CFP: PKI research workshop

Lynn.Wheeler at firstdata.com Lynn.Wheeler at firstdata.com
Fri Dec 28 18:50:04 EST 2001


both atm debit network and domain name infrastructure care capable of local
caching .... so that timelyness is within seconds to minutes (or a few hrs
as parameter within the needs of the infrastructure). the offline world for
certificates is the analogy of the letters of credit from the days of the
sailing ships. near real time with managed caching (with relying parties
forced to deal with stale credentials manufactored months or years in the
past).

part of the issue in clearing is who has the "liability" at any particular
instance; in the case of debit network caching there are very specific
procedures and processes. Are you suggesting that the certification
industry will assume liability in the case of offline clearing associated
with mars colonilization?

the process tends to be authentication, authorization, and finally
settlement and clearing. sometimes authorization, settlement and clearing
can be batched. if you are really talking about the bank account balance
resides on the earth and the access is from mars .... offline
authentication (clearing really needs to know whether the money actually
exists or not .... regardless of whether or not you are dealing with the
owner of the account) doesn't get you clearing .... and real clearing needs
to know that the money really exists (not just that a person is
authenticated)  ... and if the account balance is on earth and it takes 30
minutes elapsed time to establish it ... then that it what it takes.

More realistic is account balance caching at some near real-time location
on mars ... say within the parameters of the ATM withdrawal limit.

At one point in the PKI evolution there was the proposal that there could
be certificates analogous to the '70s "signing limit" checks .,... an
attempt to create certificates that not only provided authentication
information but also some hypothetical useful approximation to
authorization information (aka not quite reqressing totally to the pre-70s
credit card model). The issue in the "signing limit" checks was when they
found people writing 200 $5000 (limit) checks to get a million. What has
been seen since that time is near real-time purchasing department operation
(including business purchase cards that leverage the credit card system) to
provide real-time aggregation ... as opposed to sinlge event operation. In
the ATM machine withdrawal case, there are typically both single widthrawal
limits as well as daily aggregate withdrawal limits (aka the PKI proposal
for credit cards turned out to be a business process regression to pre-70s
and the PKI proposal for business checks turned out to be a business
process reqression to pre'80s).

Typically what you might have in a ATM withdrawal case .... with foreign
ATM machine (not your local bank) .... is that the owner of the ATM machine
is given a guarentee of funds from your financial institution prior to the
ATM machine releases paper money. Your bank then effectively debits your
account for the equivalent amount of funds. Then typically sometime that
evening, there is a settlement operation where there is funds transfer from
your bank to the financial institution that owns the ATM.

An offline, stale certificate .... only (slightly) addresses the issue of
authentication .... say an identification certificate ... which might not
even provide a binding between you and any particular bank or bank account.
Some sort of binding between you, your bank, and your bank account is
needed .... just for the authentication phase of what you are talking
about. There is still the authorization phase needed so that the owner of
the ATM machine believes that it can receive something (in return for
spitting out paper bills).  That effectively has to find that there are
actually sufficient funds in your account.

So a more realistic scenario would be that there is possibly dual account,
one local and one on earth ... with funds floating back and forth as needed
in evening settlements. If you are on Mars there is some local financial
branch with local record of funds that you have immediately available and
which can authorize that amount of money.

A "local" financial branch implementation and a digital cash implementation
might have a number of similar useability attributes .... aka from the
standpoint of how local funds do you have immediately available .... aka
funds are transferred into you local PDA as digital cash for immediate use
.... or funds are transferred into the local financial institution for
immediate use.





ray dillinger <bear at sonic.net> on 12/28/2001 2:29 pm wrote:


The only case in which the PKI solution is not redundant is in
offline clearing.  But getting your point-of-transaction online
is easier than paying attention to PKI.

I happen to like offline clearing -- it opens up the possibility of
new transaction types and doing transactions in places you couldn't
before.  But the practical issue is, everybody who's interested in
electronic transactions of any kind is also interested in getting
online, and when PKI's were deployed in "developing" areas (south
africa) they got dumped just as soon as the area was developed
enough for communications to support online clearing.

On the principle of people refusing to adopt something until
it relieves pain, maybe we won't see a real PKI deployed until
we need to serve markets where speed-of-light delays make online
clearing impractical.

Mars, for example, is 3 to 22 light-minutes away.  I don't imagine
someone using an ATM on Mars is going to want to wait 12 to 88
minutes for online clearing (more if the protocol is talky or the
bandwidth is busy...).  So a martian colony might be the first
practical application of PKI and/or digital cash, assuming the
colonists want to do business with Earth companies.  But a colony
looks pretty distant right now: we haven't even got an outpost
there yet.

                                          Bear






---------------------------------------------------------------------
The Cryptography Mailing List
Unsubscribe by sending "unsubscribe cryptography" to majordomo at wasabisystems.com




More information about the cryptography mailing list