Common Law +
Computation = Evolvable Contracts
BIG IDEAS FOR LEGAL & INSTITUTIONAL INNOVATION
(CONTRIBUTORS MAP FOR EACH SECTION - visualization software graphically showing type of contributor and amount)
“then the rule adapts itself to the new reasons which have been found for it, and enters on a new career. The old form receives a new content, and in time even the form modifies itself to fit the meaning which it has received”
OLIVER WENDELL HOLMES
CONTRACTS SHOULD BE PROGRAMS THAT EVOLVE TO PRESERVE FAIRNESS AND INTENT: A DEMONSTRATION OF AN EVOLVABLE TERM SHEET
A contract is an agreement between two or more parties that defines their mutual obligations to one another under a variety of conditions. It generally is enforceable through a court or arbitration. A contract is fair when it is willfully entered into with full and informed consent. Yet contracts can become highly complex in having to anticipate future contingencies and provide appropriate remedies that preserve the initial intent of the contract.
Since all contracts are expressed as static documents, it is very difficult to capture in text dynamic, evolving relationships that can be triggered by potential, future events. Good contracts try to anticipate those future events that can potentially effect the obligations and interests of the different parties. However, in a textual document this can only be done by enumerating and commenting on such contingences and remedies, often resulting in hundreds of pages of complex, technical prose that even legal professions find daunting. For many signatories, they are told, “Just sign here at yellow marker,” without actually reading, much less comprehending the full import of the contract. Under such circumstances, it is hard to see how such a signature represents “informed consent”, whether it be a young entrepreneur signing a venture capital term sheet, or a seasoned board member ‘signing off”’ on a merger and acquisition agreement.
Anther limitation of representing a contract as a text document is the inability of text to capture and generalize the intent of the drafters. Intent by definition cannot be literal, as it represents a principle of how to act rather than a particular kind of action. While such a principle or rule might be described in text, it cannot be implemented as text, since text is inherently declarative and rules are procedural, in short, algorithms.
The question being raised here is whether it may become possible and desirable to algorithmically represent protect the interests of parties to a contract without knowing in advance what events or actions might impinge on those interests? Instead of simply writing a contract, it may be more effective, efficient and fair to program a contract to act in a way that transparently, exhaustively, and fairly preserves the interests of the parties for circumstances, not previously contemplated, or in Oliver Wendell Holmes, Jr. words, ““ then the rule adapts itself to the new reasons which have been found for it, and enters on a new career. The old form receives a new content, and in time even the form modifies itself to fit the meaning which it has received.”
The recent invention of algorithms that mimic natural selection, called genetic algorithms, (Holland, 1980) offers an opportunity to program rules to generate and select among all potential contracts to met the “fitness conditions”, or in this case, the original intentions, of the parties to the contract. The complexity of this task is simply beyond the capacity of any human. For example, in the term sheet pilot demonstration described in the following section, there was a search space of 1014 potential contracts - and this was a relatively simple contract.
A good contract is one that is both fair and durable. Fairness is not just a matter of how the contract is drafted, but how it is understood. Parties need to understand how their interests are served, and they need to feel confident that they will be protected under most circumstances. Robustness in biology refers to the ability of an organism to survive and replicate under a variety of adverse and diverse conditions. Similarly, robust fairness in a contract refers to a contract’s ability to be both transparent and durable in protecting the original intent of the contracting parties.
THE ARGUMENT FOR: COMPUTABLE CONTRACTS
The argument being made here is that even relatively simple contracts – like a standard venture capital term sheet – are difficult to understand – and when extended to account for even standard business contingencies, costly to draft and negotiate. Moreover, as they grow in complexity, they can become neither transparent nor fair.
Contracts can also have unintended consequences by forcing different parties into more adversarial positions and defensive positions than either party wants or benefits form. Therefore it would be advantageous to all parties to have a process whereby the implications and outcomes of contractual provisions were as transparent and comprehensible as possible. Moreover, it would be further beneficial and fair to all parties if the process of generating, drafting and evaluating alternative contractual relationships were as independent and inexpensive as possible – in short, automated to the extent possible. Given that the space of alternative contracts can be very large and highly sensitive to variations in business scenarios and changes in specific clauses, this process should be automated to any extent possible. Moreover, by making the contract model/algorithm testable and subject to empirical data – such as the empirical likelihood of different business scenarios or other events, such contracts become less expensive to draft and more fair and durable.