Editor’s Note: The author of this post is a law professor whose research focuses on decision-making in the context of emerging legal technology.
By Brian Sheppard, Associate Professor at Seton Hall University School of Law
According to business wisdom, you must “know what you’re selling.” Many products make it easy by having well-defined “specs” that are part of the sales pitch. You can sell smart phones pretty well knowing little more than a few facts about memory, resolution, and battery life. When it comes to legal tech, however, knowing what you’re selling means understanding a complex activity – legal interpretation.
This requires some explanation. There are a lot of different kinds of products that fall under the umbrella of legal tech, but the vast majority of them seek to improve the performance of legal services. We call them “legal” services because they target people who understand that their activities are governed by law but don’t yet understand how the law should affect their decisions. Legal service providers help them understand. Legal tech companies seek to improve that process.
Some companies do this by providing greater access to the law. Research products like Ravel provide a vehicle for rapid retrieval of legislation or court decisions that support legal arguments for use in disputes. Other companies do this by putting the law in simpler terms. Document preparation software like Legalzoom provides computerized forms and plain language instructions that facilitate the creation of legally compliant instruments. Still others provide ways for machines to perform the difficult work of applying the law to a high volume of facts. E-discovery products like Veritas help determine which items are responsive to litigant interrogatories and are protected by privilege. Finally, some companies provide analysis of the available options when a legal dispute arises. Legal analytics companies like Lex Machina provide predictions of case outcomes that help people decide whether to pursue litigation and use outside counsel.
Importantly, each of these companies contemplates a significant role for legal interpretation in their client services. Some seek to improve it, some seek to do it for you, and some seek to predict it. To know what they are selling, then, legal tech companies should understand legal interpretation, particularly its capacity to be improved.
This is where things get tricky. The power and limits of legal interpretation have long been the subject of controversy.
Consider a topical example: the divergence in two recent Supreme Court confirmation hearings. Chief Justice John Roberts, a Republican nominee, famously stated during his opening remarks that “[j]udges are like umpires,” that “[u]mpires don’t make the rules, they apply them,” and that he “will decide every case based on the record, according to the rule of law.” Compare this to the statements at the hearing of Associate Justice Elena Kagan, a Democratic nominee, where she said, “The [umpire] metaphor might suggest to some people that law is a kind of robotic enterprise” and “we go ‘ball’ and ‘strike’ and everything is clear cut” but “that’s not right.” Roberts is arguing that laws are clear and, even at the Supreme Court, generally can be interpreted mechanically and objectively without having to create new rules on the fly in order to resolve a case. Kagan is arguing that laws are often indeterminate, particularly in cases before the Supreme Court, requiring interpreters to use complex and (often subjective) resources like value judgment and creativity.
This is the most accessible version of a debate that has existed for at least a century — Formalism vs. Realism. At the risk of oversimplification, Formalists believe that law is a rational system that gives legal interpreters the power to determine one correct answer to all (or nearly all) legal disputes. A Formalist is likely to conclude that legal interpretation is friendly to mechanical and often deductive, logical techniques. By contrast, Realists claim that the law is incapable of providing these right answers in a very high number of legal disputes (particularly those that make it to the appeals courts), leaving it to people like judges to rely on other things to resolve them, such as their own personal morality. A Realist is likely to believe that science can provide inductive methods that tell us how these non-legal factors will influence the resolution of legal disputes.
Legal tech companies have, knowingly or not, adopted positions in the Formalist-Realist continuum.
Several companies that aim to automate legal interpretation have adopted the tenets of Formalism. For them, playing up the systematic and organized dimensions of law makes sense. In that vein, Judicata, a legal text visualization platform, claims to be “mapping the legal genome.” Competitor, Ravel Law, similarly declares that it allows users to “unravel the law” by “turn[ing] legal data into legal insights.” Likewise, Shake, a computerized legal document creator asserts that it is “making the law accessible, understandable and affordable” through technology and aims to give legal agreements the “simplicity and convenience of a handshake.”
If computers are going to be able to handle legal interpretation at current levels of technology, it has to be the case that legal meaning can be discerned using rather simple and mechanical techniques. It is hardly surprising, then, that Robot, Robot & Hwang, a legal automation software developer claims, “[d]rawing from [the] commandment that code is law, we believe equally strongly that law can be code.”
