Interview by Tristan Koh and Ong Chin Ngee | Edited by Tristan Koh, Ong Chin Ngee and Josh Lee
TechLaw.Fest 2020 (“TLF”) will take place online from 28 September – 2 October 2020, becoming the virtual focal point for leading thinkers, leaders and pioneers in law and technology. In the weeks leading up to TLF, the LawTech.Asia team will be bringing you regular interviews and shout-outs covering some of TLF’s most prominent speakers and the topics they will be speaking about.
This week, LawTech.Asia received the exclusive opportunity to interview Wong Meng Weng, Principal Investigator of Singapore Management University Centre for Computational Law and Co-Founder of Legalese. Meng Weng will be speaking at the Knowledge Cafe on “What Computational Law Can Do For You” on the third day of TLF (30 September 2020).
The SMU School of Law recently opened its Centre for Computation Law and Legal Technology. For the uninitiated amongst us, what is computational law?
When people need help thinking through a quantitative situation, what do they use? A spreadsheet. But if people need help thinking through a legal situation, they turn to people (i.e. lawyers), rather than a tool. People trust lawyers to know the law, deal with contracts and advise them on various scenarios. However, lawyers are a valuable and expensive resource. Many want to get the answers they need without having to bother a lawyer and pay for their service. Computational law seeks to develop software to make that possible.
This leads to the question: How can computers answer legal questions?
Let me offer some analogies. In computational games, we say to a computer: “Here’s a chessboard; what’s the best move?” Now computers play chess and the game Go better than any human. In computational driving, we say: “Help me stay in my lane, at the right speed.” Mercedes-Benz, Volvo, Audi, and Tesla sell thousands of cars every day with Level 2 autonomy that are self-parking and able to change lanes at speed.
Computational law is similar. Questions like these occur every day in the legal industry:
"I want to achieve X outcome; what paperwork do I need to file?"
"I want to perform X action. Given all the contracts and laws that govern my day-to-day life, may I?"
"Some event X has occurred; what should the various parties do ?"
For centuries, clients have brought these questions to lawyers. Contracts and laws are complex. Most people sensibly outsource the work of knowing the rules, keeping up with developments and applying them to specific situations to trusted advisors.
Computational law observes that the complexity of laws and contracts is similar to the complexity of computer programs. We explore the possibility that, if our contracts and rules could be encoded in a form which a machine can understand, these kinds of legal questions could be answered by machine.
However, this approach differs from much of what is on offer in legal technology today. Some legal technology solutions are legal marketplaces or practice management tools; they are Grab and Asana for law. Other legal technology solutions are Google for law: legal analytics and contract analytics software fundamentally analyse reams of input text and offer prioritised search results. Natural language processing (“NLP”) and machine learning (“ML”) methods power the engines behind GPT-3, a ML model recently released by OpenAI. So far, GPT-3 has produced entertaining speeches and literary fragments. I wouldn’t mind reading a poem written by GPT-3 – but doubt that most would be comfortable signing a contract written by GPT-3.
We draw inspiration from Michael Genesereth, a giant in this field. He set a vision for the legal equivalent of the spreadsheet. Just as spreadsheets are widely used by people from all walks of life, we expect that our software will eventually benefit both lawyers and non-lawyers. We hope it will be a democratising and empowering “digital transformation.”
As an academic research programme, we have the luxury of looking a decade into the future and taking a somewhat contrarian position anticipating the next swing of the pendulum. Where much of legal technology today uses NLP to decipher English into some kind of structured representation, we rely on natural language generation (“NLG”) to turn code into English. We also borrow from an arcane discipline within computer science called “formal methods”, which offer powerful ways to detect bugs in programs before they are run. We adapt these methods to attempt to find loopholes in contracts and laws before they are signed or passed.
You also operate a start-up, Legalese. In its website, Legalese talks about the notion of “rule as code”. Do you foresee that when this notion becomes a reality, it will be seen as an evolution from current legal technology solutions or a quantum leap in capabilities (and thus a form of disruption)?
“Rules as code” is an international movement in legal drafting and policy-making. The idea is that some rules and regulations could be formulated with “digital twinning” in mind: the English-language version and the machine-readable version of the rules could be constructed in parallel. Low-hanging fruit for this approach can be found in domains like tax, finance, and corporate law, where many laws already read like the word problems we remember from mathematics class.
Encoding rules in machine-readable form is a necessary step toward the vision of computational law. With those rules encoded, it is possible that the task of interpreting and implementing such regulations will shift from the in-house counsel’s office to the CTO’s IT department.
Earlier this year, the EU announced a 500,000 Euro tender for exactly such a project: they commissioned a machine-readable language for regulators to disseminate reporting requirements to financial institutions. One requirement was isomorphic representation – in other words, fidelity to the English regulations – while being open-source and readable by any bank needing to report over-the-counter derivative trading activity. Currently those regulations are represented as several pages of PDF text and Excel spreadsheets which draws out the compliance implementation process to several months. If they were in a “Rules-As-Code” form, they could be integrated directly against trading software, potentially increasing the speed and quality of compliance while saving millions in implementation costs. So “RegTech” is one domain of application for computational law.
Speaking with my startup hat on, we believe that there is room for a company to be built in the size and shape of Adobe. In other words, we envision a world that has Excel for numbers; Adobe for graphic design; AutoCad for architecture; and Legalese for law.
