posted Jul 24, 2001

Linux: The electoral test that pencil and paper meet Misc.7/24/2001; 12:28:18 PM 'When Carol Boughton's Canberra consultancy, Software Improvements, won a $200,000 contract to provide an electronic voting system for the ACT's October election, it was critically important her team got the technology right.'"ACT" stands for "Australian Capital Territory". This is still a poll-based scheme... you have to come in and vote, it's not a remote thing. (This is good.)What's interesting about this is the system is all open source: 'The only platform that provided robustness and voter confidence was GNU Debian Linux, with all source code released under the General Public License (GPL).''Douglas Jones, an associate professor of computer science at the University of Iowa, in testimony in January on voting technology before the US Civil Rights Commission, adopted the axiom, "trust no one"...."Truly open source systems are valuable, but they pose threats, too, because anyone can get and modify the code."...'I chopped out quite a bit there; please read the article for full context. I wanted to point out that the phrase "anyone can get and modify the code" is deceptive, and probably doesn't accurately convey what the professor actually said. It may be true that I can download this voting software, make some changes, and compile it, but the effect that would have on the ACT voting results would be precisely bupkis. I would still need to get those changes into the real voting system, which should be virtually impossible, regardless of the details of the actual vote counting software.The machines will be physically secured at the polling location and the counting location (I hope!), and as long as all communications between those two locations are adequately and competently secured (which should be easy to do with something like ssh & certificate authentication), it will be very difficult to affect the system remotely, almost regardless of any weaknesses in the system. That leaves only local exploits... and there are other things that could be done to detect the effects of that. If I were designing this system, I'd put some paranoia checks into the counting system. Is one of the polling computers changing its tune? Is it suddenly registering too many votes too quickly?So, while in a vague theoretical sense, open source voting software might allow someone to discover holes in the system and exploit them, there are still huge (theoretically insurmountable) practical difficulties in exploiting these bugs, and even bigger ones associated with not being detected, assuming competent system design and administration. When considered against the very practical and real problems proprietary voting systems have, with their opaqueness and the power being handed over to the vendors of the system as a result (who could know if they were fudging the vote by a percent or two?), open source is the clear winner for voting software. It may not be perfect, but if you insist on using software, there's no reason to go with closed source.


Site Links


All Posts