anonymous companies

[Wei Dai <weidai@eskimo.com>]

On Tue, 14 May 1996, Hal wrote:

> It might be interesting to make a list of all the problems people can
> think of why this idea won't work, paired with proposed solutions and
> workarounds - sort of a mini FAQ for this important (some might say
> ultimate) cypherpunk model.

I'll just give one problem: the principal-agent problem.  How do owners of
the company make sure the managers operate the company in their best
interest?

Solution: reputation.  If the managers don't do the right things, the
owners arrange so that the managers lose reputation and won't get hired in
the future.  Unfortunately the science of reputation is not so advanced
that we know this will actually work.

Solution: smart contracts.  This is Nick Szabo's idea of building
contractual obligations into cryptographic protocols so that the parties
have no choice but to fullfil them.  But again we don't know whether this
will actually work for this problem.

A company implies a particular kind of persistent structure, with a
hiearchy of owners, managers, and employees.  It is far from clear to me
that this is the most likely organizational form in an anonymous digital
economy.  One possible alternative is to have no persistent organizations. 
Teams form spontaneously to work on individual projects.  Each individual
member jointly negotiates a contract with every other member, and these
contracts are enforced through some arbitration system. 

I'm not saying this is somehow better than the anonymous company model. 
It has just as many problems for which no easy solutions exist.  I'm just
pointing out that the properties of anonymous relationships differ quite
radically from our current ones, and that these differences may be large
enough so that the social and economic structures in such an anonymous
digital world may not merely be analogs of currently common structures. 

Wei Dai