Mar. 22nd, 2006

flewellyn: (Default)
As my bio says, I am a computer person, in much the same way I am an oxygen-using person. Computers and programming and technology in general fascinate me, so naturally I care a great deal about what goes on in the computer world.

One of the up-and-coming ideas, currently being advanced by the RIAA and MPAA, those same paragons of consumer advocacy that brought us the Digital Millenium Copyright Act, is Digital Rights Management, or DRM. Also known in the free software community as "Digital Restrictions Mangement", this technology is intended as a set of hardware-based cryptographic tools which will allow the computer to verify that the user is actually authorized to use certain content. In other words, a CD or DVD would be encrypted, and unless your DRM chip could verify that you had a legit copy, you couldn't use it.

Obviously, this has some people rather upset. It's a very domineering tactic, and among other things, makes it so that you no longer own the computer you buy, or the CDs or DVDs. You would use them only in the way that large, monolithic corporate interests want you to, and they could have complete control over what you were allowed to view and listen to on your computer.

At least, that's the idea. There is one teensy little problem with DRM: there is no way it could actually work.

I am not a cryptographic expert (amateur enthusiast at best), but to me the problem boils down to one of key distribution. It's very easy to build cryptographic tools, open source or not, without compromising the security, because knowing the cypher algorithm does not necessarily help you break the encryption by itself; you need the keys, or a means of producing those keys.

Traditional cryptography, also known as symmetric key cryptography, has one key which is used for both encrypting and decrypting messages. The problem there is, symmetric key cryptography has the small problem that the same key is used for both encrypting and decrypting, which means that key has to be secret at both ends. So, if the two ends are far apart, you have this problem of sending the key to the other person; how do you get it there without it being intercepted?

The newer forms of asymmetric cryptography, called "public key" cryptography, were designed to alleviate this problem. This splits the single key into TWO keys: one for encrypting, and one for decrypting. Having the encrypting key does not allow you to decrypt messages that were encrypted with it; only the decrypting key can do that. The idea here is that, basically, you distribute the encrypting "public" key far and wide, while keeping the decrypting "secret" key to yourself. Thus, anybody can encrypt a message to the holder of the secret key, but only the holder of that key can decrypt those messages. This is a very elegant solution to the basic problem of key distribution, assuring that the messages are kept private for the intended recipient.

But here's where the problem comes in. Since it was invented, cryptography has always assumed that, while the message you encrypt may be intercepted by untrusted parties, the person you intend the message for can be trusted. DRM is about locking down content, not about ensuring privacy of messages; it flips around the usual case in cryptography by distrusting the recipients. In order to make sure content is only decryptable by authorized people, you'd have to ensure that only those authorized people had access to the keys needed to decrypt. If it's public key, asymmetric cryptography, you'd have to distribute those keys somehow to the authorized people.

And any way you do that, whether by embedding them in the content or by separately enabling them, nothing prevents those authorized people from redistributing the keys to other people! Since, in DRM, the recipient is not trusted, this means that key distribution becomes a catch-22: if you don't distribute keys, then nobody can read your content, and they have no reason to buy it; but if you do distribute keys to purchasers, then they can redistribute them, everyone can read your content, and they have no reason to buy it.

Locking down the spec for DRM, requiring it in ROM, building it into the operating system...none of these things will help. The core issue is, the keys have to exist on the recipient's machine in order to read the content, and once there, they can be discovered and put to "unauthorized" use. Granted, 9 out of 10 computer users may not have the knowhow to do this, but it only takes one or two to discover the method and propagate it. DRM as a content restriction method is doomed to fail.

The technology could be put to legitimate, useful purposes, mind you, as a way for the owner of the machine to secure it against running untrusted binaries; this would allow you to, for instance, "sign" all the programs you want to be runnable with a little authorization key, and then the computer would simply refuse to run any unsigned program. This would be a huge boon for securing a system...but for the system's owner, not some third party. That, I think, would be a very good use of built-in cryptography in a computer, and should be encouraged.

But the stone cold fact is, if you have physical access to a computer, that computer cannot be secured against you. And frankly, if you're the one who bought it, I think that's a good thing.


flewellyn: (Default)

July 2014

13141516 171819

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Sep. 21st, 2017 03:45 pm
Powered by Dreamwidth Studios