Australian Voting Software Goes Closed Source

Scott Ritchie ended up delivered an angry rebuttal to Friday's OSCON presentation on the credibility of election software: What's strange is that his rebuttal came in response to a talk he himself had just delivered. Ritchie doesn't have a split personality, and wasn't simply playing devil's advocate. He found himself, though, in the strange situation of having agreed (as a last minute stand-in) to deliver a presentation he hadn't yet had a chance to read, provided by Dr. Clive Boughton of Australian software developer Software Improvement. (Boughton is also a Computer Science lecturer at Australian National University.) Between agreeing to fill in and arriving at the conference, Ritchie found that Software Improvement was switching its eVACS voting software from a Free, open source software license (specifically, the GPL) to terms "even worse than that on MS's shared source," and decided to do something about it. (Read more below.)

From Diebold's last-minute installation of uncertified software updates on its touch-screen election machines in California (leading to decertification of the company's machines in several California counties) to ethically troublesome relationships between politicians and the companies whose machines count the votes that determine their employment, the possible benefits of electronic voting seem swamped at the moment by objections (from simply prudent to caustically cynical) to its security and integrity.

Within the world of electronic voting, though, eVACS (for "Electronic Voting and Counting System") has been a rare success story both for open source development methodology and for the benefits that electronic voting can offer. The first generation of eVACS (running on Debian Linux machines) was developed starting in March 2001 in response to a request for bids by the Australian Capitol Territory Electoral Commission (ACTEC), and it was done on a budget of only AUS$200,000.

(The Australian Capitol Territory includes Australia's capitol city, Canberra, as well as surrounding suburbs and Namadgi National Park.)

Besides a respectable list of features driven by ACTEC's initial requirements (like support for 12 voting languages, and audio support for blind voters), eVACS has an advantage not enjoyed by many electronic voting systems: it's been successfully, uneventfully used to gather votes in a national election. The election in which it played a part went smoothly, and the eVACS system itself functioned as hoped.

This year, though, ACTEC asked Software Improvement to update the code for future elections, and Software Improvement decided to go them one better -- or, in the eyes of open source enthusiasts, one worse. The notes Ritchie was provided to deliver announced a change to the process under which the code is released; specifically, a switch from an open source license to something the company calls "controlled open source."

According to Software Improvement, simply releasing election-machine code under a liberal license such as the GPL is undesirable for two reasons: it means a loss of the company's intellectual property, and unfettered access could lead to a compromise of the voting system, if a determined cracker could find and exploit flaws in the code. (Software Improvement has not supplied any examples to show that this has happened, however.)

The company's use of "open source" would find little support from organizations like the Free Software Foundation or the Open Source Initiative. Software Improvement's idea of software openness is rather limited. Claiming that open source development is insufficient, even inimical to creating trust in election systems, the company now says that portions of eVACS's codebase will be released only to approved analysts, and in encrypted form, to enable viewing only for auditing purposes, rather than code contribution. Repeated viewings would be reported to the company, and only a limited number of views would be permitted before the code would self-destruct.

After delivering the prepared presentation, Ritchie took a few minutes to react to the changes it announced.

"Six hours ago, while I was reading through this on the plane," said Ritchie, "I was infuriated to read what it actually says."

Ritchie, though, is a computer-literate political science student at the University of California - Davis, and behind the Open Vote Foundation. He said he's decided to resume the project represented on that site, started with the intent to fork and bring to the U.S. the first generation, GPL'd version of eVACS.

"A long time ago, I read the first news report about Diebold, wondered why we didn't have open source election software for our voting machines. Eventually, I found out that Australia had apparently beaten us to it. It seemed like a good thing; the eVACS system was developed and released as GPL code, it was checked and rechecked by computer science people and all kinds of election officials. I said, 'Why don't we bring this to the U.S.? It's GPL, let's do it.'"

So he started the nonprofit Open Vote Foundation to bring the software to the U.S., specifically to California. Ritchie went to the meeting at the California Attorney General's office which resulted in decertification of Diebold machines in that state's 2004 election process, and his involvement in the fight against Diebold's secret-source voting machines is what led him to the open source eVACS; now he finds that the restrictions on the formerly GPL software are "even worse that that on MS's shared source. To call that open source is a bit dishonest."

"As of 6 hours ago," he said, "I've decided to start that again. It's not that hard; I mean how hard is it to say 'add one to this vote'? ... I remembered my old plan, and thought 'Let's take the old Australian code, fork it, and work from that -- and that is still an option. This is the great thing about open source software. If the old lead developer goes insane, you can always fork it, right?"

His opening line?

[Anonymous Coward]

"A Dingo Ate My Vote."

Re:When is civil disobedience justified?

