purloined letter

uri at watson.ibm.com uri at watson.ibm.com
Tue Jan 5 09:55:48 PST 1993

Edward Bertsch says:
> ->calling attention to the message BECAUSE it is encrypted? "If he went to
> ->the trouble of coding it, there MUST be something in there!!" Granted that
> ->if everyone begins encrypting, this problem will vanish... are there
> ->practical solutions in the meantime? (eg, Codes that look like plaintext?)

Well, my opinion is - the only way to go is to SHORTEN the transition
period. Switch to all-encrypted e-mail ASAP.

> a good point indeed.  I know of no software that works the way it seems
> you would like.
> ............................................This sounds like a VERY
> difficult problem, and one that is not likely to be solved any time soon
> (in the sense of having this be done 100% by software).

Agreed. Theoretically possible - practically infeasible. Plus imagine
message size... Plus it depends on how clever a scanner-program can
be - if eavesdroppers have enough CPU power, they could check for
the "validity" as well, i.e. right word sequences, not just

> Another option would be to have the message fit the letter-frequency,
> letter-pair frequency, etc... that 'normal' messages have.  The idea
> here is that messages may be scanned for unusual (i.e. non-english text)
> properties in this regard, and then scanned further by humans and/or
> computers in the order of their 'interestingness'.  So to defeat this
> kind of scanning, your 'secret' message should 'appear' to be a 'ordinary'
> message.

Again, it will, or will not work, depending on how smart the scanning
program is. There's no reason why it can't detect, that your letters
don't form valid English (German, Swedish, Arabic, whatever) words,
*or* the words don't form valid sentences...

I repeat - the surest way is to get over the hump sooner.
Uri         uri at watson.ibm.com      scifi!angmar!uri 	N2RIU

>From cypherpunks-request  Tue Jan  5 10:49:28 1993

More information about the cypherpunks-legacy mailing list