// COMPLETE THREAD

Re: Digitally Signing Physical Objects

2 expanded posts ยท every known parent and child

NODE a4837091Re: Digitally Signing Physical Objects
Tim has an interesting point on the use of digital signatures.

A variation is to use an "undeniable" signature.  This is a signature which
can only be checked with the cooperation of the signer.  However, the protocol
is such that the signer cannot cheat and try to deny a valid signature
(hence the name).  This could be used by manufacturers to authenticate
their products only to certain customers; for example, to customers who have
paid for them.

This might be especially useful for software, although Tim's idea would
extend it to any object for which the authentication is especially valuable.
PGP is distributed signed by Phil Zimmermann using an ordinary digital
signature.  This allows anyone to verify that it is a good package, free
of viruses or trap doors.  If it instead had an undeniable signature, this
verification would require interacting with Phil (or his agent) via a
protocol; but at the end the same assurance would result.  This kind of
signature would be more appropriate with a payware product.

Undeniable signatures cannot be passed on from one person to another.
If Alice verifies Bob's undeniable signature, she can't prove to Charlie
that the signature is good.  She can claim it is good, and assure Charlie
that it is good based on her own reputation, but Charlie can in general not
be convinced unless he verifies it himself directly with Bob.

Hal
NODE 19d743dbDigitally Signing Physical Objects
>Undeniable signatures cannot be passed on from one person to another.
>If Alice verifies Bob's undeniable signature, she can't prove to Charlie
>that the signature is good.  She can claim it is good, and assure Charlie
>that it is good based on her own reputation, but Charlie can in general not
>be convinced unless he verifies it himself directly with Bob.

This is the standard reason given why undeniable signatures can't be
passed on.  And it is correct, as far as it goes.  But the conclusion,
that "in general" the trust cannot be passed on, while technically
correct, is not of pragmatic consequence.

I'll start a service to perform any undeniable signature verification,
even ones for money.  I'll perform the verification, and then make an
attestation that I perfomed the verfication and whether it succeeded
or failed.  I sign this is a regular digital signature, the kind that
is infinitely duplicable.

Only a few such services need exist to assure the public of the
results of a signature verification.

True, there is a layer of mediation here, but of what practical
consequence is that?  In fact most transactions are mediated already.
If I expect to make money charging a dollar per verification, and if
there are some who will publish their experiences of the verification,
that reduces the total income I can expect to, oh, say, the logarithm
of the size of the market.  In other words, why bother?

Eric