I would like to keep discussion of NSA capabilities out of the BOF as well.

For the sake of having a focused and effective discussion I think we should
just stipulate the list of possible compromises and focus on what we can do
to address them.

Speculation as to the nature of the NSA capabilities is probably best done
in the bars and on mailing lists. What we discovered over the weekend
should cause a lot of the assumptions as to how PRISM works to be reviewed.

When we first heard of PRISM it was assumed that the data was being
voluntarily disclosed by Google etc. It now appears that it is plaintext
traffic on the Internet trunks that is being intercepted.


While it is true that the NSA probably can't do the intercepts without any
help, we can't build an Internet without intermediaries either. The
question at issue should be not whether an intermediary can default but
whether that default could be detected.

Ben Laurie's Certificate Transparency demonstrates a way to keep one set of
intermediaries under constant observation making default unlikely to
succeed unobserved. We need to start looking for equivalent schemes for all
the intermediaries we are forced to trust.

These include:

1) Generation of ECC Curves

2) Cryptographic Hardware, in particular SSL accelerators
2a) Kleptography as described by Motti Young to encode random seeds in the
modulus
2b) Disclosing private key through TLS covert channel

3) CAs

4) Standards organizations

On (4) the folk who I am suspicious of are not so much the direct
participants but the folk who I have never met that pop up and send me
private emails telling me that they agree wholeheartedly on some proposal I
am making and I should resist attempts to make some change.
_______________________________________________
perpass mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/perpass

Reply via email to