🚀 go-pugleaf

In article <1992Dec15.012350.11243@shearson.com>, pmetzger@snark.shearson.com (Perry E. Metzger) writes:
|> 
|> Summary for readers with little time: Uri Blumenthal insists that
|> stripping the ----BEGIN PGP lines from PGP messages makes things more
|> "secure" by eliminating the "proof" that you are using PGP. In this
|> reply to his most recent message to me, I explain precisely why its
|> trivial for an eavesdropper to detect that PGP messages are going by
|> in EMail.
|> 
|> uri@watson.ibm.com (Uri Blumenthal,35-016,8621267,) writes:
|> >From article <1992Dec11.030807.29118@shearson.com>, by pmetzger@snark.shearson.com (Perry E. Metzger):
|> >>>Why does PGP has those ugly lines "----BEGIN PGP...."
|> >>>and so on? PGP-2.1 is much better than PGP-2.0. Let's
|> >>>make it really good now -  GET RID OF THOSE BETRAYING
|> >>>TAGS! NOW!

I believe Uri is making a practical statement, not a scientific one.  It would
sure be hard to dissuade a judge and jury once that stuff was seen.

|> >> 
|> >> I guess you never read the docs. Those "betraying tags" have a purpose
|> >> -- they allow the system to automatically find the beginning and end
|> >> of messages. 
|> >
|> >I guess you
|> >	a) are deprived of sense of humor;
|> >	b) have too little experience with real crypto...
|> 
|> Please carefully re-read what I said, and what I say in this message.
|> 
|> >> You can feed mail messages into PGP without even
|> >> stripping the headers. 
|> >
|> >Ge, thanks for explaining! And I was sure they are there
|> >just to attract feds' attention... Oh, my...
|> >
|> >> Its all very well engineered, and the feds can
|> >> tell you are using PGP anyway by looking at the magic numbers in the
|> >> Radix 64 text.

But, that would take smart Feds.  If they "know" or guess it is PGP, they
will surely look.  And, I would sort of plan on it.  But, they may actually
overlook it if you don't make it as plain as the nose on your face.
Theoretically, it makes no difference.  In terms of practical
cryptanalysis, Uri is right; give 'em a chance to go wrong.  At the least,
they may overlook _some_ of your encrypted traffic, which is worth something.
With this, one merely does _grep_ or MS DOS _find_ and all the encrypted
traffic comes tumbling out.  Do you really wish to make it so easy?
 
|> >
|> >And what sense would a hexadecimal number prepended to an
|> >encrypted (hexadecimal) data make to an eavesdropper?
|> 
|> The data is not "hexadecimal"; its in Radix 64 format.  I can readily
|> detect the presense of Radix 64 data -- a very simple regular
|> expression filter will allow one to detect candidate lines of Radix
|> 64, and at a fast enough speed that one can sort through all the mail

Yes, but said evaluator isn't doesn't shipped with every major operating 
system.

|> going through even as high volume a gateway as, say, UUNet. Once a
|> regular expression has found such a line, its easy to check if
|> subsequent lines indeed contain more apparent Radix 64 data. One may
|> then assume that one is indeed confronting Radix 64 data, and
|> transform it into binary. Once turned into binary, one may attempt to
|> interpret the data as a PGP format file. PGP files are NOT random in
|> their first bits -- they contain magic numbers, file lengths, and
|> other such things that can be used to determine if the candidate file
|> is indeed a PGP file. As an example, I can check to make sure that the
|> first byte of the data contains the CTB for an RSA public key
|> encrypted packet, that the next two bytes contain the length of the
|> packet, that the fourth byte contains a "2" for version two, etc.
|> 
|> However, you insist that all of this is impossible, in spite of the
|> fact that your PGP program more or less does all of it every day.
|> 
|> >> I don't think there is any point in stripping them,
|> >> since it adds no security for you and will make the program a lot more
|> >> inconvenient to use.
|> >
|> >I do, since it can increase security via making it 
|> >"unprovable" that the message is encrypted/created
|> >my PGP.
|> 
|> How exactly does it make it "unprovable"? In the sense that anyone can
|> add those lines to any message they send, they "prove" nothing about
|> the contents, and in the sense that anyone can detect that a mail
|> message contains PGP data even without those lines, one would appear
|> in another sense to be able to "prove" that a message was created with
|> PGP even without those lines being present. Stripping the lines would
|> appear to do you not a lick of good -- it does not fundamentally
|> change anything.

As a matter of theory & cryptanalysis, no.  As a matter of proof in
court, I am not as sure.  Question:  Is there radix 64 data out there in
the world that is _not_ PGP data?  If there is, then the burden of proof
is on the government, probably, to show it is something else.  Or, perhaps,
the government will overlook it in the first place (faint hope, but why give
up on it).  Moreover, if one is devious or desperate enough, one may be able
to convince a judge that it is some other form of radix 64 data, even if the 
burden of proof is on the defendant and how could the government 
find such a person out in perjury without solving the cipher?  
With the tags, on the other hand, one has nothing to stand on.  One is
reduced to arguing that someone else put the tags in, which probably doesn't
go over well with a judge or jury.

Now, if the software _does not put tags on in the first place_, how can 
the cops convincingly plant anything?  If it does, a "planting" of the
data allows the cops to meet one form of dishonesty with another even if
an individual strips them off.

|> --
|> Perry Metzger		pmetzger@shearson.com
|> --
|> "They can have my RSA key when they pry it from my cold dead fingers."

-- 
   Larry W. Loen        |  My Opinions are decidedly my own, so please
                        |  do not attribute them to my employer