A Programming Language for Future Interests.

• Author: Basu, Shrutarshi

Introduction I Programming Languages and Law A Contract B Tax C Legal Drafting D Visualization II An Informal Overview A Previous Work B Orlando and Littleton C An Example III The Formal Details A Title Trees B Semantics C Conveyances D Translation E Conclusion IV Lessons for Property Law A Design Principles 1 Orlando 2 Littleton B Insights into Property Doctrine 1 Defaults 2 Syntactic Ambiguity 3 "Theorems" of Property Law C Insights into Property Theory 1 The Numerus Clausus 2 Recursivity 3 Modularity Conclusion Appendix: Orlando Reference Introduction

The formulas that govern future interests are similar to those of chemistry. They seem to be more of the law of nature than law of men except for one crucial difference: The rules of future interests occasionally make no sense. (1) Though of feudal origin, it is not a relic of barbarism, or a part of the rubbish of the dark ages. It is part of a system; an artificial one, it is true, but still a system, and a complete one. (2) The logician must be rather like a lawyer ... in the sense that he is there to give the metaphysician ... the tense-logic that he wants, provided that it be consistent. He must tell his client what the consequences of a given choice will be ... and what alternatives are open to him.... (3) Every law student and every law professor has a different reaction on reaching the unit on estates in land and future interests in Property. For some, it is the worst part of the course. They find the system of reversions, possibilities of reverter, and remainders vested subject to complete divestment to be an alien language: dull, desiccated, and divorced from the practical realities of the rest of law. (4) For others, it is the best part of the course. Here, there are no counter-arguments and indeterminate multi-factor tests, only rigorous deduction and clear right answers. (5)

These two groups, polar opposites though they may be in their approach to law school, share an intuition: there is something logical and computational about estates and future interests. Whether they want the computer to serve as a junior associate that calculates the consequences of conveyances so they don't have to, or as a sparring partner that plays along with them, they share the sense that there is something about this particular system of legal doctrines that makes them particularly suited for automated algorithmic analysis. A life estate and a remainder fit together like a lock and a key, with the mathematical certainty that establishes 2 + 2 = 4. Couldn't someone program a computer do this?

We did.

Our system, called Littleton, (6) can interpret stylized conveyances like O conveys to A for life, then if B is married to B, but if B divorces to C and correctly report that B holds a contingent remainder in fee simple subject to executory limitation. It knows that O holds an implied reversion; that if B marries while A is alive then B's remainder is upgraded from contingent to vested subject to complete divestment; and that if A conveys their interest to D for life, then D's interest will be limited to the shorter of A's and D's lifetimes. It can even apply the Rule Against Perpetuities to strike interests that could vest too remotely.

We designed Littleton to be useful to teachers trying to explain the system of future interests and to students trying to learn it. We have put a web version online at https://conveyanc.es. Just type a conveyance in the box, click on "Interpret," and Littleton will display an easy-to-understand diagram of the resulting interests. It comes with documentation and a tutorial of demonstration conveyances, and has been validated against examples drawn from one of the leading student guides, Linda Edwards's Estates in Land and Future Interests. (7) We have also placed Littleton's source code online, and released it under the permissive MIT license, allowing anyone to use and improve it however they want. (8)

But that's not even the interesting part.

Rather than using an existing programming language to write a program to model future interests, we treated the formalized, ritualized language of first-year Property conveyances as a programming language itself. Each term in this language, which we call Orlando, (9) has a precisely specified syntax and semantics. The expression O conveys to A in Orlando is like x = y * 4 in a traditional programming language like Python, Java, or C: a command that causes the computer interpreting it to update its state in a predictable, objectively determined way.

This makes Orlando into a a domain-specific language (or "DSL"). (10) Just like JavaScript is useful for writing interactive web pages, Ink (11) and Inform (12) and Twine (13) for creating text adventure games, Solidity for smart contracts, (14) or Flash for animations, (15) Orlando is a language for expressing property conveyances.

