From: hughes@ah.com (Eric Hughes)
Date: Fri, 19 Aug 94 16:49:23 PDT
To: cypherpunks@toad.com
Subject: In Search of Genuine DigiCash
In-Reply-To: <9408192310.AA10603@TeleCheck.com>
Message-ID: <9408192324.AA13829@ah.com>
MIME-Version: 1.0
Content-Type: text/plain


	   I think that very few would have the initiative to lay out the
   money for a no-transaction cash system.  With credit cards and checks 
   there is a transaction trail that you can follow to spot and get rid of
   fraud.  

I trust that for "transaction" above you mean "audit".  You still have
transactions and you still have audits.  It's just that this
information does not allow for the derivability of the customer's
transaction.

Assume four accounts in the books of an issuing bank: one asset
account, cash, and two liability accounts, one for a customer and one
suspension account for digital banknotes issued by not yet redeemed.

The withdrawal transaction posts a debit to a customers demand deposit
account (decreasing it) and a credit to the suspension account
(increasing it).  Now suppose the customer buys something from a
merchant, and the merchant redeems the digital banknote cash.  The
deposit transaction posts a debit to the suspension account
(decreasing it) and a credit to the cash account (also decreasing it).

As you can see, there are perfectly good journal entries for each of
the two transactions just described.  What is missing is an audit
trail to determine which debit to the suspension account corresponds
to which credit to the suspension account.  An assurance that these
match up is provided by two properties.  First, for each banknote
issued there is one and only feasibly computable modification of it
that is acceptable for redemption.  (In Chaum's scheme this is the
unblinding.)  Second, a database of the banknotes as redeemed is kept,
which prevents multiple redemption.

   Will it be a replacement to ATM and credit cards or would it be a concurrent
   working solution?

Concurrent, of course.  There's very little point to scrap any
existing system as a system.  Individual merchants may decide not to
support older systems eventually, but that is a different issue.
Nonetheless, I have argued at length at other times that digital cash
will not be viable as a physical retail system very soon.

Where digital cash is immediately useful is online as a retail level
wire transfer system.

Chaum:
	 Such a verification procedure might be
	 acceptable when large amounts of money are at stake, but it is far too
	 expensive to use when someone is just buying a newspaper. 

Maybe a physical newspaper today, but the cost of networking is
dropping and the cost of computation is dropping.  I personally don't
expect that off-line digital cash techniques will ever actually be
economically most efficient.  Existing alternates (e.g. credit cards)
work well enough today, and by the time PDA's work well enough and are
cheap enough to be universal, the cost of an online verification will
be down in the fractions of a cent.

Eric