Talk:James H. Ellis

From Wikipedia, the free encyclopedia

The paper apparently lacked an author, so the ancestry of the idea may not be further traceable. The system conjectured in the paper (C43) was probably SIGSALY, which was actually built. It is not clear from Ellis' account whether he knew of it.

Sorry to cut this again, ww, but it likely wasn't SIGSALY. The idea mentioned was just theoretical, and it was the receiver added and then subtracted noise from the channel. In SIGSALY, it's the sender who adds the noise to the message, and then the receiver removes it — very symmetric key. — Matt 23:05, 24 Jun 2004 (UTC)

Matt, I keep trying... Sigh... The information about c43 was taken directly from Ellis' own paper and was added to buttress the previously removed material. On the question of whether SIGSALY was or was not c43 I have no information, but I seriously doubt there were two projects at Bell Labs with the same approach.

My point is they weren't the same approach. One was a one-time pad, the other was a purely theoretical exercise. — Matt 15:15, 25 Jun 2004 (UTC)

. American WWII crypto R&D may have been lavish in UK eyes (it certainly was in re the bombes, aside from political control issues) but it's not plausible that it was that lavish. That Ellis does not make the connection between the two (if they were two) is, as nearly as I can make out, accountable by secrecy re SIGSALY and corresponding compartmentalization. Open source was precisely 180 away from the way these folks operated and continue to operate insofar as I can make out current practice. Said speculation in this case, if fairly identified, is appropriate as semiexpert evaluation of the currently unknown probabilities.

As for your description of how c43 worked in contrast to SIGSALY, your comment is opaque. How would the receiver having added and then subtracted noise to the channel have impeded an Adversary? You've lost me. ww 14:48, 25 Jun 2004 (UTC)

Well that's the cunning bit. Alice adds her noise e to the line for, say, 30 seconds, while Bob sends the message M. Thus, the line carries the signal, M + e. So both Alice and Eve get M + e. But because Alice knows the noise e, she can subtract it to recover the message: M + e - e = M. Eve is stumped, however. SIGSALY, by contrast, requires both Alice and Bob to know e. So SIGSALY is very symmetric key crypto, while C43 foreshadows public key cryptography. — Matt 15:15, 25 Jun 2004 (UTC)

Matt, Much becomes clear that I'd speed read over. Got to take that Evelyn Wood refresher one of these days. Sorry about that.

This would be impossible to do in many electrical situations and I see why it went no further than a report. I'm still curious to know whether Ellis was aware of SIGSALY or not.

As for your suggestion that they are fundamentally different, well... From an information theoretic perspective (I can just smell Shannon in the background of both SIGSALY and c43, can't you), they're just about the same, cryptanalytically. From the perspective of who has the noise (ie, one time pad key material) and who does just what with it (ie, operational issues and security issues), there's a good bit of difference, of course. Very related concepts -- if you approach from a particular perspective. ww 18:12, 1 Jul 2004 (UTC)