home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky sci.crypt:6592 alt.security.pgp:480
- Newsgroups: sci.crypt,alt.security.pgp
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!n8emr!colnet!res
- From: res@colnet.cmhnet.org (Rob Stampfli)
- Subject: Re: Zimmermann's responses to Sidelnikov's PGP critique
- Message-ID: <1993Jan9.172046.17709@colnet.cmhnet.org>
- Organization: Little to None
- References: <1993Jan8.173701.8858@ncar.ucar.edu>
- Date: Sat, 9 Jan 1993 17:20:46 GMT
- Lines: 38
-
- In article <1993Jan8.173701.8858@ncar.ucar.edu> prz@sage.cgd.ucar.edu (Philip Zimmermann) writes:
- >> - the sequence of random numbers has strong prevalences on
- >>bytes (up to 0.05 ... 0.1 on material of 10000 byte) and strong
- >>correlation dependence between contiguous bytes;
- >
- >Really? How so? What does "strong prevalences" mean? Is he talking
- >about the internal random number source in random.c, used for making
- >RSA keys? Or is he talking about the output of the IDEA cipher? In
- >either case, evidence should be presented that allows others to
- >reproduce his results. The random.c code for getting randomness from
- >keyboard latency has been tested pretty well, and it uses MD5 to
- >enhance the raw randomness from the keyboard timings. It looks
- >pretty good to me. Does the IDEA cipher running in CFB mode output
- >text that appears nonrandom? This is disturbing, if true. Biham and
- >Shamir have thus far not succeeded in finding weaknesses in the IDEA
- >cipher. Perhaps Dr. Sidelnikov has found one. I'd like to see some
- >evidence of this claim.
-
- It is not my intent to add fuel to any fire, but this seems like a good
- springboard to voice a potential concern I have had with the Unix port of
- pgp. Pgp seeds certain random number vectors by asking the user to type
- some characters at the terminal and then measuring the times between
- keystrokes. I know Phil gave some thought to this for pgp1.0, and it
- probably works well in a DOS environment. However, some flavors of Unix do
- not have the ability to measure time to the same granularity as DOS, and,
- combined with the multiprocessing environment of Unix, the times gathered
- may not be representative of the typist's true signature. I have not looked
- at the code, but, empirically, I have gotten quite a few instances of '?'
- when performing this ritual, which I believe means the number was thrown
- out because it was too small.
-
- Has anyone indeed paid any consideration to the randomness of the numbers
- so generated under Unix? It would be comforting to know that someone has
- looked into this, and even more comforting to know they concluded that
- this is, indeed, not a problem.
- --
- Rob Stampfli rob@colnet.cmhnet.org The neat thing about standards:
- 614-864-9377 HAM RADIO: kd8wk@n8jyv.oh There are so many to choose from.
-