Drawing on the computer science discipline of programming language theory, we treat Orlando like any other DSL. (16) Littleton's processing is divided into stages:

* First, Littleton parses a conveyance written in Orlando, recognizing the individual clauses and their relationship. The language O conveys to A for life, then to B, for example, consists of two separate grants, linked by then. The first has a quantum (for life) attached to it; the second does not.

* Next, Littleton creates a data structure (which we call a title tree) that keeps track of the current interests and their relationships. The title tree corresponding to

O conveys to A for life, then to B until C marries.

is shown in Figure 2.

* Littleton then applies substantive rules of property law to update the title tree as further events occur. That is, while the syntax of Orlando is given by the stylized language used in conveyances, Orlando's semantics are those of property law.

* Littleton analyzes the title tree in accordance with various rules used by lawyers and law students, so that the various interests can be properly named. For example, it classifies remainders as contingent or vested based on whether a condition precedent must be satisfied before that node in the title tree can be reached.

* Finally, Littleton displays the current state of the title by rendering the title tree in a graphical format that hides many of the internal details and emphasizes the viable interests and the conditions on those interests. The resulting visualization is designed to be readily comprehensible to lawyers and law students who need not be aware of the sophisticated processing taking place under the hood. Figure 3 shows an example of Littleton's output.

Treating conveyances as a programming language yields insights into property doctrine, into property theory, and into legal theory more broadly. Doctrinally, Orlando brings the entire system of future interests into clearer focus by capturing the linguistic structure of property grants in a succinct and intuitive way. A confusing mess of doctrinal minutiae resolves itself into an orderly collection of well-specified rules. Facts about conveyances that previously became apparent only after detailed study are now immediately obvious--for example, that a grantor can recursively stack up an indefinite number of successive life estates. It is even possible to prove "theorems" of property law, such as that a fee simple is forever.

Theoretically, the fact that this fragment of property law can be formalized in this way is striking: other areas, like trademark law or international human rights law, almost certainly cannot. Orlando's simple but generative structure provides a new kind of support for a line of scholarship, associated with Thomas Merrill and Henry Smith and with the New Private Law movement, that emphasizes the modular and standardized elements in property's conceptual structure. For example, Orlando's design embodies the numerus clausus principle: that property interests only come in a finite set of forms.

Finally, Orlando is a proof by example that legal scholars can learn from programming-language theory. Law and programming languages can be to law and computers as law and linguistics is to law and language: a subfield that draws on the insight of another discipline to identify and systematize recurring structures of pervasive importance to law. The linguistic parallel between the natural languages of law and the artificial languages of software offers a fresh way to reflect on how law, lawyers, and legal texts work. In property and beyond, defining a programming language to model a body of law should be part of legal scholarship's methodological toolkit.

This Article provides a detailed exposition of a core subset of Orlando and Littleton, and a discussion of why they matter to legal scholars. (17) Part II introduces Orlando informally; Part III explains the formal details underneath the surface. Part IV discusses the design philosophy of Orlando and Littleton to show how they hold lessons for property law and property theory. And Part I surveys the scattered scholarship applying programming languages to law to argue that other scholars should consider creating their own legal DSLs.

1. Programming Languages and Law

   Computerizing legal reasoning is by no means new. There is a long-standing research program on the use of artificial intelligence (AI) systems for other areas of law. It has proceeded along two tracks, corresponding to the division within AI between systems using formal logical reasoning, sometimes called "symbolic" AI or "good old fashioned AI" (or GOFAI), and systems using statistical methods, sometimes called "subsymbolic AI" or, more recently, "machine learning." (18) Legal scholars draw on both tracks. (19) Orlando is squarely in the former tradition, so we focus on it here.

   The use of AI systems to automate logical legal reasoning goes back decades. (20) Many scholars, legal-automation companies, and even teams of students have built "expert systems" that can walk the user through...