Berkman, MIT, Primavera and GitHub

• In 2014, in connection with a legal hackathon at MIT, I was referred to Primavera De Filippi at Harvard's Berkman Center because of her work on smart contracts and blockchains. Primavera saw the connection between smart contracts and legal text, but also the problems of the various frameworks that CommonAccord then was in. She recoded the app, rapidly, in perl, enabling use of flat files, for the first time since the original RoR version. This permitted us to host the app and content as a repo on GitHub, to run the app locally or via services such as Heroku with a very minimal install. It enabled a fully decentralized data model based on repositories, working just like collaboration on software by the open source community. It was also the first version with lazy evaluation of files. This enables unlimited use of the object model. Her coding was done rapidly, famously at a McDonalds in Buenos Aires and overnight at the Media Lab, and has performance issues and very little error-handling. But it is perfectly accurate. It can be the measure of compliance.
• Since that recoding, I figured out solutions to a few problems. The most important of these are:
  1. Widgets for common format elements - notably sections and subsections. Most legal documents, and many others, are outlines. These widgets make it very easy to onboard and maintain a new document. They invite consistency in format without imposing it.
  2. Organizing the file system into an image of the web (at /W/) and GitHub (at /GH/) so that materials are organized by author (actually by host), rather than trying to develop a generic naming scheme. This has been greatly simplifying and liberating. (There are also a few subject matter classifications, such as /D/ for transactions; /U/ for persons, places and things; /S/ for shared, common materials, /P/ for private materials.)