From: szabo@netcom.com
Date: Sat, 17 Aug 1996 08:24:44 +0800
To: AltInst@cco.caltech.edu
Subject: Credit enforcement
In-Reply-To: <199608152033.QAA03582@prob>
Message-ID: <199608162123.OAA16545@netcom.netcom.com>
MIME-Version: 1.0
Content-Type: text/plain



Steve Omohundro writes on alt-inst:
> I'm intrigued by the notion that in a digital economy
> contracts might be enforced by design rather than by the good faith of
> the participants. 

I have been intrigued by this possibility as well.  See my essay at
"http://www.best.com/~szabo/smart.contracts.html".   Steve has
put his finger on one of the basic outstanding problems
in this area, namely the enforcement of credit.  Currently
there are several partially effective processes:

* Reputation (especially credit reports): often effective, but
only to a point, as it is often hard for the debtor to accurately judge 
the future reputational effects of an action (eg failure to pay a bill,
taking out too large a loan, etc.) that has clear, local,
beneficial effects today.   This is more imbalance in knowledge
between current and distant consequences among individual 
consumers, but even among large organizations with high credit
ratings it is not an irrelevant factor.

* Secured transactions: liens, etc.

* Garnishment of future income

* Law enforcement, especially to enforce transfer of control over 
liened assets, garnishment, etc.

These processes have a fundamental property in common --
they violate the privity of credit transactions -- they bring
in third parties to track reputations or enforce repayment.
Credit transactions seem to entail a fundamental imbalance in 
incentives that can only be redressed by bringing in third parties.

Secured credit need not violate privity if the physical control
over the securing property can be shared.  So that, for
example, automobile credit can be secured as long as reposession is
possible.  The trick is to make repossession by the creditor
easy but theft by third parties difficult.  I have proposed 
"smart liens" along these lines, electronic security measures
strong against third parties but with a "back door" for
creditors.  This well-specified, shared control over "smart property"
more accurately reflects the agreement involving that property, so that 
there is less need for third parties.  To even more accurately 
reflect the contract, we need a mechanism to eliminate the creditor 
control once the auto loan has been payed off.

Alas, there is less incentive to provide these kinds
of contractual process improvements in a market where
government subsidizes the enforcement of contracts.

Similar mechanisms might be possible for other kinds of
security (houses, escrow accounts, etc.), but many valuable
kinds of credit are unsecured, and we run into privity problems when 
it comes to garnishment of future income.  Here, we are invoking third 
parties, namely the debtor's future contract counterparties.
Any mechanism seemingly needs to involve them, but both principals
have an incentive to enter a private, ungarnished 
contract in preference to one involving the creditor.
(ie, the amount of garnishment is a surplus to be divided
between principals who can route around it).

A way to a solution, if it were feasible, would be to give the creditor 
shared control over the entire scope of of the debtor's income
capabilities - or, a bit closer to practicality, over the entire scope of 
his digital income capabilities.  A secondary solution is some combination
of wide scope and limited compromise of privity.  After all, money itself 
is a compromise of privity, since the contract parties rely on third 
parties to clear and maintain the value of the currency.  Money's compromise 
of privity is well-defined, however, not an open-ended release of information
and physical control, even over one's own person, as often occurs with 
credit reports and law enforcement respectively.  Our challenge is to find
privity compromises with such well-defined limits to enforce
credit transactions.  

One possibility here is a "garnishable currency".  All banks 
have an interest in enforcing credit, so they can
make deals with each other to enforce credit via the garnishment
of debtor bank account deposits at any participating bank.  However, 
substantial amounts of garnishment (and if it provides a lower cost way of 
enforcing credit the amounts will be substantial) gives rise to an incentive 
for banks to fail to participate.  Here the need to commonly clear a currency 
between banks can be used as a barrier to entry for such defectors.
The currency is simply declared garnishable, and any banks who
wish to deal in the currency must participate in the garnishing
process.  This currency wins against competitor currencies
in a free-banking market because it provides a better means
to enforce credit, allowing greater credit expansion at lower
isk.

On the other hand, traditional coin and currency transfers, and 
some kinds of digital cash transfers, need not 
involve deposits to bank accounts linked (usually by True
Name, but a "debtor nym" could also work) to one's unsecured debts,
and given that there is a market for these kinds of transfers 
for other reasons, its existence allows banks to defeat auditing of 
garnishment by other banks participating in that currency.   Abuses 
of financial auditing for the purposes of extortion, inside information, etc.
will likely maintain a major market for the non-deposit 
payment methods.

Steve proposes "interval money" that would expire.   A similar idea
has been proposed by Tim May, a "time release" form of money that 
becomes good only after a certain date.  These can
probably be implemented by a digital mint expiring or activating
special issues of digital cash, or by a third party issuing
escrowed keys at specific times (since these keys are encrypted
against the escrow agent, and that agent doesn't know what they
will be used for, the escrow agent has no incentive to cheat).
A technical issue here is whether the digital signature space is large 
enough to encompass one unique signature per unique credit transaction
deadline into the indefinite future.  While these protocols are
intriguing, and potential building blocks for a solution,
the institutions using them proposed so far seem to still rely on
a solution to the deposit garnishment problem as discussed above, that 
is the problem of sufficiently disincentivizing the debtor from using 
non-garnishable alternatives while maintaining the privity of
payments to those who are not unsecured debtors (the ability for 
these people to use non-deposit payment methods).

Nick Szabo
szabo@netcom.com
http://www.best.com/~szabo/