[aussersterne]

The patriotic one. You know, the one with the good hair. The one who was a member of Skull & Bones. The who's strong on defense, wants jobs for working Americans, has beautiful, intelligent daughters that love him, and still believes in the American dream.

Oh, wait...

Re:When is civil disobedience justified?

[TrentL]

I'll destroy some machines if you let me blame it on you.

Re:When is civil disobedience justified?

[Anonymous Coward]

Yes, destroying the voting machines in not civil disobedience... turning them into a beowolf cluster to play Doom 3 on, now that is civil disobedience.

Re:Why not an AVM?

[provolt]

Why not model these voting machines after ATM's?

You know you're right. I wonder where we could find an ATM company? They have the knowledge and skills. I wonder where we could find one of those [diebold.com]. They'd be really good at it.

Re:His opening line?

[ackthpt]

"A Dingo Ate My Vote."

"To vote against the incumbent, hit the monkey!"

my tcl contribution to open source voting

[Anonymous Coward]

#!/usr/bin/wish

set kerry 0
set bush 0

canvas .c

label .l1
label .l2

button .b1 -text Kerry -command {incr kerry; .l1 configure -text $kerry}
button .b2 -text Bush -command {incr kerry; .l2 configure -text $bush}

pack .l1 .l2 .b1 .b2

When Voting Machines Go FPS

[MooseByte]

"Yes, destroying the voting machines in not civil disobedience... turning them into a beowolf cluster to play Doom 3 on, now that is civil disobedience."

Except you'd be in a tight deathmatch, frag your opponent with a headshot from behind for the final kill, but somehow you'd end up on the ground headless and they'd be doing the victory dance.

I'll start worrying when...

[Anonymous Coward]

...Dubya gets elected to a third term.

Re:The original American way...

[bytesmythe]

...you've still got that last solution: weapons. ... I say we vote with the tea.

Everyone get back! I've got a 25-pack of decaf Irish Breakfast, and I'm not afraid to use it!!

*dunk dunk dunk*

Umm... Did anyone bring cream and sugar?

If voting machines were Open Source...

[ngunton]

... then we'd have fifteen different interfaces that all do pretty much the same thing, but they would each have their quirks and none of them would do it quite right. The software would take five years to develop from scratch, and at the end of it we would have a huge virtual machine-based system that executes XVL (Extensible Voting Language), which is horribly complex and slow, but allows for very fancy voting platforms, in theory. But as a result, the old voting hardware will be too slow and limited to run it, so we'd need all new machines based on the latest processors. We'd also have to wait a while for all the drivers to become available, and the Debian Voting Project wouldn't release the code until it ran properly on *every* platform, including PDP11 and ZX81. Meanwhile the FireVulture project will aim to develop a super-lightweight version of the codebase that will be fast and sleek, but it will run into problems due to schisms in the team, caused by differences of opinion about whether the code should be LGPL, GPL or BSD license.

The eventual system will work very well and be extremely stable, but by the time it is in widespread use the developers will have started on Version 2.0, which is a total rewrite from the ground up (they now feel they understand the problem much better, and can see that the original API needs to be redesigned). So Version 2.0 is totally incompatible with Version 1.0, and much confusion ensues as States try to decide which "standard" to go with.

Meanwhile, Microsoft comes out fast and dirty with Microsoft Vote and although it doesn't work too well at first (version 1.0 has a glitch where everyone who's first name begins with "L" is deleted), it works "well enough" and with the buckets of money that MS dumps on the States for new MS-compatible hardware, they quickly gain dominance in the market.

The Open Source projects try to shift their focus to work with the MS hardware, chasing Microsoft's lead and running into a brick wall with the closed XML format that is encrypted and depends on hardware DRM to work.

Apple brings out the iVote, which is a small device that lets you simply plug into an Apple voting machine anywhere and vote quickly and easily. Plus, it works. And quite a few people buy it and rave about how great it is, but because only Apple is allowed to make the actual voting machines, very few of them get manufactured and as a result the iVote falls into betamax territory.

In the end, everybody uses MS Vote and complains about how closed it is, the Open Source crowd eventually gets their act together and comes out with a fantastic system that kicks butt but nobody cares any more, and that was that for the United States of America, thanks and goodnight.

Re:eVOTING? Bah thats simple!

[Anonymous Coward]

That has to be the sorriest excuse of a program I've ever seen.

What Diebold system did you steal it from?

Re:When is civil disobedience justified?

[sploxx]

Well, the answer is easy. A video camera to prevent vandalism ;)

Re:When is civil disobedience justified?

[Anonymous Coward]

Why vote for the lesser evil: Cthulhu 2004!

Re:eVOTING? Bah thats simple!

[TrixX]

This must be the code used 4 years ago: In the case of a tie, it gives the victory to Bush!

Re:When is civil disobedience justified?

[FLEB]

On Deibold letterhead, no less!