Software Is Eating Law

Alexis Chun
May 7, 2018

There are gaps in the legal industry that are ready for innovation, says Alexis Chun, the founder of Legalese, a company that aims to apply computer science to the law. Chun gave a talk earlier this year at TechLaw.Fest 2018 in Singapore, along with ROSS Intelligence CEO and Co-Founder Andrew Arruda and other luminaries in Singaporean legaltech. The following is a recording of her talk at the event, along with her slides and a written transcript.


Video Transcription:

Hello. Thank you ROSS Intelligence, and thank you, Andrew, for asking and trusting me to do this. I’m immensely honored and grateful and must apologize for the background noise, because it’s absolutely pissing it down in Singapore right now, where I’m recording this.

I’m Alexis, and I hail from a little project called Legalese. We currently have our home base in Singapore, because that’s where my co-founder Meng and I are from. We’re also open-source, fiercely international, aspirationally teal, if you’ve read Reinventing Organizations, and softly holacratic. But that’s Legalese the project, Legalese the company.

Today, I’m going to give a sort of big-picture idea about “Software is eating law,” because this is something that we’ve been banging on about for quite some time.

To do this, I’d like to enlist a couple of men to help me get the idea across. For the avoidance of that, I’m being at least 15 percent facetious.

So, these six men that Kipling wrote about: What, Why, When, How, Where, and Who.

Let’s start with an easy one, let’s start with the what.

Specifically, what do people do when they need legal help? Well, they go to a lawyer. Perhaps even someone a bit like this one. That’s me on a good hair day. And if you look at the TL;DR of my legal background, that’s what we at Legalese call the Lawyer Version 1.0. this person would have worked in various places, done some sweet secondments and consulting gigs, so that means the clients like them or trust them. They went to law school and got called to the Bar and did alright in all of those things.

But, if we just explore this a little bit further and think about what really happens when they go to a lawyer, it’ll look something a bit like this:

Well, lawyers charge about $600 an hour for input, not output.

If you think about it, it puts us in a rather awkward position because lawyers serve the clients on one hand, but on the other are compensated by the law firm on a billable hour structure. That means the incentive structure is such that the lawyer is incentivized to be the most inefficient lawyer that they can get away with. That’s how they pay the bills, that’s how the partners pay for the third yacht, and that overcompensating birthday party for the kids that they have but never quite see.

Being humans, as they are, they also make mistakes.

They’re error-prone, and they can occasionally overlook considerations, capture unintended weakness or inconsistency. They have bad days. They have deadlines that they have to rush for. And sometimes it really just depends on whether they’re on their first cup of coffee or the fifth.

They have an imperfect recall, they rely a lot on experience.

And on the other hand, they also rely a lot on professional indemnities to cover their asses and a whole kitchen sink of assumptions, which if you’ve seen the ledger of engagement from a law firm or from a lawyer, you’ll know what I mean.

Of course, some things have changed, right?

A hundred years ago they used a typewriter, and today they use Microsoft Word.

That’s technology.

But that’s still kind of an awkward realization. 100 years on, not much really has changed if you realize that lawyers still use technology to only help with the typing but not with the thinking.

And if one may be permitted to speak frankly, in a world:

  • Where Cars drive themselves (arguably)
  • Where software flies our planes (arguably), and
  • Computers are beating doctors

It seems quite mad to pay a human $600 an hour to copy and paste words on a page. Some of these folks certainly think so.

Which brings me to the next two points: Who are the alternatives, and where?

These are quite easy, so I’m going to tackle them together.

Where I’m from, Singapore, which is a tiny island in Southeast Asia, we see Legal Zoom exploring the space with us where they’re offering U.S. legal help from the U.S. for businesses and families in Singapore.

So very clearly the where is everywhere to everywhere.

But who are these players?

The landscape is fairly well mapped out. Thomas and Reuters has an entire legal texture of the landscape, so there’s Legal Geek who are based in London, but they also map out the legal startup space all over the world.

Closer to home, at least where I am in Legalese, we list down in our website the legal tech startups and people that we’ve encountered. And also chiefly at the bottom at debt pool to remember those who have since left us.

And I see you thinking, “Gosh, that’s a lot of stuff that’s been going on. When did all that happen?”

And…. I’m afraid I have to break it to you that it’s been going on for quite some time. The idea of software eating law has been around for at least 60 years.

This is a paper from the Yale Law School from 1957, and talks about how symbolic logic is a “Razor-edged tool for drafting and interpreting legal documents.”

In 1960 there was the First National Law and Electronics Conference in California, where it says, and this quote is quite pretty, it says: “The day should come, when it will come I don’t know, but the day should come when you will be able to feed a set of facts to a machine that has cases, rules of law, and reasoning rules stored in it, and in which the machine can then lay out for you, step by step, the reasoning process by which you may be able to arrive at a conclusion. You can then study it and then decide whether the machine is right or wrong. In some cases, the machine may not tell you exactly what the conclusion may be, but may say there is a probability that such-and-such is correct, and this probability is 90 percent.”

This was in 1960.

In 1986, the British Nationality Act was codified as a logic program.

