W3C home > Mailing lists > Public > w3c-ietf-xmldsig@w3.org > April to June 2000

RE: Manually Signed Digest as an XML signature type

From: <tgindin@us.ibm.com>
Date: Fri, 9 Jun 2000 18:52:28 -0400
To: Philip Hallam-Baker <pbaker@verisign.com>
cc: Philip Hallam-Baker <pbaker@verisign.com>, Barb Fox <bfox@Exchange.Microsoft.com>, "Joseph M. Reagle Jr." <reagle@w3.org>, w3c-ietf-xmldsig@w3.org
Message-ID: <852568F9.007DA769.00@D51MTA04.pok.ibm.com>
     First, your last statement that if there is no strong binding to the
document no electronic signature mechanism is useful is valid and, I hope,
not even controversial.  If great ease of forgery (much easier than that
for conventional signatures, for example) can be demonstrated for a
signature technique it is good for nothing at all.
     I ordinarily understand the term "biometric mechanisms" to refer to
measurements of non-voluntary characteristics (including voice prints)
rather than to speech recordings or handwriting, and I think most people do
as well.
     The essential question about the mechanisms suggested here is "how
easy and undetectable is a forgery in this case"?  Your statement that "if
I have even one sample of handwriting I can produce forgeries" is true to
some extent, but if taken as an absolute it would render any handwritten
signature on a document (probably on a handwritten document, but certainly
on a typed one) completely untrustworthy.  However, such signatures are
routinely accepted for many purposes.

          Tom Gindin


Philip Hallam-Baker <pbaker@verisign.com> on 06/09/2000 06:08:33 PM

To:   Tom Gindin/Watson/IBM@IBMUS, Philip Hallam-Baker
      <pbaker@verisign.com>
cc:   Barb Fox <bfox@Exchange.Microsoft.com>, "Joseph M. Reagle Jr."
      <reagle@w3.org>, w3c-ietf-xmldsig@w3.org
Subject:  RE: Manually Signed Digest as an XML signature type



     These are both biometric mechanisms.

     If I have a recording of a persons voice saying each of the
hex digits I can reasonably expect to created a sufficiently good
forgery of that person saying any document digest.

     If I have one sample of handwriting I can produce forgeries.

     I don't even need a computer to do this.

     Thus I reject your assertion that their is a strong binding
to the document. If there is no strong binding to the document I
don't see why the XML Dig Sig mechanism or indeed any binding
mechanism has interest.

          Phill

-----Original Message-----
From: tgindin@us.ibm.com [mailto:tgindin@us.ibm.com]
Sent: Friday, June 09, 2000 5:40 PM
To: Philip Hallam-Baker
Cc: Barb Fox; Joseph M. Reagle Jr.; w3c-ietf-xmldsig@w3.org
Subject: RE: Manually Signed Digest as an XML signature type


     Electronic signatures are not restricted to biometrics.  In
particular, a sound recording of an individual speaking a simple form of
words including a recital of a specific document's digest would be
feasible, would be an electronic signature, and would be properly bound
to
the document's digest in such a way as to constitute a signature of the
document.  Similar methods involving handwriting may also be feasible.
The
difficulty in these schemes is not that they are not valid signatures,
but
that the last step in validating the signature is not cryptographic and
may
not be readily automatable.
     My suggestion was not intended primarily for biometrics, as like
many
others I have not yet been convinced of the usefulness of biometrics for
the execution of signatures as distinct from access control (including
access control for key storage).

          Tom Gindin


Philip Hallam-Baker <pbaker@verisign.com> on 06/09/2000 04:34:47 PM

To:   Tom Gindin/Watson/IBM@IBMUS, Barb Fox
<bfox@Exchange.Microsoft.com>
cc:   "Joseph M. Reagle Jr." <reagle@w3.org>, w3c-ietf-xmldsig@w3.org
Subject:  RE: Manually Signed Digest as an XML signature type



All,

     I would slightly modify Barb's statement but to my knowledge
the restatement would not have any effect whatsoever on any
Electronic Signature scheme I have seen to date.

     The problem I have with electronic signatures is that a
biometric
alone cannot provide a cryptographically strong (i.e. immune to attack)
binding of the signature to the specific document.

     I have seen many, many biometric schemes that claim to construct
such a binding but on examination every single one that I have seen to
date either relies for comms security on the cryptographic key alone or
is
subject to relatively unsophisticated attack once the algorithm is
known.

     The only technique I have seen thus far that is promissing in
this
area is to use a biometric technique to gate access to the public key

     My very strong predjudice is that should sufficient motivation
exist that a proof may be constructed to this effect.

If we define

1) a cryptographic system to be one in which there is an
asymetry in the complexity of computing a function and its inverse.

2) a biometric system appropriately

My guess is that it is possible to construct a proof that any system
in which there is an asymetric work function associated with the
creation and verification of a signature validating the document
falls into this category.

     That is just my opinion, I could be wrong.

               Phill
(snip)
Received on Friday, 9 June 2000 18:53:11 GMT

This archive was generated by hypermail 2.2.0 + w3c-0.29 : Thursday, 13 January 2005 12:10:09 GMT