Comments from Frankston, Reed, and Friends
Saturday, August 03, 2002
DPR at 9:07 AM [url]:
Scapegoating the wrong people
"Security czar" Richard Clarke points the finger of blame at the wrong people. He blames the users, software manufacturers, and designers of the Internet for its security risks. This "patriot" thinks that by loudly proclaiming a Big Lie, he can help create a secure infrastructure.
Since the NSA itself, in 1976-77, blocked a fully worked out end-to-end encryption approach created at MIT for TCP, we might want to point the finger elsewhere.
Perhaps at the government itself.
Quite a number of us who participated in the early Internet protocol design were from the computer security research side, and did our best to make the Internet architecture secure from the start. But the NSA (I am told) told DARPA that any attempt to introduce security mechanisms into TCP/IP's architecture would be viewed very negatively. (This happened at about the same time that Rivest, et al. received a mysterious threatening letter from a senior military official claiming that their work on the RSA cipher must be stopped immediately).
Despite this, the TCP and IP designers insured that the architecture of TCP and IP were such that end-to-end encryption and other crucial protections , along the lines of the banned proposals, could be introduced at any point. I know. I did a lot of this work.
And in fact, IPSEC was later invented along similar lines, as an option. But part of the difficulty with implementing IPSEC is that it is too late - popular fads such as NAT and stateful inspection firewalls have been deployed too widely. Firewalls (which provide faux security at best) make real security much harder to deploy, because they require that end-systems expose too much information in the clear. Truly secure protocols (even IPSEC) don't work very well with firewalls.
Later, in the '80's and '90's, when I was a VP at Lotus, my friend Ray Ozzie wanted to put end-to-end RSA encryption in Lotus Notes. Again the government required that the civilian users get a weakened form of encryption. And the government blocked PGP. And more recently, the government called for Lotus to introduce security holes in Lotus Notes that would weaken users' protection.
In one respect I agree with Mr. Clarke - it is important to have good security in the Internet. But as a representative of the gov't security community, he should stop pointing fingers, because the real finger needs to be pointed back at himself.
Many, many of the folks who worked on secure systems architectures in the '70's foresaw these vulnerabilities in the so-called "civilian sector" and called them to the attention of policymakers, and also urged solutions, only to have government interests block such proposals. I know, I did. And I still do, most recently arguing for a "cryptographic checksum" in SIP and other new protocols that allows end systems to detect unauthorized modifications to data in the middle of the network, and for use of encryption-based approaches in wireless networks to prevent selective disruption of civilian traffic by other civilians for personal gain or sabotage purposes.
It makes me more than a little angry to see a public figure who works for the government implicitly blaming the very people who pointed out the problem and tried to solve it.
The reason I don't work in the security field (despite my recognition of its importance, and my own early work in secure protocols) is that the governments (US and others) made it impossible to do good work. I'm sure that others who might have made contributions, or did make contributions, made the same career decisions. To do good work in the field when your own government opposes its use is quite discouraging. To see ideas that might benefit the public security blocked at every turn by military and corporate interests. Ask Ed Felten, Ross Anderson, Whit Diffie, Ron Rivest.
If Clarke would start taking responsibility for our own government's failure to take security of our information economy seriously, and ask for help, rather than blaming the victims, that would be a start. But he strikes me as the kind of "leader" who thinks that he can acquire loyalty by torturing his troops.
Wednesday, July 31, 2002
DPR at 2:35 PM [url]:
Learning in 3D
I just read Andrew Glassner's Other Notebook. This is the the most fun book I've read in years! It's a sequence of explorations, using 3D computer graphics, of some very interesting ideas, ranging from Celtic knot designs to the hidden structure of motion blur in movies to quantum mechanics. A beautifully illustrated book, but one where I learned something new on every page. If anyone ever says that 3D graphics is only "eye candy" - show them this book. If only it were a Dynabook!