There is some recent discussion of some sort of "multilevel" internet, with e.g. a level defined for banking in which access is only granted after special authentication with "true names", special credentials, and an explicit consent to permit packet searches:

Former CIA Director: Build a new Internet to improve cybersecurity - Nextgov

"The most important value of a dot-secure domain is that, like dot-gov and dot-mil, now we can satisfy consent under the Fourth Amendment search requirements for the government's defenses to do their work within that domain, their work of screening for attack signals, botnets and viruses."

searches of the .gov domain are conducted by the Einstein program, an intrusion prevention and detection system under the direction of the Homeland Security Department that monitors only federal traffic for signs of unauthorized access.

Has anyone seen any technical proposals for how to design something along those lines which might actually result in improved security? Or is this just bad reporting or clueless bureaucrats speaking?

This is sometimes associated with a ".secure" top-level domain, but without a lot of other mechanisms that of course wouldn't do anything to help security.

And assuming that the websites in this .secure level of the Internet all use SSL all the time, what good would it do for the government to have legal permission to search packets?

Alternatively I guess I could imagine some sort of big semi-public VPN to which access is controlled - has that been explored? But finding mechanisms that are resistant to phishing or DDoS still seem pretty hard.


My guess, from reading the article, is that the envisioned network is mostly the Internet as we know it, except that thanks to some trickery with regards to the (US) constitution, the presence of a ".secure" at the end of the name allows a (US) agency to deploy various eavesdropping measures. The proposed system should allow only "authorized" users, which probably means that whoever wants to operate a ".secure" machine will have to give a copy of its SSL private key to the CIA/NSA/whatever, or something like that. They could call it "Revenge of the Clipper Chip".

There is nothing really technical here; only legal. The whole things is impressive in its americanity (a US legal gimmick for the benefit of the US agencies, with "consent" stemming from the word "secure" -- that's not .sécurisé or .seguro or .sichern or .terjamin).

  • 2
    That would be truly awful. I'm guessing that the certificate you present with your "true name" would be based on granting explicit permission for some sort of surveillance, perhaps just by the bank you're connecting to at the other end of their SSL cert, and permission for them to share that with the authorities. Also, if they make that a requirement for a .secure domain registration (from whatever part of the world) then they could also maybe argue that they could search all packets to .secure hosts??
    – nealmcb
    Jul 11 '11 at 19:06

Apart from the fact that this is clearly only meant as a surveilance mechanism:

How would the user know and check that he is on a ".secure" domain? How is this any different from checking that you are on bank.com instant of badbank.com? You have the exact same problems on client side.

If you now think you can fix this by OS, hypervisor or client-wise enforcement, user education, or by placing big red warning dialogs: This has all been done before. It is either not practical or of very limited practical impact.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.