Features

net.wars: Security versus security

If there is one subject that dominates the thoughts of everyone who uses the Internet – which is most people and businesses these days – it's security. The biggest need of everyone but law enforcement is to improve it. Yet the FCC would rather poke holes in it.

Most of the time when we talk about privacy, we talk about the trade-offs that have to be made between privacy and convenience, or the trade-offs between security and privacy. It's inconvenient for me, for example, that my British cable company is so unable to trust me - after I've given them my name, address, phone number, and password - that they won't reveal to me my account number so I can pay my bill online. But they are protecting my privacy (they say; who really knows?).

Equally, it's inconvenient for law enforcement agencies everywhere if they can't wiretap any or all of us at will; and if we object they can complain that their needs are a matter of national security and we're just trying to hold onto our privacy (what's left of it).

But what happens when national security has to compete against national security?

In a paper (PDF) published a couple of months ago, Susan Landau, a distinguished engineer at Sun, examines the problem posed to the Internet's infrastructure by the 1994 Communications Assistance for Law Enforcement Act (CALEA), specifically its application to Internet telephony (or, in the catchy way we have with acronyms, VOIP). In August 2005, in answer to a request from the FBI, the FCC ruled that VOIP must comply with CALEA. That means that all VOIP services must have built into them the ability to turn on a wiretap at any time, at the drop of a court order.

The point that Landau – and, in a jointly signed letter that echoes her paper, network security expert (and Usenet inventor) Steve Bellovin and cryptographer Matt Blaze – makes is that it's one thing to demand that VOIP applications comply with CALEA, but it's quite another thing to require that the Internet be completely re-engineered down to the very genetic makeup of its protocol stack, which is what her research indicates would be required.

"Applying CALEA to VOIP," she writes, "requires embedding surveillance technology deeply into the protocol stack." This is for a simple reason: if you put the wiretapping capability where it logically ought to go, given the Internet's layered design, in the application layer, it is impossible to ensure that someone won't be able to write an application to bypass it. We've seen that pattern all sorts of times over the Internet's short history, from Usenet's alt hierarchy on out.

 The IETF considered years ago whether it made sense to try to build a "wiretapping protocol", and concluded that it didn't. That doesn't make them radical lunatics who want to aid terrorists; it just means that after due consideration their technical expertise told them that any time you try to create a hole in a security system you have . . . well, a hole. It's kind of like the heroine of Nora Ephron's novel Heartburn and capers: after much experimentation she concluded that anything that tasted good with capers in it tasted even better without capers in it. But these capers can be exploited not only by the people who built it but by all sorts of other people besides.

"This is in conflict with the goal of freedom from security loopholes," the IETF concluded mildly.

That's what's so weird about these proposals. As I said earlier, this is the one subject that dominates the thoughts of everyone who uses the Internet - security. The biggest need of everyone but law enforcement is to improve it. So whose side is the FCC on, really?

There are the usual points to make. Those of us who do not live in the US will most likely still feel the effects of CALEA as manufacturers design their equipment to conform to its requirements. If the Internet is actually re-engineered – if anything still works afterwards – everyone outside the US will have to decide whether to adopt the new standard, build a gateway/firewall with huge signs saying "Here there be dragons", or chop the Internet in two.

More, the requirements the FCC laid down for traditional telephony don't fit the Internet, whose design is exactly the opposite (smart ends, simple network) of the Public Switched Telephone Network (simple ends, smart network). The number of organisations opposed to this thing is nicely laid out by the Center for Democracy and Technology, which is challenging the ruling. The opponents include all those radical privacy groups like, you know, like the US Telecom Association or the Satellite Industry Association.

It is not unprecedented for governments to conclude that sacrificing some law enforcement abilities is worth improving security for the masses. Around the time CALEA was originally passed, the big, hot, passionate topic on the Internet was the freedom to use strong cryptography. Eventually, over the protesting bodies of law enforcement, it became clear that what public-key cryptography folks had been saying was right all along: for electronic trade to flourish, the masses had to have cryptography. The peacetime uses outweighed crypto's use as a military weapon.

Now, twelve years later, we have yet to hear of a case in which law enforcement was unable to function because of the widespread use of cryptography (although, to be fair, most traffic is not encrypted). We do, however, have a beautiful example of what unofficial wiretapping can do when security isn't good enough to prevent it.

Tags: FCC, security, cryptography