I'll confess that I sill read the paper edition of the NYT and I just read the story (http://bit.ly/bkjXIx) on anonymity. What struck me most was the section on the "Internet plumbing" and the idea that we must lock down the central plumbing so we can trust it. The idea of trusting the Internet is dangerously naïve. As I noted in http://rmf.vc/?n=IPGENI you can't find trust in the network any more than you are safe because you can trust the road system.
Trust is a social construct and one that doesn't fit well in simplistic reductionist paradigms. It's not just about the Internet, it's also true of the financial system and we've seen the fine mess we get when we let the naïve try to reduce it to a set of simplistic equations. Why do I trust Amazon and why (and how) does Amazon trust me? A simpler problem why do I trust Visa and why does Visa trust me? With Visa it's about money and the nice thing about that is that Visa can simply cover some risk. The bad thing is that we use the concept of "identity" to cover simplistic heuristics but our acceptance of the story allows us to continue to use Visa. Same goes for Amazon though they go further in assuring the integrity of the transactions. Note that SSL is not an in-the-network protocol. HTML and HTTP are not Internet plumbing as such. Sure, you can consider them that way but Joffe seems to be talking about the routers and doesn't make this vital distinction. The reason I used the title "Zombies Ahead" is the street signs that kids reprogrammed. If you think about it how different is a US-1 sign from routers on the net. Why do we trust those signs and not others. It's not as if there are no risks-people have died because of stolen STOP signs. This is a pragmatic issue. The term "web of trust" is a mechanistic reductionist view of a complex social and emotional phenomenon that doesn't deal with the difference between trusting people with your life and with your money. It doesn't permit us to laugh at Zombie signs while believing US-1 signs. If we are to fulfill the promise the Internet in making the Internet itself something that emerges from the edge than we have vital interest in avoiding any dependency central plumbing by learning how to come to terms with semi-trusted paths. To be sure the problem of trust among strangers at a distance is real and it's much easier to con people electronically than by replacing an ATM machine with a phony. Remember "con" comes from "confidence game". I'm not proposing the solution here but I am arguing that we need to recognize that the network is part of the real world. Procrustean measures that act as if the net is just a mechanical system in need of mechanistic solutions will likely do real harm. For that matter those who take their simplistic notions (such as a belief in unambiguous biometrics) into the real world are likely to leave us with a brittle dependencies that make it hard to deal with the ambiguities upon which our social fabric depends.
