Maryland Plans Code Review for Voting Software

asmithmd1 writes "We already knew Diebold software is insecure, now the Baltimore Sun is reporting that the Governor of Maryland has asked SAIC to review the software in Diebold voting machines. Diebold has graciously allowed SAIC access to their proprietary code. Why isn't this code open source by law?" In a related story, a trade show for closed-source electronic voting systems is doing their best to keep critics out. Update: 08/07 15:23 GMT by M : Diebold's website security is less than outstanding.

Comments

the problem is...

[borgdows]

even if the code is opensource, how can you be sure the voting machine executable has been compiled from the genuine source code ?

Re:the problem is...

[digitalunity]

I won't trust digital voting. A lot of people won't. I seriously hope someone hacks it in November next year to such an extreme amount that the politicians see the error in their stupidity. This won't work. I hope the results are wacked out like this one

Re:the problem is...

[ajs318]

I agree with the original person. I can't state it loud enough that THE INTEGRITY OF ELECTION RESULTS IS A FAR GREATER CONCERN THAN ANY CORPORATION'S RIGHT TO SECRECY! The mechanism by which our leaders are chosen must be absolutely open to public scrutiny and any government that does not believe this so, deserves to be overthrown.

even if the code is opensource, how can you be sure the voting machine executable has been compiled from the genuine source code ?

I've looked at this one before and it is a problem, because the C compiler may be rigged so that when you try to compile it from source, then it modifies itself subtly so as to insert various backdoors; in other words, the code you get from the compiler does not match up against the source you compiled. Then, it does not matter how "clean" any of the source code is; because the compiler might modify the code during compilation. Even if you run the original, clean compiler source through it, chances are that the compiler could spot this and mung it, giving you a "dirty" compiler.

Throughout the following, I'm assuming you - or someone you trust - can spot malicious C code just by looking at it, and can write assembler code you know is safe. You don't have to be able to look at someone else's C-generated assembler and know whether it's safe.

You first need to write a simple C interpreter in assembler. Note, it only has to interpret; it doesn't have to compile. As long as the assembler instructions it generates do the same thing as the C source code you feed it - even if much more slowly than a compiled version - then it is good enough for the time being. It can even waste as much memory as you can spare. The most important thing is that you know the temporary interpreter is safe. Then you take the source code for the compiler you want to compile - you know this is safe, but the pre-compiled binary might not be safe - and run it through the interpreter. Now the output from the interpreted compiler is actually a compiled compiler, and it's safe. You haven't run the "dirty" compiler binary, which might have modified the compiler.

Now you have a compiler which you know for certain isn't going to produce binaries which don't do what the source said. And that's the first step to trustworthy computing. Maybe get someone we all trust to sign the code by encrypting it with their secret key {so when you decrypt with their public key you recover the original; recall that P(S(x)) = S(P(x)) = x}. Problem is, you can't trust anyone with election results, because the stakes are so high.


On the other hand, why bother with voting machines at all? In this country, we count votes by hand. It may not be high-tech, but it works and it's harder to subvert. Hand-counting of small batches of papers {which are kept, in case of dispute, until the next election is out of the way} is not significantly slower than machine counting. Anyway, what's a few hours here or there when a term of office can last for four or five years? To throw an election, you would need to bribe several people, not all of whom are politicians. The ballot paper {taken at random from a book of identical ones and by a different person than the one who sees your voter ID - the only communication between them is a slight nod} is the only record of the vote, and the voter has already had the opportunity to verify it before depositing it.

For how often elections are held, it probably is less work to keep on doing all this stuff by hand than it is to put the safeguards in place that would make machine voting trustworthy.

Re:the problem is...

[Jeremi]

Who cares about the code? If the machines generated a voter-verified paper trail, you could check the results for accuracy. Since they don't, you can never be sure that the results were correct, no matter how carefully the source was scrutinized.

I know what I am doing next election

[chrisgeleven]

Voting via absentee ballot. At least there is a greater chance of my vote not being screwed up or changed.

Anyone who thinks these voting computers are going to be flawless and secure by Nov. 2004 is greatly mistaken.