We believe that the software stack we are building will one day serve as the foundation for a suite of tools and applications that will help solve access-to-justice and access-to-law problems in ways that are fundamentally digital. While others seek to solve access-to-justice by throwing more pro bono hours at legal clinics, we want to put a legal clinic on your phone or your web browser – no need to trouble a lawyer at all. Already we are building integrations with “self-help” tools like DocAssemble. We are also designing various tools for scenario exploration, contract visualization and expert system explainer which help end-users deal with the simple “20%” scenarios that come up 80% of the time.
We believe that the software stack we are building will one day serve as the foundation for a suite of tools and applications that will help solve access-to-justice and access-to-law problems in ways that are fundamentally digital.Wong Meng Weng, Principal Investigator of Singapore Management University Centre for Computational Law; Co-Founder of Legalese
Frederk Pohl said: “A good science fiction story should be able to predict not the automobile but the traffic jam.” We look beyond the technology towards the changing social institutions of 21st-century legal practice. Openness is a hallmark of the computing profession. A beginner programmer is confronted with a wealth of resources which is simply unmatched in any other field: that’s why there are so many self-taught programmers and so few self-taught doctors. Many people have called for a “Github for contracts”, and we believe that such a resource will naturally arise once we create the conditions for it. That is why our research programme at SMU is focused on developing a domain-specific language and libraries for computational law, which will let us express the syntax, semantics and pragmatics of laws in code.
When computational law becomes reality, what are the implications on legal education and training for present and future generations of lawyers? Does this make it necessary for students to pick up some rudimentary knowledge of coding or computational thinking? What are the necessary skills for the next generation of lawyers?
Richard Susskind’s book, Tomorrow’s Lawyers, forecasts a number of technology-driven innovations. These innovations are technologically feasible. What we need is collaboration and commitment to make it happen.
One of the first applications of computational law is already widespread: accounting software which people use to file taxes. In Singapore, that software is found in the Inland Revenue of Singapore’s systems. In the U.S., that software is retailed to consumers in the form of applications like TurboTax. In both cases, the computer understands tax law as well as any human: it knows the rules, it knows the exceptions, it knows how to optimise.
We envision a future where such tools are available not just for taxes but for every genre of laws and contracts. The implications for the “customers” of the legal industry are exciting. In that future, consumers of legal services are empowered to deal with mundane and simplistic matters themselves, saving their engagements with law firms for higher-value, more complex problems that challenge lawyers to operate “at the top of their license.”
If you ask a professional: “what software could you not live without?”, a photographer might answer “Photoshop”. An architect would say “AutoCAD”. A business executive would say: “Excel”. Each of these software packages is a tool for thinking, a device for intelligence augmentation. But what is the corresponding tool for lawyers? Word doesn’t help a lawyer think, not the way Excel helps an accountant think. So the gap is obvious if you look at it this way.
What made you get involved in computational law and what would you say to law students / practicing lawyers who are interested to take a deeper dive into this space?
As a customer of legal services I have paid six-figure sums to law firms over the years. I see these fees as friction in business and a burden to individual citizens and consumers. My background is in Internet infrastructure innovation, developing and deploying protocol and software standards at Internet scale. I see legal as a problem in Internet infrastructure (they call this the “Law of the Instrument”: I’m holding a hammer, everything is a nail.) The rituals of wet-ink signatures are being displaced by e-signatures; paper is being replaced by PDF; and that is only the beginning!
My lawyer friends tell stories of tremendous waste and inefficiency in legal services. The labour-intensive law firm was an unavoidable business structure at a time when all knowledge work was performed by humans. But we have already entered an era when knowledge work is routinely performed with the assistance of software. To quote Andreessen Horowitz, “Software is eating the world.” The legal profession is one of the last holdouts. As an entrepreneur I find it fascinating to encounter an industry which appears to me to be actively resisting productivity tooling. I am of the view that the legal industry is ripe for disruption: sooner or later some product is going to appear that is complete enough to sell not to law firms, but past them.
What does this mean for students and practising lawyers? One strategic goal of the Centre for Computational Law at SMU is to provide support for a nascent cluster of the next-generation legal technology industry, right here in Singapore. We believe that for Singapore to maintain its position as an international hub for financial and legal services, it is necessary to get ahead of the curve in legal technology innovation. The continued shrinking of the pool of positions available to law-school graduates may be an irreversible sign of the future of the traditional law firm. The encroachment of accounting and consulting firms into areas previously thought exclusive to law firms is another such sign. We believe that the latest generations of law students, who have had some exposure to computational thinking as part of their schooling, may find alternative careers in legal engineering, careers that look more like software development and IT “devops”, than the traditional progression through a law firm as associate to partner, or entering in-house.
Today, lawyers who receive a PDF will often write back and say, “Would you mind sending a Word document that I can edit?” Tomorrow, we believe lawyers receiving a Word document will ask for a contract written in code. While they will still engage in contract review, drafting, negotiation, and legislative interpretation, the way they do these things will be supercharged by the tools we will offer them, so there will be less drudgery and tedium.
As William Gibson said, “the street finds its own uses for things“. If tools originally developed for lawyers are embraced by non-lawyers, we certainly cannot stop them.
We hope that a few years from now when our language is ready, we will have opportunities to teach computational law in schools, and empower graduates to deliver legal services with the assistance of our tools, in ways that (to borrow Clayton Christensen’s term) are disruptively efficient for currently un-served segments of the market.
This piece of content was produced by LawTech.Asia as an official media partner for TechLaw.Fest 2020. Click on the banner below to access TechLaw.Fest, and register for the event (it’s free – for limited slots only)!
Editor’s Note: A longer version of this interview can be found at https://github.com/smucclaw/complaw/blob/master/doc/interviews.org.