Re: transaction costs in anonymous markets

-----BEGIN PGP SIGNED MESSAGE-----

In article <Pine.SUN.3.91.951016011355.2241A-100000@eskimo.com> Wei Dai said:

 "Maintaining anonymity requires that communications be done through 
 special untraceable protocols."

I think it is important to keep in mind exactly what we mean by anonymity
in these discussions - there seems to be a bewildering array of flavours.

Take three situations where anonymity might be useful:

1. The State Authorities come and take you away for subscribing to a
publication advocating the end of socialised medicine.

2. The State Authorities come and take your assets away to pay for socialised
medicine.

3. The Private Hospital comes and takes your assets away to pay for your
lifesaving operation.

[Avoiding 1 is (in my opinion) an unmitigated good; avoiding 2 is obviously
desirable; but avoiding 3 could well be counter-productive.]

This points to two categories of anonymity:

   Meat anonymity - there is no way of knowing which human is responsible; and
   Asset anonymity - there is no way of linking a particular human, transaction
                     or activity to any particular asset, or vice versa.

The point being that sometimes one will be desirable without the other.

Concentrating on meat (or combined?) anonymity, there are (at least) two
sorts;

   Legal anonymity - where the split between asset and meat is the result
                     of the interposition of some legal structure (eg. Trust,
                     corporation etc), the ultimate owners/controllers of
                     which are unknown; and
   Digital anonymity - where the identity of the actor is mathematically
                       unknowable (at least, at the limit).

Obviously, these sorts can be layered, with legal entities transacting as
digitally anonymous parties, and the owners of those legal entities protected
by digital anonymity.

In addition, we can distinguish between digital anonymity that allows no
link between separate transactions (anonymity proper[?]), and digital
anonymity that does allow links to be drawn between separate transactions
or activities (pseudonymity).

Similarly, we can distinguish between the various anonymous properties of
transactions:

A. Private Transactions. Each party to the transaction knows the identity
of the others, but no third party can know who is transacting with whom.

B. Unilateral Anonymity. One party to the transaction is known, but the other
is unknowable. An example might be subscribing for a digital security with
ecash - the issuer is known, but the purchaser is anonymous.

C. Bilateral Anonymity. The identities of both parties to the transaction
are unknowable. This might be the case on a cypherpunk stock exchange...

                                        "The current state-of-art (i.e. 
 RemailerNet) adds several hours of transmition time to each message to 
 achieve effective untraceability.  Contract negotiation, for example, 
 becomes very difficult under these circumstances."

My experience of contract negotiation is that a delay of hours would be
immaterial. Don't forget that email, fax machines, and even the telephone
are relatively recent phenomena (not to mention the photocopier and the
word-processor). Most super-tight deadlines are self-inflicted.

"If these costs remain high, but anonymous markets develop regardless, it 
will be interesting to see how these costs affect the structure of the 
markets.  Will special protocols for contract negotiations develop to 
minimize the number of round-trip messages?"

Doubtless. In areas where complex transactions really have to be entered
into at short notice, a lot of the negotiation goes on in advance, often
by trade associations. ISDA, for instance, is forever producing standard
terms and appendices so that all manner of arcane derivative transactions
can be more or less negotiated over the phone.

Cheers,

Dm


-----BEGIN PGP SIGNATURE-----
Version: 2.6.2

iQCVAwUBMISijFlo3j8JHzalAQENEAP/UJTSvhAR534mh1KUQZQC1JYxZeROUaDf
6Nd42yxCmLbbSa72dEAB69jdfDHBj2sj6Jf6jL+HAGTrfG6I/dNJuKu9MaaCw64z
MKe0efTCmD2caehOv9lf5tI2uJ0vAql8UdCC5gj3NhkxKr/uEAvcbIlt/Z60mmEu
4P2hxCv1u/o=
=U8gP
-----END PGP SIGNATURE-----

[Palmtop News Reader - Beta Version 3]

On Wed, 18 Oct 1995, David Murray wrote:

> B. Unilateral Anonymity. One party to the transaction is known, but the other
> is unknowable. An example might be subscribing for a digital security with
> ecash - the issuer is known, but the purchaser is anonymous.
> 
> C. Bilateral Anonymity. The identities of both parties to the transaction
> are unknowable. This might be the case on a cypherpunk stock exchange...
Or we could use the client/server terminology when possible.

-----BEGIN PGP SIGNED MESSAGE-----

In message <Pine.3.89.9510180711.B37215-0100000@aix2.uottawa.ca>, s1018954@aix2.uottawa.ca said:
> On Wed, 18 Oct 1995, David Murray wrote:
> 
> > B. Unilateral Anonymity. One party to the transaction is known, but the other
> > is unknowable. An example might be subscribing for a digital security with
> > ecash - the issuer is known, but the purchaser is anonymous.
> > 
> > C. Bilateral Anonymity. The identities of both parties to the transaction
> > are unknowable. This might be the case on a cypherpunk stock exchange...
> Or we could use the client/server terminology when possible.

I was actually vaguely referring to contract theory - unilateral and bilateral
executory contracts.

And this is where this discussion folds into Bob Hettinga's point about
non-repudiation. How can an anonymous party credibly bind its future actions?

One way is to hide the identity of the human/actor making the promise/contract
but leave their assets where the other party (the promisee) can get to them if
the anonymous party defaults. (The reputation of a pseudonym is an interesting
version of this.) A pledge of digital securities is a possibility; some sort
of protected pool of assets is another.

Alot of transactions/contracts only involve one party making a promise that
they have to perform in the future. When I get on a bus, once I've paid the
fare, only the bus company still has to perform its half of the contract
(although, as something of a complicating factor, I could still _breach_
the contract). This is why I used the subscription for a security as an
example - once I've paid my ecash, it's only the issuer that has to make
the coupon payments and redeem the security in the future - the contract is
unilaterally executory, with the known party still to perform.

But alot of contracts involve performance over time by _both_ parties - they
are bilaterally executory. It would seem that, to be effective, both of the
parties (or at least sufficient of their assets) must be known.

So, unilateral/bilateral anonymity is kind of the complement of unilateral/
bilateral executoriness.

Perhaps client/server terminology is better.

[And before someone points it out - alot of these issues have been explored
centuries ago, and the answers led to commercial innovations like bills of
exchange, bills of sale, bills of lading etc. I plan to spend more time
learning from legal history in the hope we are destined to repeat it...]

Whatever.

Dm.


-----BEGIN PGP SIGNATURE-----
Version: 2.6.2

iQCVAwUBMIYWxVlo3j8JHzalAQFgJQQAndaHZT4pUsx2WojXsMaiI7cBblyDE2S6
v8b74NzuGxZpDsUjtQFTuqhbIbqPjpLuIxn7c0DjNAPoa0w3LwCOvtjUkDh4HiKj
UnlXu61I6Tf3KKUr4GmBadWmADdzfnPPXl7HFHr0GfH7a17ET97CVJzoYQTYl5xu
rI3y3gYe8e8=
=QDQf
-----END PGP SIGNATURE-----

[Palmtop News Reader - Beta Version 3]