[Cryptography] Preliminary review of the other Applied Cryptography

[Christian Huitema]

> It's not because I find DNSSEC personally offensive or anything, but
because
> (and this is a very, very cut-down version of the longer reasoning in the
> book) it's a huge amount of effort that achieves almost nothing.  

I think that you are restating a variant of the end-to-end argument. The
Internet architecture has grown rather complex. A simple transaction like
"loading a secure web page" will involve a bunch of actors, DNS resolvers
and servers, web proxies and caches, firewalls and routers. One approach to
security is to try to secure every little link of these complex chains.
Another approach is to treat all that as a hopeless soup, and just trust an
end to end verification to make sure that the system is behaving more or
less as expected -- which is pretty much the TLS approach.

I love the end to end approach, but it has two big issues. First, if the
soup is in fact poisoned, the only current end-to-end option is to refuse to
eat it, e.g., refuse to open the web page. Much better than being poisoned,
but leaves you hungry. We would need a way to "retry getting the page using
a different set up," and we don't have that with TLS today.

The second issue is that if we only do end to end security, the TLA's can
still go sift through the soup and learn a lot about what we are doing. 

-- Christian Huitema