These are certainly exciting times for the legal profession. Our public square is noisy with talk about how technology is going to revolutionize law and transform the profession as we know it. Much of this talk is, however, long on optimism and short on specifics. Last week’s ReInvent Law conference in New York City was a great example of this, having produced many exciting proposals as well as a few well-placed criticisms about our legal future, but not a convincing road map for the way forward.
I’ve read more articles and blog posts about the future of law and technology than I care to admit. None of them has satisfactorily answered the question posed to me by lawyer Nick Trenticosta when I told him that I would be teaching my law students to program computers: “Computers! What do computers have to do with it?” In other words, why should law students learn to code? Don’t they already have enough on their plates just learning the law? Can’t they just hire someone to make the software for them?
I’m not sure I gave Nick a satisfactory answer to his question at the time, but I now believe that at least part of the answer is that computing will be more far more pervasive, far more essential to the law in the future than we may now suspect. Computer-based technology may not be just a tool to be used by lawyers, but may come to be an essential part of what law is, how it is formulated and dispensed. The most obvious way this may come to pass is in the application of algorithms to the law. Algorithms have become so common in other fields (banking, policing, public health, to name a few), I would not be surprised if within a decade or so, lawyers will spend a good part of their days struggling with them.
An algorithm, expressed in its simplest form, is simply an equation that leads to a result. Much of what we do in everyday life is already observed and regulated by algorithms. If you think about it, law is supposed to be a kind of algorithm. If certain conditions are met, then certain results should follow. In the push to standardize and optimize law, it’s not hard to imagine that laws as we know them might be replaced by complex algorithms, ever more detailed and specific to certain case scenarios. Envision a time when your state’s legislature appoints committees, amply supported by technology-savvy clerks, who design algorithms for the law. The algorithms are tested, debated and approved and then uploaded to the State’s central repository for use by lawyers, judges, and the public at large. Much of the advocacy that is done by lawyers then focuses around applying and modifying these algorithms.
How might such a system work? To begin to answer this, let’s indulge in a bit of science fiction. Here are two not so fanciful vignettes about how the practice of law might look in the coming decades.
The Custody Case
Suzanne is a family law attorney. She’s just received a Preliminary Judgment from the judge who is handling her client Linda’s case. Pulling the report up on her screen, she sees the judge has applied the state-approved algorithm for determining custody arrangements in cases where both parents desire custody of the child. The news is not good. The algorithm has determined that it would be best for Linda’s ex-husband to be the primary domiciliary parent. Using the (hopefully open source) software that Linda has for parsing these determinations, she sees that the algorithm has assigned significant value to the fact that Linda is only occasionally employed and has poor prospects for full-time employment. The Legislature has reviewed several studies which show that parents with long-term full-time employment tend to have children with better outcomes in later life and included this in its “best interests of the child” algorithm.
In response, Suzanne opens up her text editor and suggests that the following facts should be added to the computations: 1) her client is enrolled in a part-time nursing program and intends to graduate within two years and 2) her client’s child by another father has just graduated with honors from high school. Both of these are important facts which were not considered in the preliminary judgement, Suzanne argues, and the algorithm should be run again to include them.
Suzanne then lodges this objection (which in Github terms we might call a pull request) which is uploaded to the court’s repository for further consideration.
The Appellate Judge
Judge Alexander has just settled down to another lovely day in the Court of Appeals. It’s a wonderful job, lazily supervising the algorithms which crunch all the facts the opposing lawyers have inputed and reviewing the suggested rulings generated by the algorithm. He can usually do all of this from the coffee shop down the street from his house, but he’s come in to chambers today for lunch with the chief judge. As he sits down, his law clerk comes in to tell him about a case where the computer has screwed up and something needs to be done.
In the case, a juvenile defendant convicted of spray painting gang symbols on the side of a local store has been given the extraordinary sentence of 10,000 hours of graffiti removal as part of his community service. His lawyer has lodged an objection urging application of the Cruel and Unusual Punishment algorithm. In support of his objections, the lawyer has inputted the fact which emerged at the restitution hearing that it only cost the store owner $300 dollars to remove the graffiti. Assuming his client were to be paid a minimum wage of $7.25, his client is being made to do $72,500 worth of work for a $300 damage. This sentence, says the lawyer, calls for an amendment to the algorithm, as the sentence is clearly disproportionate to the crime.
The judge knows that these Cruel and Unusual Punishment claims are usually pretty easy. The algorithm looks at the crime and the predetermined sentencing ranges, crunches in any social and environmental factors about the defendant which were adduced at sentencing and produces a finely tailored sentence. The appellate court hardly ever overturns these sentences. In this case, however, Alexander agrees that this has been an abuse of the trial judge’s discretion and that the algorithm that recommended affirming the sentence is wrong. He sends a note to the other judges on the panel (in git terms, he makes a commit to the case repo), suggesting that the Cruel and Unusual Punishment algorithm be disallowed in this case. Having done that, and with more time on hands before his lunch, he then drafts a proposed amendment to the algorithm which suggests that, in restitution cases, the economic value of the defendant’s community service work should be calculated against the amount of damage he created. He lodges a pull request with the Legislative Criminal Justice Committee’s repo which will be duly considered in the next legislative session.
This computer-driven vision of the future, while far removed from the dusty books and shoot-from-the-hip judgments of law’s past, is hardly Orwellian. Technology is used to augment, not replace, human intelligence, just as Douglas Englebart famously proposed in 1968. Lawyers are still engaged in their true calling – advocacy. But many of the tasks which were left to analog human judgment in the past have been formalized, streamlined, and made more predictable.
Certain effects would inevitably follow from an algorithm-based legal system. Foremost of these is that lawyers whose work involves mere case processing will fall out of demand. Cases which involve the routine collection of non-controversial inputs to achieve a largely preordained goal will be handled by computers. Lawyers become superfluous in a simple divorce, for example, where the parties simply make a couple of sworn inputs on a court’s CMS which are then analyzed by the algorithm to produce a judgment. There will be a great premium, however, on good advocates who understand how the new technological system works and can use it to the benefit of their clients.
This leads us to the second important effect, namely, that in such a future it truly will be important for law students to learn to code. In this future, computers are not just auxiliary tools to help in the practice of law. Instead, the law becomes a natural result of the peculiar language and processes of computers. Law students, as future practitioners and shapers of the law, will need to have a full understanding of these processes in order to ensure that law reaches its true goal, justice. As Prof. Brian Kernighan of Princeton University, one of the creators of Unix, once remarked,
The people making policy decisions in our country ought to understand computing because it is a pervasive part of our lives. And until you battle with the machine, you don’t really understand how precisely you must talk to a computer to make it do what you want. And you don’t understand all the things that can go wrong.
In a legal algorithmic world, it will also be as important as ever that computing tools and data formats be free (as in beer) and open source. If the software which runs the system is slow to evolve because the source code is unavailable, or too expensive for the average citizen to acquire, then many important voices will be left out of the justice system entirely. In fact, access to justice issues are one of the most challenging facets of emerging legal technology and deserve constant scrutiny. The many legal informatics and design centers which are sprouting up across the country should focus their energies on creating these tools now, before proprietary solutions capture the market.
Whenever I give talks to lawyers about technology, I tell them that, no matter how technology advances in the legal profession, the role of lawyer as advocate is never going away. Advocacy is about imagining different realities and presenting them to other humans in a convincing way, sometimes in the face of seemingly incontrovertible evidence. It is a fundamentally human activity that computers will be hard-pressed to do well any time soon. With this reassurance, we lawyers should embrace the possibilities computer-based technology in the law and begin planning our future in earnest – with specifics.