Not open source because...

[JohnGrahamCumming]

Why isn't this open source by law?

Well because the US is a capitalist country and because currently most people seem to believe that the best way to make money in software is by keeping the code proprietary and because US government favors money-making corporations.

I agree that if it were open source it would be far more likely the security problems would be discovered quickly.

So how about creating an open source alternative... anyone ready to register an OpenVote system on SourceForge?

John.

It makes sense

[Doesn't_Comment_Code]

It makes sense that they don't want their code to be open source, because then ALL the bugs will be found. When open source code is developed normally, people notice bugs/security holes a few at a time and fix them. But when software has been closed source for a long time, it's bound to have tons of bugs and holes. Opening the code up to public scrutiny would unveil A LOT of problems. And that's just not good for PR... especially in voting.

My guess is they just want someone to look through the code, maybe suggest a couple quick fixes, and then give the OK, so they can reassure the public. They don't really want to get to EVERY hole in the code. They probably just want show that they get numbers close enough that we should keep using/buying their stuff.

I don't care about the code...

[JayBlalock]

But for the love of god and all that is holy, WHY are they fighting so hard against paper records? It makes no sense. (unless you are conspiracy-minded) Seriously. I just can't come up with any decent reason that Diebold et al would be so strongly against hooking a printer up to the system to produce a physical record. Much less why our elected officials would buy into such an idea.

If you want it open...

[TWX]

you need to build it. They're not interested in building it open source apparently. Remember, Diebold makes ATM machines and other commercial products, and they have stiff competition. By the design of their business plan their software won't be open.

So, if you want to see an open source implementation of voting software, something that you can argue is perfect and be able to show the world such, you need to make it. Diebold and their competitors won't.

If you can build the software to make a secure voting system, someone else can design the hardware once the software is ready. That seems to be what people are missing here. Design the system right and the hardware will be built to work. Design the hardware first, and the system will be dependent on whatever wacky design is chosen.

Open Source doesn't solve this problem!

[xphase]

"Why isn't this code open source by law?"

This wouldn't fix the problem of faulty(by design) hardware, lack of audit trails, and no trust in the delivery method.

Sure with open source we can see the code, but that doesn't help if it is compiled by a compiler that you can't see the code for, run on microchips that you can't see the code for, and administered by people you can't trust.

The ``but it should be open source'' comment that gets thrown around in every single story about electronic voting does not take into account everything that happens to the code _AFTER_ we would be able to see it.

Anyway,
here is a link to a page on Electronic Voting:
Dr. Mercuri's Page on Electronic Voting

--xPhase

Paper vs paperless

[Ioldanach]

It is still possible to have a valid election, even with a closed source voting system. The key is to have the voting machine spit out a piece of paper where the voter can see the votes written down and then confirm them. It doesn't even have to be a paper the voter handles, it could be behind glass so the voter merely can see that what they voted for is on the paper. Then, in the case of a contested election, the checks can be made against paper as well as the bits. In a case where the ballots don't match, paper overrules the bits.

Granted, I think an open source system is the only sensible way to go, and the people writing them should be protected by copyright and patents, not secrecy. After all, if they're all required to be open, its going to be awfully hard to hide the source code you stole.

Re:Open Source != Secure

[WindBourne]

Security through Obscurity is not Inherently Evil.
In this case, it is.
With normal elections, the abilty to tamper with the results are minimal. The reason is that the votes do not end up in one place. In addition, you have several different parties counting at all times (typically Dems and Republicans). With this approach, it offers a single point of entry. It is possible for the votes to be rigged here either directly or indirectly. Somebody could offer 10 to 100 million dollars to the coder or even CEO within the company (keep in mind that some elections are spending a great deal more money to win them these days; apparently it pays off). Or since it may have a unknown opening, only the bad guys would know. Only the problem is that the bad guys would be a party member.
BTW, the orginal bribe may be to simply forget about an opening, rather than the introduction of one. Then the party would simply introduce something to manipulate it nationaly. At that point, they would manipulate close elections. Totally undectable.