In my last blog, I discussed the DIKW pyramid and how the CAD world has advanced through the layers while the legal profession was going much slower. I mentioned that design synthesis was my boss Jerry’s favorite topic. We would spend hours at his desk in the evening while he described his vision for design synthesis — which would become the norm in just a few years.

Jerry’s definition of design (or document) synthesis was quite simple — it was the processing of the information found in one document to produce or update another document where that processing was not simple translation. In the world of electronic design, this meant writing a document that described the intended behavior of a circuit and then having a program that would create a manufacturable design using transistors, capacitors, resistors, etc. from the behavioral description. In the software world, we’ve been using this same process for years, writing software in a high-level language and then compiling that description into machine code or bytecode. For hardware design, this was a huge change — moving away from the visual representation of a schematic to a language-based representation similar to a programming language.

In the field of legal informatics, we already see a lot of processes that touch on Jerry’s definition of document synthesis. Twenty years ago, it was seeing how automatable legislation could be, but wasn’t, that convinced me that this field was ready for my skills.

So what processing do we have that meets this definition of document synthesis:

• In-context amending is the most obvious process. Being able to process changes recorded in a marked up proposed version of a bill to extract and produce a separate amending document
• Automated engrossing is the opposite process — taking the amending instructions found in one document to automatically update the target document.
• Code compilation or statute consolidation is another very similar process, applying amending language found in the language of a newly enacted law to update pre-existing law.
• Bill synthesis is a new field we’ve been exploring, allowing categorized changes to the law to be made in context and then using those changes and related metadata to produce bills shaped by the categorization metadata provided.
• Automated production of supporting documents from legislation or regulations. This includes producing documents such as proclamations which largely reflect the information found within newly enacted laws. As sections or regulations come into effect, proclamations are automatically published enumerating those changes.

In the CAD world, the move to design synthesis required letting go of the visually rich but semantically poor schematic in favor of language-based techniques. Initially there was a lot of resistance to the idea that there would no longer be a schematic. While at University, I had worked as a draftsman and even my dad had started his career as a draftsman, so even I had a bit of a problem with that. But the benefits of having a rich semantic representation that could be processed quickly outweighed the loss of the schematic.

Now, the legislative field is wrestling with the same dilemma — separating the visual presentation of the law, whether on paper or in a PDF, from the semantic meaning found within it. Just as with CAD, it’s a necessary step. The ability to process the information automatically dramatically increases the speed, accuracy, and volume of documents that can be processed — allowing information to be produced and delivered in a timely manner. In our society where instant delivery has become the norm, this is now a requirement.