Let's do the information analysis again.

The public keys that happened to be resident in browsers ... and other
distributed applications .... happen to be public keys in account records
.... that the physically encoded in certificate format happens to be a
physical encoding artifact not a business issues. In fact, almost all
standard discriptions say that these public keys are handled in an
"out-of-band" process that isn't part of the normal (offline) TTP/CA
certificate business process (i.e. as part of a three part
transaction/message, aka the actual message, the signature on the actual
message, and the signer's certificate obtained from a TTP/CA).

The business process of the public keys contained in these account records
.... typically included as part of some out-of-band business process ....
just happens to be in certificate encoded format. If I were to register
public keys and as part of that registration encode them in certificate
format as stored in the account records (in the AADS model) ... they could
all be considered to be certificates also. However, just because they are
in certificate encoded format ... doesn't make them part of a offline
TTP/CA infastructure.

My actual statements ... in further detail is that in the online AADS model
... the traditional offline TTP/CA business process involving the
three-part message/document/transaction has the certificate part as
redundant and superfulous .... aka part I is the actual
message/document/transactio9n, part II is the digital signature, and part
III is the (offline) TTP/CA certificate issued to the entity signing part
I.

The discussion about offline TTP/CA having a vanishing market niche ... for
selling offline certificates ... is in no way invalidated by the existance
of account-oriented public keys that happened to be encoded in certificate
format in your model.

For the most part these "static" certificates have been acquired by some
out-of-band business process (not included in the traditional offline
TTP/CA business operation) and also tend to be self-signed. Again, just
because data has been encoded in certificate format doesn't automagically
create any logical conclusion that offline TTP/CA business operation has a
large market opportunity in an online world (aka severely confusion the
encoding method with the actual business process).

When my wife were doing this stuff that came to be called electronic
commerce with this thing called SSL .... we realized that we were inserting
public keys in both browsers (the traditional SSL based stuff that most
people are aware off). However, there was also this thing called a payment
gateway ... and we were doing mutual authentication (this was before SSL
had a definition for mutual authentication). This part could be considered
one of the first B2B infrastructures (although must people aren't really
familiar with it since it was much more of a fixed backroom process). The
encoding of the public keys used certificate format ... but there was no
use of any offline TTP/CA business processes. The business software
supporting the payment gateway had the payment gateways public key inserted
into the software by an out-of-band process .... and other than the COTS
software required certificate format for the public key encoding .... there
was no aspect of the process that resembled any offline TTP/CA business
process. Correspondingly each merchant site that was allowed to use the
gateway was predefined and had account record created. Again the use of
certificate encoding for the public key was purely a side-effect of the
COTS software library being used and the actual business process bore no
resumblance to any offline TTP/CA business operation.

The offline TTP/CA business processes have other similarities with AADS
models (other than possibly using certificate encoding for public key
storage). Basically the offline TTP/CA business process inclused something
called a registration authority. This basically does some up front
validation of public key stuff before storing the public key in a TTP/CA
internal business process database (aka account record). Effectively both
offline TTP/CA business processes and AADS implementations can use very
similar business process for the validating and recording of public keys in
account records (weither or not they are certificate-format encoded when
they reside in the account record). Such account records can be in
traditional database management infrastructures or just simple tables that
are typical of the form used for public key storage by internal browser
processes or say by some of the PGP applications. Whether or not they use
certificate-format encoding for the stored public keys and whether or or
not the account record is a real database entry or just a much more simpler
linear table doesn't affect the logical business process (and still doesn't
automagically turn it into a offline TTP/CA business process).

The place where AADS starts to differ significantly from the offline TTP/CA
model is with the elimination of the appended certificate as the 3rd part
of a signed message/document/transaction.

Now, a traditional account record .... doesn't usually stop with the
identification of the TTP ... but some may ... and some could actually use
certificate-format encoding of the stored information ... but they
typically have to have much more context than just simple a TTP
identification. There is a lot more business context involved ... that has
to be wrapped around this. At a minimum, it isn't sufficient that you know
the identification of the entity .... there has to be some business process
that established this is a TTP entity ... and not just some random entity.
If all I had was the name of the entity .... and nothing about it be an
acceptable TTP entity .... the infrastructure could be susceptable to
fraudulent impersonation (aka just because i have a certificate that has my
identification and my public key doesn't make me an acceptable TTP). So
there has to be some additional context ... either the whole table of
publickey/identification entries has to be known to only be acceptable TTPs
... or each entry has to have additional information distinquishing between
entries for acceptable TTPs and entries for other entities.

misc. refs about this stuff we did with this small client/server startup
that was doing this stuff called SSL, HTTPS, and electronic commerce ....
the year that we worked with them they moved from menlo park to mountain
view and changed their name from mosaic to netscape.
http://www.garlic.com/~lynn/aadsm5.htm#asrn3
http://www.garlic.com/~lynn/aadsm5.htm#asrn2
http://www.garlic.com/~lynn/aadsm5.htm#asrn1
http://www.garlic.com/~lynn/aadsm5.htm#asrn4

to re-iterate .... the method of encoding the data doesn't establish the
business process of the use of that data. just because some data happens to
reside in an account record in certificate-format encoded format doesn't
automagically turn things into offline TTP/CA business operation.



[EMAIL PROTECTED] on 12/10/2002 1:33 pm wrote:

Lynn,
In ad-hoc peer-to-peer B2B-networks relying on a set of TTPs,
a static certificate have at least one important function:
- identifying the TTP

So you can delete the certificate, keep the public key but you
must add something else to let the RP find the on-line TTP.
An URL maybe?

Anders

To remove yourself from this list send a message Unsubscribe to
[EMAIL PROTECTED]




Reply via email to