Re:Zeroing in on unbreakable computer security
by kens on Wednesday August 21, 2013 @01:32PM
(#6251951)
A few years ago researchers found a way to eavesdrop on a commercial theoretically-secure quantum key distribution system. The trick was to replace the polarized light pulses with bright pulses that overwhelmed the sensors, sidestepping the whole quantum business. The point is that even if a system is theoretically unbreakable, it's only as secure as the implementation, which may have nonobvious flaws.
Re:Zeroing in on unbreakable computer security
by Fortaymedia on Wednesday August 21, 2013 @11:00AM
(#6250761)
It starts off saying that it is unbreakable i.e the title, then a little way through goes onto say that its virtually impossible then towards the end it mentions that the Americans 'probably' wont be able to break it. It sounds like its just a exaggerated title to get people to click :/ but the theory behind quantum encryption is amazing.
As long as the key is securely shared between the two parties, this scheme provides perfect secrecy, and the proof of this is trivial (check the link).
The main issue (and the reason OTP is very rarely used) is the key distribution. Quantum communication can solve this however by providing a channel which can be proven (to an arbitrarily high probability) to have not been eavesdropped or intercepted. Sadly, this doesn't scale at all, but it does give us a perfectly secure scheme to build up from.
Re:Zeroing in on unbreakable computer security
by betterunix on Wednesday August 21, 2013 @01:38PM
(#6251990)
Another reason one time pads are rarely used is that there it is easy to screw them up, since the key material must never be reused. In practice this is often violated.
Re:Zeroing in on unbreakable computer security
by sirkneeland on Wednesday August 21, 2013 @12:20PM
(#6251387)
"perfect" is a dangerous word. Whether it's someone saying a "perfect" product or "perfect" society or "perfect" anything, I tend to run in the opposite direction. What can I say, I'm a digital Burkean..
Re:Zeroing in on unbreakable computer security
by gcommer on Wednesday August 21, 2013 @12:38PM
(#6251534)
Philosophically I may agree, but in this case "perfect secrecy" is a mathematically defined term with a specific definition.
Basically, it is mathematically impossible for any encryption scheme to ever provide better data secrecy than the OTP, so I think calling it "perfectly secret" is perfectly alright :)
Re:Zeroing in on unbreakable computer security
by comex on Wednesday August 21, 2013 @11:59AM
(#6251207)
Just an obligatory reminder that as cool as quantum cryptography is, it's very unlikely that anyone (including NSA) can break today's run-of-the-mill classical encryption, and an XOR cipher plus a hardware RNG (such as, unless you're somewhat paranoid, the ones built into modern Intel processors) is essentially completely unbreakable.
Re:Zeroing in on unbreakable computer security
by draz on Wednesday August 21, 2013 @12:02PM
(#6251235)
true, unbreakable now, but probably breakable in a few years....
You store whatever SIGINT you pick up now, and understand it in a few years. A lot of it will still be valuable in 5,10, 15 years
Re:Zeroing in on unbreakable computer security
by comex on Wednesday August 21, 2013 @12:13PM
(#6251329)
True. But
(a) These quantum links are physical cables that have to be laid between sites. If you lay your own normal cable rather than using the Internet, you probably won't be intercepted either. Or...
(b) While shuttling around massive XOR pads on hard drives is probably too annoying for most individuals, it's quite feasible for the kinds of organizations that could buy this stuff and a whole lot cheaper, unless they have enormous quantities of data to transfer; though the scheme can then be broken by passively copying the pad rather than continuously actively subverting the quantum link, I'm not sure this makes much difference in practice.
Re:Zeroing in on unbreakable computer security
by derekp7 on Wednesday August 21, 2013 @01:31PM
(#6251948)
A bit off topic, but I noticed that the title that was originally submitted for this story is the HTML title of the linked article. It was then changed to the article's <h1> title. When submitting articles to HN, which title is actually the preferred one to submit?
Re:Zeroing in on unbreakable computer security
by txutxu on Wednesday August 21, 2013 @12:30PM
(#6251478)
I think I've write fun things in this life (in the sense of code to do stuff). But I can't imagine how the "job" of persons behind all this stuff could be... mmm...
Re:Zeroing in on unbreakable computer security
by hownottowrite on Wednesday August 21, 2013 @11:18AM
(#6250914)
The CNN Headline is sort of misleading... The International Conference on Quantum Technologies was held in Moscow, but the company profiled (ID Quantique) is based in Geneva.
References, which explain this much better than I did: http://www.vad1.com/lab/full-eavesdropping-2011/ http://www.nature.com/ncomms/journal/v2/n6/full/ncomms1348.h...
http://en.wikipedia.org/wiki/One-time_pad
As long as the key is securely shared between the two parties, this scheme provides perfect secrecy, and the proof of this is trivial (check the link).
The main issue (and the reason OTP is very rarely used) is the key distribution. Quantum communication can solve this however by providing a channel which can be proven (to an arbitrarily high probability) to have not been eavesdropped or intercepted. Sadly, this doesn't scale at all, but it does give us a perfectly secure scheme to build up from.
http://en.wikipedia.org/wiki/Quantum_key_distribution
Basically, it is mathematically impossible for any encryption scheme to ever provide better data secrecy than the OTP, so I think calling it "perfectly secret" is perfectly alright :)
(a) These quantum links are physical cables that have to be laid between sites. If you lay your own normal cable rather than using the Internet, you probably won't be intercepted either. Or...
(b) While shuttling around massive XOR pads on hard drives is probably too annoying for most individuals, it's quite feasible for the kinds of organizations that could buy this stuff and a whole lot cheaper, unless they have enormous quantities of data to transfer; though the scheme can then be broken by passively copying the pad rather than continuously actively subverting the quantum link, I'm not sure this makes much difference in practice.
http://www.idquantique.com/contact-us.html