Yesterday I attended the “Computable Contracts: In Theory and In Practice” workshop at the Universidade do Minho Law School, sponsored by Singapore Management University and Stanford Center for Legal Informatics (CodeX). Thank you to the organisers!

It was great to catch up on the latest advances in computable contracts, which ranged from syntactic sugar on top of Prolog, to using GPT-4 to automatically create LawScript programs, to creating logic programs inside Google Sheets, and everything in between.

I presented a brief overview of the work we do at Accord Project.

ap-icail-2023Download

My risky hot-take is that that we don’t have a good way to represent temporal / stateful multiparty state-machines with side-effects, of the type we see in contracts, certainly not at the level of detail required for traditional deterministic execution.

  1. English is ambiguous (and lawyers punch out at a certain level of detail)
  2. Structured natural language ends up looking a lot like code
  3. Logic programming is hard
  4. Real-world ontologies are large and complex
  5. Code is impenetrable, hard to read, certainly for non-technologists
  6. Flow charts are big / complex / hard to maintain
  7. Rules are difficult to write and maintain
  8. Spreadsheets get big quickly. Are they maintainable?

Will LLMs save us? They may make accurate guesses at what we struggle to express ourselves. Will that be good enough?

Below are just a selection of the materials that caught my attention. Lots to process!