An industrial revolution is coming to law
8 October, 2015
This post explains one way an industrial revolution is coming to law. Expert systems are changing many aspects of how law is practised. Michael Mills, a leading authority, explains how.
Bankers who pitch business to customers around the world are subject to restrictions in each country they target. The regulations are intricate, though not analytically complex, vary considerably from one country to another, and change from year to year. Penalties for violation can be more than a minor nuisance—including a ban on doing business—and reputational risks with regulators are real.
How do bankers avoid tripping over the restrictions? They don’t read 1,000-page compliance manuals. They may ask compliance officers, who may hand them a few pages from the manual. Or they may wing it.
This is an application designed to deliver convenient, practical, on-the-run advice to the bankers, and thereby avoid non-compliance and reduce risks. It was built with our software, the Neota Logic Server.
How the industrial revolution is coming to law
What is this application? It’s an expert system, which we define as:
experts’ knowledge & judgment
into practical answers
other people can use.
Conceptually, it looks like this:
Expert systems are everywhere: evaluating your medical test results on an iPhone in the pocket of your doctor’s white lab coat; and doing your income taxes. Expert systems are driving the industrial revolution in the professions, law included.
The curse of constant cost
Let’s look at a diagram—the revenue and cost lines in our profession.
What happened to these lines in the Great Reset?
First, the top line was shoved downward by cost-conscious general counsel.
No surprise, law firms cut costs—mostly support staff and space—to push the cost line down and preserve margin. Alternative legal services providers like Hive Legal brought reduced costs to market too. Flexibility as to assignments and location has attracted very highly qualified lawyers, often from the best traditional firms.
Second, we got process engineering and project management. With these tools, we have the least-cost people doing only the stuff the client really needs, in the right order, with minimal handoffs and rework. And we shoved that red line down again.
So, this is better, but not good enough. What’s still wrong?
The straight lines.
The cost per unit of output has shifted downward. An hour costs $600 instead of $800, and that hour produces two-tenths of a contract rather than one. But when quantity climbs, total costs climb at exactly the same rate. Every unit of output costs the same as the last one.
The happy curve of declining average cost
That’s not how the big boys do it.
The great engine of Microsoft is that the marginal cost of delivering another copy of Windows is close to zero. Not zero, there’s lots of R&D for the next version. But the average unit cost declines as quantity rises, so quantity is a good thing.
Yes, more hours from a fixed-cost resource, i.e., associates, reduces the average cost, but only up to the limits of human capacity for working without sleep. When that limit is reached, the curve is stuck.
Economies of scale, as Henry Ford taught us.
Alas, quantity in legal services is not necessarily a good thing. We have the diseconomies of scale—internal coordination costs, quality variation—but not enough of the economies, other than branding and cross-selling (when it works).
In short, law practice missed the industrial revolution. We didn’t build power looms, and we certainly didn’t build Jaquard looms, programmed by holes in paper cards (the model for the 80-column, cropped-corner punch cards of computing’s adolescence).
Forget about billable hours, alternative fees, and ABS’s. The problem is constant cost.
Richard Susskind’s illustration of what he posits is the inexorable evolution of practice is well-known. It tracks from the bespoke (the one-off, novel, inventive, bet-the-company stuff we all love to do) along the line through standardized (e.g., a forms library for our lawyers) to systematized (checklists for deal types, project management) to packaged (defined as a service, offered to clients on a fee basis disconnected from hours). But not to commoditized (the work product is indistinguishable among practitioners)—that’s LegalZoom, and that’s not us.
It is, of course, the challenge of law firms and their leaders to keep moving ahead, to spot new products and opportunities, to capture the new and accept that the old will wither away.
This analysis tracks roughly to Jeff Carr’s well-known bucketing of legal work:
He says Process and Content trend to free, so companies should not pay for it, or not pay much. I disagree that process and content are to be disdained as a cheap commodity. To the contrary, there is deep and valuable expertise in both process and content. Nonetheless, the distinction is useful here because it illustrates the line between activities that can be systematized and packaged (on the right) and those that, for the most part, cannot (on the left).
Another way of looking at the industrial challenge in law practice is this:
Years ago, companies tended to give their loyal outside counsel most of the work in the middle and upper ends of the bell curve. However, as costs rose, more of work to the left of the middle red line was taken in house or sent to alternative legal services providers. Firms lost share with their clients. One can view innovation around process and content as an opportunity to recapture share, to recapture revenue in offerings that are systematized and packaged.
