The context of this CodeX Conversation is inspired by Richard Tromans’ article, “Law Isn’t Code – And That’s Good Thing”. CodeX Fellows Klaudia Galka and Megan Ma, will refer to it as an anchor to their conversation.

We begin with Tromans’ assertion:
“Contracts are approximations of an agreement that will be expressed in the real world“

Tromans describes how contracts are “not always followed to the letter;” that there is interpretation at play; and deviations that “common sense will allow.”

Klaudia: Insurance contracts specify under what terms and conditions an insured individual will be compensated by an agreed amount. The real world is complex and constantly evolving. So is everything else, including the insurance policy which gets renewed every year. It is in the interest of both parties, often also mandated by the regulator, to avoid any legal uncertainties. Most of the processes in insurance have been digitized and now is high time for its core product – the underlying insurance contract. The benefits of having a contractual ‘digital twin’ are huge!

Megan: It seems to me that the intention behind code-ification is of utmost importance. If other elements within the insurance ecosystem are digitized, would offering a machine-readable version of the contract help connect other parts of the ecosystem to one another? More importantly, would it foster congruency as a whole? There is frequently the assumption that code-ification would flatten and eliminate the interpretative space. This may be so if the purpose was to eliminate uncertainty through simplification. But, what if the intention was for clarification? This then becomes a different story.

“A standard clause is not the same as code“

Klaudia: Relative to natural language, code is a simply different, more structured way of expression. Code still allows for variation and gives room for adaptability. It is like a box of Lego, with standard clauses akin to a pile of bricks. You can do so much more with Lego!

Megan: Modularity is key and standards come in many forms, including at several levels of granularity. Standardization at a clause level works differently than standardization at a conceptual level. Moreover, translation from law to code is not necessarily about isomorphic mapping, or a 1-to-1 semantic match. What matters is ontology, a proper conceptual transfer.

“The success of legal tech will have to be a compromise between the absolutism of computer code and the subjectivity and variability of human language”

Klaudia: It has been proven that in almost any discipline, the most powerful combination is the one of human and machine. Sometimes, legal tech places itself too much in the ‘self-driving cars’ category. We are not there yet. For now, we should instead focus on the tangible benefits legal tech can bring to the industry. Some key themes which should be explored further include: Are there some computational approaches that work better than others when dealing with natural language? Could the approach depend on the use case (e.g., contract maturity, contract complexity, type of contract, pre/post-negotiation stage)?

Megan: I find that the notion of ‘absolutism of code’ is misleading. Code is well-defined and definite, and in many cases, places constraints on defining and definition. This does not mean that code cannot account for the variability in human language. Perhaps a better question to ask is whether we sufficiently considered the expressibility of code beyond function? That is, is code merely used to execute tasks, thereby functional in nature; or can code be understood as a drafting tool, capable of writing and communicating thought?