Open Source" Red Herring

The Linux Operating system with all it's window managers, something called OpenGL for drawing pretty pictures on screens etc is on the order of over a couple million lines of code. The Touchscreen app from Acme is probably on the order of 50,000 lines of code. So when Mr. Nice Guy Acme turns over the source for the Touchscreen app to an independent reviewer to compile and test, everything will work great and no fraud found. WHY because the reviewers are not looking at all the source for, in this case, Red Hat's Linux.

In this simple example, Acme Voting machine programmers had changed less than a 100 lines of code out of a couple million from Red Hat and only ONE module. They had two versions of a "common" Linux module. One that committed voter fraud and one that didn't. The two could be made to be the same size, same date, and same CRC (geek method that tries to come up with a unique number for a module, but can be fudged around).

Please do not fall into the trap of thinking if you can review the source code the product is "safe" from voter fraud. That is only a Red Herring IMHO.

[Wickerman]

1. well stated

I even got it. Fuck, one more reason not to sleep tonight.

[KittyWampus]

2. Really, FloridaGeek Must Know Their Stuff Cause Even I

understood the main point...

[never cry wolf]

3. Well, I am in the minority here so far

I did not really understand very well. But that is not all that surprising. I am very ungeeky software-wise. I am more into architure, the building construction type. I have installed hard drives and RAM chips and can do a defrag and virus scan, etc. Nevertheless I have followed BBV and think I get the gist of thsi enough to kick it so the BBV activists more knowledible than I can peruse it.

[ex_jew]

4. I say BS !

Just what was this "time of day call" ? TIME macro ? STCK ? And you're telling me that with the millions of man hours IBM invested to keep MVS tight, they just happened to leave a trapdoor ? Now of course independent software vendors might very well pull this stunt, but IBM ?
Why would they need to ? They have supervisor calls, program calls, authorized programs, whatever.

I would really like some evidence to back up your assertion.

[Florida_Geek]

8. It was not the time of day system call

IBM used, but I picked that one for something non-computer types could understand. But there was a very standard system call that did do what I said. :)

[progressivebydesign]

5. Okay. A question..

With all of the emphasis on making a paper trail, or an ATM type receipt to show who you voted for. Aren't computers such that you could program anything to show one thing on the screen, one on the receipt, then show something else in the tally???

[Andy_Stephenson]

6. That is why you have a random mandatory audit

of 5% of the machines. To catch any fraud.


EDIT: BTW call it a ballot...not a trail or reciept.

[Florida_Geek]

7. Right have a piece of paper printed out

is 90% of the solution, the audit is the last 10%. Even without the audit, the fear of a recount using the paper ballot would stop fraud, IMHO. Most people would not scan the printout for every thing they voted on, but they would probably check the one or two major people they voted for.

Old War Story - Free Money from Diebold ATMs in Tampa.

In the late 80s or so, my sister was moved into auditing ATMs for a major bank in Tampa. The person that should have audited the ATMs was over 6 months behind. In the end there was a two month period the bank gave away free withdrawals from this banks Diebold ATMs do to a bug in the Diebold software that the bank did not catch in a "timely audit" that allowed something like 1 in 5 withdrawals not to be processed correctly. The banks main ATMs were IBM, but they had Diebold standalone units in grocery stores. Added to the slow audits was the bank's ATM programers that only wanted to work on the IBM ATMs and did not test the Diebold software changes.

[RedEagle]

10. No, Not an Audit of the Machines....

...random audits of ALL the BALLOTS.

Remember how the Secretary of State fudged the random audit requirements (when he capitulated a bit) in Washington State?

The definition of random audits was of ONLY poll-site electronic voting devices.

Since Washington State votes heavily absentee, and most paper ballot systems are central count in the State, according to the County information on the Secretary of State's web site, there will likely be fewer touch screens than optical scan. The very, very specific audit language in the legislation meant that ONLY the touch screen machines would be audited- leaving out what may be close to 80% of all the other ballots in the state.

Call for random audits of ALL the BALLOTS.

Don't get machine language in there or they will mess up the intent.

