Rep. Billy Tauzin is circulating a draft of a bill that would restrict digital technology. One effect of the bill would be to mandate “broadcast flag” technology.
The bill has not yet been introduced.
Archives for 2002
Misleading Term of the Week: "Trusted System"
September 19, 2002 by
The term “trusted system” is often used in discussing Digital Rights/Restrictions Management (DRM). Somehow the “trusted” part is supposed to make us feel better about the technology. Yet often the things that make the system “trusted” are precisely the things we should worry about.
The meaning of “trusted” has morphed at least twice over the years.
“Trusted system” was originally used by the U.S. Department of Defense (DoD). To DoD, a “trusted system” was any system whose security you were obliged to rely upon. “Trusted” didn’t say anything about how secure the system was; all it said was that you needed to worry about the system’s level of security. “Trusted” meant that you had placed your trust in the system, whether or not that trust was ill-advised.
Since trusted systems had more need for security, DoD established security criteria that any system would (theoretically) have to meet before being used as a trusted system. Vendors began to label their systems as “trusted” if those systems met the DoD criteria (and sometimes if the vendor hoped they would). So the meaning of “trusted” morphed, from “something you have to rely upon” to “something you are safe to rely upon.”
In the 1990s, “trusted” morphed again. Somebody (perhaps Mark Stefik) realized that they could make DRM sound more palatable by calling it “trusted.” Where “trusted” had previously meant that the system’s owner could rely on the system’s behavior, it now came to mean that somebody else could rely on its behavior. Often it meant that somebody else could force the system to behave contrary to its owner’s wishes.
Today “trusted” seems to mean that somebody has some kind of control over the system. The key questions to ask are who has control, and what kind of control they have. Depending on the answers to those questions, a “trusted” system might be either good or bad.
Comments on White House Cybersecurity Plan
September 18, 2002 by
As a computer security researcher and teacher, I was interested to see the White House’s draft cybersecurity plan. It looks to be mostly harmless, but there are a few things in it that surprised me.
First, I was surprised at the strong focus on issues late in the product lifecycle. Security is an issue throughout the life of a product, from the initial conception of the product through its design, implementation, revision, use, and maintenance. The usual rule of thumb is that an ounce of prevention is worth a pound of cure – that attention to security early in the lifecycle makes a big difference later.
Despite this, the White House plan emphasizes remediation late in the lifecycle, over prevention earlier in the lifecycle. There is much discussion of intrusion detection, management, application of patches, and training of users; and not so much discussion of how products could be made more secure “out of the box.”
In the short run, these late-lifecycle methods are necessary, because it is too late to redo the early lifecycles of the products we are using today. But in the long run a big part of the answer has to lie in better product design, a goal to which the plan gives some lip service but not much concrete support.
The second surprise was the section on higher education (pp. 33-34 if you’re reading along at home).
Cybersecurity is a big mess, and there is plenty of blame to go around. You would expect the plan, as a political document, to avoid direct criticism of anyone, but instead to accentuate the positive by pointing to opportunities for improvement rather than inadequate performance. Indeed, that is the tone of most of the plan.
Universities alone seem to come in for direct criticism, having “many insecure systems” that “have been … exploited by hackers” thereby “[placing] other sectors at risk.” Contrast this with the section on “large enterprises” (pp. 19-22). Universities “have been” exploited; large enterprises “can be” exploited. Universities “place other sectors at risk”; large enterprises “can play a unique role in developing resiliency”.
But the biggest surprise in the higher education section is that there is no mention of the fact that computer security education and research are taking place at universities. The discussions of other stakeholders are careful to genuflect to those sectors’ worthy training and research efforts, but the higher education section is strangely silent. This despite the fact that many of the basic technologies whose adoption the report urges were invented at universities. (Think, for instance, of public key crypto.)
This general lack of attention to the educational system is evident elsewhere in the report too. Consider discussion point D4-12 (emphasis added):
How can government and private industry establish programs to identify early students with a demonstrated interest in and/or talent for IT security work, encourage and develop their interest and skills, and direct them into the workforce?
That’s what we do at America’s schools and universities: we help students identify their interests and talents, we encourage and develop those interests and skills, and ultimately we help students direct themselves into the workforce. On the whole I think we do a pretty good job of it. We’re happy to have the help of government and industry, but it’s a bit dismaying to see this identified as somebody else’s job.
Lessig/DRM/End-To-End Debate: Resolved?
September 17, 2002 by
Larry Lessig and I had a brief blog-discussion last week about the meaning of the end-to-end principle(s), and how end-to-end applies to DRM. The discussion continued off-line, and we ended up in pretty close agreement. Here is my version of what we agree on:
(1) End-to-end is not a single principle, but a cluster of related principles. Some are engineering principles, and others are policy/economic principles. It is good to be clear about what version of end-to-end you are using.
(2) The MPAA/Hollings approach does harm by forcing all computers to implement certain functions, even though those functions are not needed by all law-abiding network users. This violates the engineering end-to-end principle that says that functions should not be required unless needed by all.
(3) The MPAA/Hollings approach does even more harm by forbidding a great many non-infringing functions from being implemented at all. This offends both engineering and policy versions of the end-to-end principle, all of which favor giving end users flexibility in how they use the network.
(4) DRM is generally a bad idea, but some DRM systems are worse than others.
White House Cybersecurity Plan: On Life Support?
September 17, 2002 by
The White House’s “National Strategy to Secure Cyberspace,” initially slated for release on Wednesday, has been delayed, the Washington Post reports. This comes on the heels of the removal of some of the report’s proposals, and a leak of the draft proposal.
It looks like the report will end up as an eloquent expression of good intentions, coupled with few if any effective action items. Once the decision was made that the report would be changed to make all of the stakeholders happy, this result became inevitable. There are just too many agendas in play to reach any kind of consensus on this issue.
This is not necessarily a bad thing. The government can improve the security of its own systems, but there is little it can do to make ordinary non-government computing more secure. Our main problem is that the market doesn’t reward vendors for investing the large amounts of time and money necessary to build highly secure systems. There isn’t much the government can do to change that.
