🚀 go-pugleaf

tcmay@netcom.com (Timothy C. May) writes:

>- Can encryption-within-encryption be detected? How?

My understanding of information theory is that there isn't any way to
prove that a given stream of bits _doesn't_ contain a message. You can
prove that it has various random properties, but maybe the existence
of the random properties is what's being modulated to transmit the
information.

>- If noise is sent to foil traffic analysis, how will the
>Denning/Rivest proposal deal with it? (The general issue that a person
>can always shrug and say he was sending random bits. Are there
>technical means of proving otherwise?)

In "Codebreakers" Kahn talks about the message censorship regime
enforced during World War II on all international communications to
and from the US. They restricted the commercial codes that could be
used, arbitrarily replaced blank sheets of paper with blank sheets
from other sources, and generally blocked messages that they couldn't
otherwise read. "Random bits" as it were, weren't allowed in outgoing
international traffic. This was, after all a "war emergency."

>- Psychologically (further afield from sci.crypt's charter), what will
>be the effect on public confidence in such things as digital
>signatures and timestamping if it is known that in principle the
>government (or others) could obtain the private keys?

There's something fundamentally frightening about having a third party
appear from nowhere with knowledge of things you thought were secret.
If it happens seldom, or under really restricted circumstances, people
in general will probably tolerate it. On the other hand, if they're
regularly faced with unexpected disclosures or just inconvenienced by
the mechanisms that support eavesdropping, then they're more likely to
complain, IMHO. I submit as evidence the extent to which people trust
organizations like banks and credit card companies.

Rick.
smith@sctc.com            arden hills, minnesota