[BevHarris]

12. Yes. The devil is in the details.

Unlike accounting audits, where you go on and audit some more if you have a hunch that something isn't right, these jokers use audit rules to LIMIT what can be checked.

Bev Harris

[SoCalDem]

9. Paper ballot with time stamp & unique number on TS "vote" cast

..Voter reviews paper ballot (looks like atm receipt..but just names of who you voted for)..

Voter places "paper ballot" into box on sign-in desk..

at the end of the day.. The "ballots" are manually counted and a number is posted..

Machines are "counted" and a number is reached.. If the sign in says that 325 people showed up, there better be 325 pieces of paper in the box..and the machine count better say 325..

For the major races..president, senate, congress, the paper ballot totals should be tallied..

Blackboard & chalk should do nicely..

Then they should be verified by what the machine count says..

This method should be used UNIVERSALLY for this election, especially..

MOST people "know" that things are rotten in our election system and the "machines" will not make them more comfortable until they KNOW FOR SURE that it's right..


It's EASY.. If they need more poll workers, HIRE THEM.. Goodness knows there are enough unemployed people around these days.

[hunter]

19. Not so easy...

"If they need more poll workers, HIRE THEM.. Goodness knows there are enough unemployed people around these days."

This might not help. If election officials are simply hiring extra eyes and hands they will probably use a labor contractor. The contractor may or may not be honest. The people the contractor hires may or may not be honest.

The people working for the lowest bidding labor contractors may be so desperate for work that they will consciously or subconsciously overlook a wide variety of election "irregularities."

I think a lot of labor contractors keep a mental list of workers who "don't cause trouble" and they send these people out on jobs that don't quite smell right.

[SharonAnn]

16. Absolutely right! I can do that in programming and even hide it

(somewhat). I'm sure that programmers more knowledgeable than I can do it in a heartbeat and make it impossible to find.

That's the best description:
1. Show one thing on the screen
2. Show something else on the receipt
3. Do something entirely different for the tally.

On edit -

That's why robust audting procedures are absolutely necessary. At least if there's a Voter Verified Paper Ballot created and retained, there can be a manual count of the ballots and a match of them against the electronic tally.

Robust auditing includes several things, with random machine recounts and random precinct recounts as part of it.

[yella_dawg]

11. The solution is obvious

Publicly owned voting hardware / software that is fully tested and validated. The problems are due to possible conflict of interest in the vendors.


A few points:

A paper trail is absolutely necessary.

Networking of voting machines to expedite tallies is unacceptable in any configuration. In a network, one compromised machine invalidates an election.

Dedicated hardware with embedded code that is validated as a unit is desirable.

I spent years doing aircraft weapons systems. The accuracy and security needed to tally votes electronically is possible and realistic. But only in a system designed for the purpose. Existing voting systems are extensions of general purpose computers.

I disagree that open source is not a good solution. The simple solution to the problem presented is to compile a validated OS at the point of use. The point is, whether you use embedded Linux or slash marks on tree branches to compile votes, the people must apply substantial effort to insure that the count is accurate. Electronic voting is a good solution. It just isn't an easy solution.

[BevHarris]

13. For the above, check out the most high-profile project yet

for open source, FREE public voting software.

It's called the Open Voting Consortium and has people like Dr. Doug Jones support. http://www.openvoting.org

Will be on the Jim Lehrer Newshour this month, and has hit the New York Times, and the Washington Post, already.

Bev Harris

[TinfoilHatProgrammer]

28. hurray for the open voting consortium

Too bad they wrote their "product" in Python, which is prohibited by FEC standards and can't be certified.

Nice try. Back to the drawing board.

JC

[Florida_Geek]

17. Well one small problem with that,,, what is coming out of the EPA

and others Federal dept. with BFEE heads running them.

It would have to be not controlled by the Feds.

[sendero]

18. As a long-time computer professional...

.... who has also spent years doing software testing - I can tell you there is no way to absolutely prove the integrity of a system by testing.

For example, using a technique similar to the one outlined in the original message, I could have a simple "flag" in the program that told it whether to work correctly or shave like a bastard. Without looking at every line of the ACTUAL code (and even looking at all of the code, including the OS, comm, EVERYTHING, it might be missed) you would simply never find an error in testing.