A couple of months back, Meng and I found this textbook that was published in 1979 called Computer Science and Law.

What’s interesting, we found, was that every chapter, as you can see in the contents page, has become a legal tech startup or domain today.

  • Chapter Three, The taxman project, is TurboTax.
  • Chapter Four, modeling of legal rules, has led to the expert system Chatbots of today.
  • Chapter 13 is the grandfather of an entire industry, which we call legal document assembly software.

And many academic papers have also been written about the idea of software and law.

So it’s really not new, which brings me to the next question of why?

  • Why software & law?
  • Why is software eating law?
  • Why the isomorphism?

Our insight at Legalese, or suspicion, is that lawyers are kind of like the lost tribe of programmers. There’s a close parallel to be drawn here if you bear with me.

50 years ago, software was proprietary, much like law today. There was no high-level languages, there was no open-source, there was no Git. If you wanted something done, you hired a team of programmers to do it from scratch, and fees would start at $100,000.

That’s where most law firms are today.

Most corporate departments in law firms are basically a small team of assembly programmers writing proprietary code by hand. And if that is true, then it might be pretty easy to predict and extrapolate the next few decades of the legal industry and software factors in it.

Our suspicion is that the legal industry’s future will look a lot like the software industry’s past.

Today, lawyers use different words to mean the same thing, or the same words to mean different things.

That’s also called pseudocode.

Don’t believe me? Let’s try a simple example.

“Tabling” something. We’re not going get into multilingual translations and all that math for now. We’re just going restrict ourselves to one language, English.

What does it mean to table something?

Well, the very lawyer answer is it depends.

And it does.

If you’re in the U.K. it means let’s table it, let’s talk about it.

In the U.S. it means let’s never talk about it.

I’m going let that sink in for a second.

So if you think about it, the isomorphism, it’s quite clear. Because contracts and programs are both fundamentally specifications for the distributed execution of business processes.

Except, as I’ve showed quite limitedly but nonetheless lawyers have a language problem.

Which brings us to the last serving man, How.

How will software eat law, and how do we get there?

We enter the domain of computational law, which is something very close to what we do at Legalese, and we start by designing a domain-specific language for law. That is a programming language specifically designed to capture the semantics, deontics, and pragmatics of law. It would be a cross-discipline synergization of over 70 years of academic research and industry insight.

It’s an intersection between computer science, law, mathematics, and logic.

And I see you asking, “So? So what if you’ve got a programming language that’s designed to express law?”

Well, because it’s specifically designed to express law, you can then use that language to write law. And when I say law, I don’t just mean contracts. I’m also talking about the statutes, the regulations, the state rules, the business logic, the business processes, and also to quasi-legal documentation that are all involved with executing legal contracts. If you can write all of these things that make up law, that means they all have a common denominator, which allows you to go from pseudocode (natural languages) to real code.

If I put on my programmer hat and I look at legalese like this, I think, “Surely we can take these kinds of pseudocode out of Microsoft Word to a text editor, add some curly braces, pretty syntax highlighting, and perhaps this is what an agreement really wants to be and really can be when it grows up. It can be a program. Because with a program, you can get from syntax (which is what does it say) to the semantics (as in what does it mean) to the pragmatics, which is really what the clients want to know about and want to be advised on. What does it mean for me?

If you’re still underwhelmed, think about this.

We can now apply the batteries of tools that computer scientists have at their disposal to law. These are tools that lawyers don’t even have names for. Over the last 50 years, software engineers have developed a massive suite of tools to make their lives easier. Look at all of these things.

In comparison, what does the legal stack look like?

Track changes, Microsoft word, Latin.

But if you look at all of those gaps that I’ve marked out with opportunity, they are opportunities for innovation.

When the laws change, for example, we can call that a software update. We can even make it a self-updating package; it’s been done before!

  • If your documents depend on each other, that’s not a problem. We can build in dependency management between them.
  • We can even layer the development architecture modularly the way software development already does today.
  • We can do static analysis.
  • We can test and calculate the effect of an immediate change to your system.
  • We can test code without executing it. That means finding bugs at compile time, because bugs at runtime, that’s called litigation.

That’s kind of like magic, isn’t it?

And a rather well-honed software tradition. We see that with the printed page and design with postscript, databases, 3D drawings, accounting, chip verification, and at Legalese that’s what we hope to do with law.

Andreessen said (I hope you recognize who this is) “Software is eating the world.” And that’s where we got our “Software is eating law.

And also, a couple of years ago, on my birthday, my co-founder Meng went on Twitter to have a proper rant about how law should be eaten by software and can be eaten by software. And Andreessen replied and said, “Bingo.”

That’s one of the stories we tell about why we decided to do a startup, because Andreessen said “Bingo.”

So that’s all from me.

You either die a lawyer, or you live long enough to see yourself become the villain, as I have to my ex-colleagues.

Thanks for having me! Please feel free to write to us or tweet us if you’ve got any questions or you want explore any of these further.

Thank you for your time!

Alexis Chun

A recovering lawyer and co-founder of Legalese (it’s cheaper than therapy). Legalese is marrying computer science with legal by designing a domain-specific language for law.