Schneier on Security
A blog covering security and security technology.
« Security Information Management Systems (SIMS) |
| World Series Security »
October 25, 2004
Doghouse: Merced County
Merced County is in California, and they explained why they chose Election Systems & Software (ES&S) as their electronic voting machines. There are a bunch of vague selection criteria, but this one is quite explicit: "Uses 1,064 bit encryption, not 128 which is less secure."
I am simply too appalled to comment further.
Posted on October 25, 2004 at 3:25 PM
• 19 Comments
To receive these entries once a month by e-mail, sign up for the Crypto-Gram Newsletter.
... another fun one is the usage of the phrase "military-grade" used in many marketing spiels when it comes to explaining the level of security in relation to encryption...
From the web link:
"Software security will be accomplished by logic and accuracy testing of each unit and is open to the public for review."
Does that mean the software is open for public review, or the testing? I wonder who can get/see the results or whatever is open?
So what did you expect? These people don't know and don't care to learn. Just giving a 20-second sound byte is enough. Sounds good on TV doesn't it?
Check out the complete list of offenders at:
Only Merced and Tehama Counties are going to use "modernized" voting systems this election.
I wonder if anyone on the Merced review committee (Management Auditors, Information System Manager, and Election Managers and staff) be held directly responsible?
Hopefully someone in Merced will call for an official recount, per their rights, as stated in California State Elections Code: Section 15360, Section 15627, and Section 15630 (http://www.leginfo.ca.gov/cgi-bin/calawquery?codesection=elec&codebody=&hits=20)
"I am simply too appalled to comment further" (1064 bit versus 128).
It appears to me that by not commenting further, Mr. Schneier misses an opportunity to educate.
Isn't key length a factor in the security of an encryption algorithm?
Isn't it just another example of much bigger problem? That the cryptography (and security mechanisms in general - who understands e.g. how the connections are established and what are IP packets, not mentioning any security mechanisms on top of it) is indeed a very complex and difficult science area! It is not only about voting machines but also about courts and law (to get an example) when solicitors are not able to question expert witnesses without help of other expert witnesses. How can a judge understand what really happened when most of expert witnesses don't have a clue.
Cryptography has got a nice property (when properly implemented) that it prevents misbehaviour (no jury of common people do not have to solve disputes). But systems are getting complicated. Users are slowly starting to require privacy and a lot of scientists still believe that the law as we know it can be used to ensure this - let's just state policies and if someone violates them we can punish her/him. But can we?? We also can use cryptography but the systems will be so complicated that only a few will uderstand what is going on. And some say that privacy is an unreachable illusion. (Try traffic analysis and privacy in Google).
Yeah, what's my point?;-) - law enforcement and legal or justice system and maybe even guarantees of democracy as we know them can't keep the pace with the digital world if people move there activities there. It's like trying to stop Neo by bullets in the Matrix films.
Counterintuitive as it may be, perhaps this is one aspect of why cryptography should be locked down somewhat? Promoting it as "security" to the layman (and lay-reporter) is as we both know ridiculous, and that ends up in things like the marvelous quote above. People like that don't need to know about keys and algorithms, they just want to know it's "secure". I dunno, I've had to deal with more than one situation where people were insisting upon a certain item (be it a 1024bit key, or a firewall etc) simply because somewhere along the line they were told it provided "security". So while they're stuck on this one thing, while not understanding its irrelevance to the total solution, I end up having to educate and placate. Most of the time I'd rather just give them a black box, versus having them dissect a solution they don't understand in to keywords and specifications...
Strange, the "encryption" criteria Bruce talks about no longer appears on the page he linked to. The page appears to have been last modified on October 25,2004 at 21:45 UTC, about 40 minutes before I wrote this comment (the clock on this website appears to be somewhat strange, the time now is 22:25 UTC).
Nigel Tufnel has left Spinal Tap and is now an election manager in Merced county (This voting machine goes to 11!)
I am equally appalled by the ignorance displayed as knowledge on the linked page re: crypto as I am by the statement on the same page that reads "your vote is as important as your dollar". Any election official who thinks my vote is on equal standing with my money (and not far, far above it) needs to be fired and put back in 8th grade civics class where there may be a glimmer of hope that they understand the real value of my vote.
Seems to me that they had someone without much knowledge do up the web page. They put up obviouis rubbish and once pointed out they corrected it. Now the issues surrounding the reasons for picking the system are still valid, but lets not harp on about the webpage, but rather look at the issues.
ps. I'm from Ireland and we have gone through a process where the government tried to introduce electronic voting, but eventually had to back off, after looking to get approval from an independent commission. The commission refused to sign up, and the entire project got mothballed - after spending approx €50 million.
The page has been edited to remove the moronic crypto-key boast, but the new gobbledygook is even worse. I can't make out what they're TRYING to say with such gems as, "Did not use SMART card technology which requires the voter to open the ballot, instead it uses a Personnel Electronic Ballot (PEB) which is controlled by the poll worker." What are those things? I know what a "smart card" is but what is a "SMART card?" Is "SMART" an acronym for "Stupid Municipal Authorities Regulating Trash?" Because the people running the voting there would likely be doing less damage to the Republic if placed in charge of sanitation -- about which they could not possibly know less.
Davi - the key length determines how long it will take to brute-force the encryption. So a longer key is more secure against that one specific attack. BUT brute-forcing already takes a long time. On average you will have died of old age before you can brute-force a 128-bit key. So increaseing the length of time that one specific attack would take even more is not a good use of resources, especialy for something as time-sensitive as an election. WHen systems are hacked it is either by finding defects in the system or briding/blackmailing/tricking the people in the system to give you access, not by breaking the encryption.
To be fair, it could be that the rest of the system is very well designed, and the increased key length is there because Merceds marketing dept wanted a big number to put on their website. I would hope that a technology company hires marketing people who know about technology, but sadly this does not often happen.
I just started reading this site and would like to learn about this stuff. Can anyone suggest resources for me to understand what you're talking about? I'm a techno zero.
"On average you will have died of old age before you can brute-force a 128-bit key."
Actually, our solar system will have long died before you can brute-force a 128-bit key.
If every person on the planet (approx 5 billion) had a 1 Teraflop computer, all of which were linked together to form a massive distributed computer, and we were given a known plain-text, known cipher text scenario, and it takes only 100 calculations to test a key, it would take just over 100 billion years, on average, to brute force that single key.
This is why you worry about people who say "we use a 1,064 bit key, because 128-bit keys are insecure".
Of course, if they are talking about public/private encryption, and by "key-size" they mean "key-generating prime number size", then they'd have a reason to use >1000 bit prime numbers... but something tells me this is not the case....
Steven Alexander, I can't access that link, can you post a transcript of that conversation? I'm interested in what she said.
Schneier.com is a personal website. Opinions expressed are not necessarily those of BT.