Conversely, many companies that sell legal analytics services or software have sided with Realism. Several of them claim that their data on individual judges allows them to predict legal outcomes through proprietary algorithms. This “moneyball” approach is agnostic when it comes to what the laws governing a case mean and maintains that non-legal factors – like the identity of the deciding judge – have a significant bearing on how a case will actually be resolved.
Whereas Formalist automated legal interpreters base their value proposition on their ability to access the inner logic of law, legal analytics companies view law as just another datapoint in the prediction game, shunning the inner legal perspective. It unsurprising then, that some of these companies embrace the opacity and mysteriousness of the law in selling their products; it makes their end-run around legal meaning all the more attractive.
For example, Quovant emphasizes, in its words, “the uncertainty in modeling the financial implications of results, and the variability caused by statutes and precedents” as well as“the modern practice of US law . . . based on statutes and precedents that are hard to model for the purposes of predictive analysis.” Similarly, Argopoint claims on its website that “[t]he legal environment, with its constantly changing components and variables, may seem like an area where it is difficult to quantify activity and results.”
There are even shades of Realism’s characteristic non-conformity in some of the promotional language for legal analytics. Lex Machina explains how adding its “bottom-up data” to the “traditional, top-down” Formalistic approach is “new, it’s unorthodox and it’s extremely valuable.”
Does this mean that legal automation companies are conservative and legal analytics companies are liberal? Of course not. These companies seek to sell products, and adopting a potentially exclusionary political outlook is probably bad for business. Besides, we shouldn’t overlook that some of these companies, such as Ravel, sellbothautomated legal interpretation and legal analytics products.
But is it possible to occupy both sides of the philosophical spectrum? The trouble is that the value of these products is tied to whichever of these competing and politicized positions is right. Broadly speaking, the more that the dictates of law, alone, determine the outcomes of cases, the less valuable analysis of non-legal factors, like individual judge data, will be. If the law provides obvious correct answers, then there is little value in paying a legal analytics company to crunch data on individual judge preferences.
Looking at it the other way, the less law is systematic and rule-based, the less valuable a computer-based interpretation of the law will be, at least until computational methods improve. A highly variable or ambiguous legal system limits the value of software that uses natural language processing to interpret legal texts because it reduces the reliability of computerized interpretations. The result is that those products are confined to the most simplistic and straightforward dimensions of law.
We can surmise that companies are not eager to choose a side in the Formalist/Realist debate. They might hope instead to stake a position in the center.
Fitbit provides a helpful example. Because it aims to make it easier for people to be physically fit, business wisdom dictates that it should understand what fitness is. The problem is that the parameters of physical fitness are debated. Even recently, studies about whether there is such a thing as too much exercise and whether one can be fit but overweight have created quite a stir. Yet, Fitbit, a successful company, has been able to able to duck the issue. It doesn’t seek to define “fitness;” rather, it modestly asserts that knowing how much you move its wristbands will help you “achieve your health and fitness goals, whatever they may be.” This is possible because, despite disagreement at the fringes of the concept of fitness, there is a core of agreement. Almost everyone agrees that being fit is linked to physical activity and that moving your body (and any attached wristbands) is a physical activity.
The question is whether there is an uncontroversial core at the center of the legal interpretation debate. Maybe so. Regulators of the legal profession generally assume both that a typical legal dispute allows lawyers to raise opposing arguments and that lawyers can tell the difference between frivolous and non-frivolous arguments. In other words, they believe that we might not easily know what the single correct answer is in a case, but we can narrow it down into some decent opposing options.
Unfortunately for legal tech sellers, moderate positions like this aren’t very impactful. You won’t set the world alight by saying that your software converts the law into plain language that, while potentially wrong, is at least non-frivolous. It’s much more exciting to take a page out of Logickull’s playbook and say something like,“We solved it. We automated discovery.”
At this point, it is helpful to remind ourselves that legal tech is very young, and intemperate rhetoric about a product’s capabilities might be needed for a startup to make a splash. As the market matures, perhaps we will see more companies shift towards the middle. Or we could wait to see which side of the debate prevails.
The views expressed in this column do not necessarily reflect those of Big Law Business or its owners. [Image “For more legal resources, visit Bloomberg Law.” (src=https://bol.bna.com/wp-content/uploads/2016/03/Screen-Shot-2016-03-09-at-5.43.34-PM.png)]