Law as code
Law doesn’t need blast furnaces or warehouses. We have the pure digital product—lightweight, malleable, portable, computable.
Ray Kurzweil’s picture shows computers getting smarter. They are, though I’m not waiting for the Singularity:
Thanks to Moore’s Law, we’re nearly up to mouse brain level. Fortunately, that may be enough for what lawyers do—which is, indeed, rocket science but actually doesn’t tax modern computers very hard.
Traditionally, lawyers have seen law and all its artifacts as text to be read by humans. Today, in a computational world, we can see law as data and algorithms.
Machine learning and other A.I. tools
It’s true that “Artificial Intelligence is whatever hasn’t been done yet,” but A.I. is a fertile field in many disciplines—robotics, machine learning, and expert systems among them. We lawyers don’t build cars, so we don’t need physical robots.
Machine learning—the forest of amazing algorithms that can master big data (and not-so-big data too)—is truly a marvel in all its many manifestations, from giving vision to robots to helping oncologists create better treatment protocols.
Machine learning has come to law. It is taking over e-discovery, indeed would have taken over entirely but for the Luddite resistance of many lawyers and clients. It is beginning to do useful work in other contexts, such as Lex Machina for patents and Kira Systems for contracts.
Machine learning could be doing useful work analyzing the (pretty big) data sets you have in your firms—clients, matters, time, task codes, narrative time descriptions, documents, email (especially email), HR systems, and so on.
Here’s an example in employment law—using big data techniques to analyze issues around immigration and industry.
Maybe, as IBM’s general counsel said last year, Watson will pass the bar exam in a few years. (Or perhaps before Watson signs up for his bar review course, the profession will finally concede that the exam is a worthless measure of fitness to practice, as Watson’s ability to pass it would prove.)
Neota Logic works with machine learning systems to induce rules from big data sets, but most of the time we work with human experts to represent their expertise and their judgment in software so it can be applied by their clients at scale. That expertise can be of at least three kinds:
- About substantive law, regulation, and policy—what’s the operationally useful answer to this (relatively) routine question? (Clients say, “We want answers, not memos.”)
- About process—what is the most efficient, lowest-risk, highest-return method of doing this deal or litigating this case?
- About documents—what text in this contract best confers the rights and obligations the client wants? what risks do these provisions create?
In these contexts, the black-box, probabilistic algorithms of machine learning are not by themselves sufficient, because people want to know:
- what to do
- about this problem
- in these.
Lawyers are determinists and skeptics. Clients are result-driven and busy. They want answers. Expert systems are transparent, auditable, defensible, and repeatable.
What problems are suited to legal systematics?
This diagram emerged from conversations with several general counsel about how they see the landscape of problems with which they must cope. If it’s a critical problem, top left, of course the answer is “hire the best, roll the troops, start up the engines.” In the middle and lower right sectors, problems are individually less consequential but far more frequent and in the aggregate highly consequential indeed.
Example—Foreign corrupt practices
In order to assist companies with global operations to comply cost-effectively with the U.S. Foreign Corrupt Practices Act (and similar laws and policies) when they do not have dedicated FCPA compliance staff, as only the very largest companies do, the firm of Foley & Lardner devised a new practice offering that combines technology with traditional but technology-enabled legal services.
Example—Independent contractor classification
Classifying a worker as an independent contractor or an employee is an increasingly critical determination, as state and federal tax authorities and regulators increase audit and enforcement activity. An expert system can aid in that determination and produce a defensible audit record.
Summing up Neota Logic and the industrial revolution in law
This is what Neota Logic does:
We enable law firms to
leverage without associates
and bill without hours.
The industrial revolution isn’t actually coming to the practice of law – it’s here now.
About the author, Michael Mills
Michael Mills is the co-founder and chief strategy officer of Neota Logic Inc., developers of a software platform that enables lawyers and other professionals to build expert systems. Michael was for 20 years the chief knowledge officer and co-head of technology at Davis Polk & Wardwell, after practicing as a litigation and bankruptcy partner at Mayer Brown.
About Neota Logic
Neota Logic creates software for expertise automation, applying rules & hybrid reasoning to to professional advice, decision management, and intelligent businessprocesses—in law, human resources, compliance, tax, , financial services, insurance,and other fields. Neota Logic applications reduce risk, increase efficiency, and ensurecompliance for businesses, government, and not-for-profit organisations.
This article was first published on 27 February 2015 in Neota Logic’s Resources site.
About the author