On election day, I could connect to the box in some way and "set" the cheat flag and off we go. I could also program the box to work perfectly except between 7 and 7 on election day.

Open source and/or a voter-verifiable receipt are the only real solutions IMHO.

[yella_dawg]

21. High reliability military software

is inspected line by line, by a committee. I'm not suggesting a quick fix. I'm suggesting a laborious but valid technique to guarantee voting machine reliability to a high degree of correctness. In a correct system, you'll never be in a position to trip your "flag", since proper procedures in secure software isolate the developer. Tools flag switches as focus points for testing. Secure, correct software is possible to a high degree. But not without a powerful, validated development process. And not when there is any possible conflict of interest overseeing the process.

The point is, the current electronic voting systems are a bad joke.

Paper receipts are an absolute necessity. Any software is vulnerable when there is no possibility of discovering a fault.

[sendero]

22. Unfortunately....

.... even if the code is inspected to death, it is very tough to ensure that the code that is inspected is actually installed on a given machine.

[Florida_Geek]

23. That is the REAL problem

That why a paper receipt is a major answer to the problem.

[ParanoidPat]

24. It's a paper BALLOT damn it! NOT a 'receipt'!

A paper ballot 'receipt' looks like this....



.....which is USELESS in a recount! :(

That's a "receipt" that I get for my BALLOT! Asking for a 'receipt' is not the same as asking for a paper BALLOT!

The devil IS in the detail. :evilgrin:

It may seem like a small detail to some but when it comes to getting the proper legislation to protect our voting rights it makes all the difference in the world. Please always refer to it as a paper ballot not a receipt! That word is what separates good legislation from very bad 'feel good' legislation. Both have been written, back the laws that demand a PARER BALLOT! :) (and the accounting that will use them to keep elections fair!)

[yella_dawg]

26. Not at all...

Software deployment specialists validate the hardware / software package, insuring a known system is in operation. You solder a certified ROM chip onto a board and test it. Happens all the time, which is my main point. The military software development process already handles most, if not all, of the security issues that surround electronic voting. A defined and well documented process exists. Why the F*** don't we use it?! That in itself reveals a serious flaw in our approach to electronic voting. We've reinvented the wheel... And it wobbles.

[TinfoilHatProgrammer]

29. high reliability military software?

After reading this thread it seems clear that we shouldn't trust any military software that isn't open source... there's no way it could possibly be reliable if it's developed in secret, particularly by government employees. ;)

JC

[PATRICK]

14. The tactics themselves are disturbing

The resistance and then the pointedly insufficient "cures" hawked as more of the good as gold software in and of itself reduces the discussion to the voting public versus a dishonest used cars salesman.

Not enough simply to argue their side, they continue pasting one false veneer over another EVEN accepting that changes must be made, that a problem exists- so long as the true fraud potential remains totally inviolate.

It is hard to characterize this methodology as anything else except a dedication to fraud, especially when they pretend to go along with the basic need for improvements. Were they honestly conceding the point they would make real fixes with real accountability. Instead they concede the argument and the pressure and then do nothing. This can have no other motive than intent to commit fraud.

Unless you nail their hides to a wall this protection of fraud potential means more than the integrity of the their product. That is, their systems will flourish when they deliver elections to their chosen ones, not when they deliver a genuine product to the electorate. Again the votes of the many mean nothing against the tyranny of the few.

[BevHarris]

15. This is the most eloquent and accurate description I've heard yet

and I want to know: Can I use it? This is good stuff.

Bev

[shance]

20. This is great information. Needs to be sent to Randi Rhodes/Air Americ

She continues to talk about a receipt and how the computer will print out a receipt that stays in the computer, lalala.... doesnt seem to grasp this is just another rather misleading and fake solution to relieve doubting voters minds, and throw us back into ignorant voting bliss, continuing to allow business as usual for companies like Diebold and ES&S.

[leftchick]

25. well, as I told our local Democratic precinct leader..

We need paper and pens for Nov. 2004. For Cripes sakes we are using them in our primary why not in November? :grr:

[RedEagle]

27. Back Up

Kick!