Re: Newbie naive question, perhaps - - be kind

I'll just jump in here, even though both the other answers so far appear to
be pretty good.

"Arthur" <Art7@att.not> wrote in message
news:_v94f.436759$5N3.333722@bgtnsc05-news.ops.worldnet.att.net...
> I'm merely a potential end user of some form commercial encryption
> program primarily to protect some sensitive psychiatric case histories
> (as well as other files) from the curious eyes of the computer service
> kids.

This is suprisingly very important information, because it gives both the
value of the information and the attackers. The information indicates that,
unless you're psychiatric cases involve important individuals, the
information is actually fairly easily secured. More on this later.

> Although the 2 questions I have show my naivety in the field of
> cryptography and may sound simpleminded, this group seems to have the
> source of the best answers. Doing a bit of "homework," I found little
> help* in reviewing several months of posts here (too technical for me)!
>
> 1) Some of the commercial programs, such as Cryptainer, seem to suggest
> that their encryptions are essentially unbreakable: trillions of years
> of computer time would be required to break their Blowfish and AES
> encryption schemes in the program.

I took a look at the Cryptainer site
(http://www.cypherix.co.uk/cryptainerle/, assuming this is the same one you
were discussing), it does ring some of the bad bells. In particular their
Technical specs page says:
      ENCRYPTION ALGORITHM
     Cryptainer LE runs a 128 bit implementation of the Blowfish algorithm
in Cipher Block Chaining (CBC) mode with a block size of 64 bytes.

      CONFORMITY WITH STANDARDS
     Conforms to the following standards
      SHA NIST FIPS
      Key setting PKCS 5v2
      HMAC RFC2104 HMAC test vectors RFC2202
      IBS - ICS 36.080

Problems with this:
1) Blowfish operates on 64-bits, and Blowfish with CBC would operate on
64-bit blocks, not bytes as stated.
2) The given standards have nothing to do with Blowfish (except PKCS 5 can
be applied, and I haven't read IBS), this makes it substantially suspect.
3) IBS - ICS 36.080 (which I haven't heard of before) only appears
legitimately on Cypherix website, making it highly suspicious.

It could be that both of these are marketing errors (i.e. engineering told
them something, marketing interpretted it incorrectly), but they are
sufficient to make me question things.

> Yet they suggest long passwords or
> pass phrases to make hacking these passwords "more difficult." They
> encouraged written questions from potential buyers, but have not
> answered my question: isn't the program as weak as its weakest link,
> e.g. the password?? ("Wheel of Fortune" comes to mind: "I'll buy a vowel,
> please.)

Yes it is. The claims about 128-bit encryption, etc. are only telling you
about the strength of one link in the chain, the passphrase (don't trust
anything that can only use passwords) you choose will likely be the limiting
factor, and Diceware (http://world.std.com/~reinhold/diceware.html) is a
good way to approach the problem.

> 2) if these encryption schemes are so unbreakable, and commercially
> available, why haven't I heard news items describing "terrorists" and
> their use of unbreakable encrypted e-mail ("tomorrow at 10:15, Sidney,
> we light the fuse")?

It's because they still leak, even though the information is heavily
protected. Using the psych evals that you will be storing, it is fairly safe
to assume that they will be named for the individuals, given this
information, simply looking at the filenames (unencrypted) would be enough
to reveal who the contents deals with. In the case of terrorists this
information is the primary information (e.g. X has had direct contact with
bin Laden, question him on location), the actual contents are actually less
important.

> Can I assume that "hackability" or unbreakability
> is merely a matter of degree, and that the police or local computer
> repairman will in all likelihood be intrigued in my newly encrypted data
> files and e-mail and therefore try all the harder to see what's within?

You can assume that most of the programs out there are snake-oil, and that
even the good programs are often used poorly; both of these options quickly
lead to broken security.

The general, and long-standing, recommendation has been to use PGP wherever
applicable. PGP has been heavily analysed, and while occassionally there are
holes found, generally they are non-critical. Combining it with Diceware
passphrases results in a system that should be more secure than needed, and
quite usable. Just don't ever lose your passphrase, no one will be able to
recover it. There are other good solutions out there (and quite possibly
some superior), but PGP has been so substantially analyzed that I feel it is
probably the most trustable. For a free version GnuPG implements much of the
same portions (the difference won't matter to you).
                    Joe
Received on Mon Oct 17 20:48:21 2005