David M. Adlerstein is counsel at New York City law firm Wachtell, Lipton Rosen & Katz, where he focuses on mergers and acquisitions, corporate, securities law and regulatory matters.
In this opinion piece, Adlerstein discusses the current state of smart contracts, providing what he believes is a useful definition for the concept that can help answer key questions about the emerging technology.
I invite the reader to perform a simple experiment: Ask five people with reasonable working familiarity with blockchain to list five key potential benefits of the technology.
I'd wager that at least three will include "smart contracts" on that list. Ask that same group to define what a smart contract is, and you'll get at least three different answers (likely ranging from an agreement for the transfer of an asset on a blockchain, to the simple execution of code on the blockchain, to the koan-like "code is law").
While it is clear that technology (notably blockchain) is opening new horizons for the creation and performance of legal agreements, there is no consensus as to what people mean when smart contracts are discussed. This naturally sows confusion, including as to the legal status of smart contracts.
This article attempts to enhance discussion by positing a working legal definition of smart contracts, and, using that definition as a framework, offering preliminary answers to three fundamental questions of interest to technologists and lawyers alike: Are smart contracts contracts? Are smart contracts smart? And are smart contracts legally recognizable?
The working definition of a "smart contract" is a consensual arrangement between at least two parties for an automated, independent commercial result from the satisfaction or non-satisfaction, determined objectively through code, of a specific factual condition.
A few preliminary observations on this working definition. First, it is prescriptive; I do not contend that this describes general current use of the term "smart contract." Moreover, while blockchain promises to facilitate arrangements of this type, I submit that a smart contract can exist outside of the context of a blockchain (and the foregoing questions remain pertinent as to non-blockchain-based arrangements) and that a smart contract can have as its subject matter something other than the custody or transfer of an asset.
Importantly, a non-consensual or single party arrangement is, per my definition, not a smart contract (consistent with the legal conception of a contract, as discussed infra). An arrangement where either effecting the result of the arrangement or making a determination of whether that result will occur requires human intervention beyond the bounds of the arrangement is also not a smart contract.
Finally, the "independent commercial result" is a necessary element of the definition, for the simple reason that computer code always consists of deterministic "if-then" statements (for example: if I am playing a driving video game and move my control pad to the left, then the car in the game will move to the left – but this is obviously not a smart contract).
Are smart contracts contracts?
Within the American legal system, a contract is "an agreement between two or more parties creating obligations that are enforceable or otherwise recognizable at law." For an agreement to be enforceable or otherwise recognizable at law (as opposed to, say, an unenforceable agreement with me that senior citizens shouldn't invest their life savings in dogecoin), three particular elements must be present:
- An offer (basically, expression of a willingness to enter into a binding agreement subject to the offeree's acceptance of the proposed terms);
- An acceptance of those terms
- A mutual exchange of value – so-called "consideration."
Legally binding agreements identify the parties and subject matter with particularity, and include mutual promises, which may be absolute or subject to conditions. A legally binding agreement may be written, including, as discussed below, in an electronic form, or oral (except in limited circumstances).
And finally, the legal system offers remedies for the breach of a legally binding agreement, such as a requirement to pay damages or, in certain circumstances, a court order mandating performance under penalty of law.
My contention, as per the working definition posited above, is that smart contracts can be, but are not necessarily, legally binding contracts.
In Nick Szabo's classic conception, a humble vending machine transaction is a proto-smart contract, consistent with the working definition: insert a dollar and automatically receive a can of soda. Now in the case of this vending machine transaction, there actually is a legally binding contract.
Stocking the vending machine with beverages and inviting passersby to insert a dollar to purchase one constitutes an offer. Actually inserting that dollar constitutes acceptance. The dollar and soda, respectively, are consideration. And while it would be petty to sue over it, if the machine eats my dollar without giving me a soda, I would have legal recourse.
But a smart contract may represent only a component of, or means of performing a component of, a legally binding contract, rather than an entire contract.
Per the offered working definition, the feature of an adjustable-rate mortgage providing for automatic deductions of mortgage payments due from a bank account would constitute a smart contract: if the reference interest rate changes, the amount of the payment will automatically adjust upwards or downwards.
But this adjustable payment mechanism is only a component of the mortgage agreement, not the agreement as a whole, which would be evidenced separately. For example, this payment arrangement, though it implies the existence of offer, acceptance and consideration, does not itself fully evidence the existence of those components, or other key terms of the mortgage, such as an identification of the mortgaged property, disbursement of the actual mortgage loan or resolution of disputes.
The writing evidencing a legally binding agreement can be electronic (with an electronic "click-through agreement" representing a typical example). In the context of blockchain technology, it is possible to record an agreement – or a cryptographic hash of an agreement – as metadata within a blockchain.
While this may in and of itself offer distinct advantages, particularly in establishing for posterity what the definitive terms of an agreement are, recording a legally binding agreement within a blockchain does not, without more, create a smart contract.
Only to the extent that performance of that agreement is automated through code based on satisfaction or non-satisfaction of an objective precondition is there a smart contract.
Are smart contracts smart?
While contractual performance is today often automated for certain uses (for example, periodic auto-payments), sophisticated commercial contracts are rife with "if/then" provisions, dependent on the state of objectively verifiable facts, often requiring manual administration and susceptible to misapplication or inadvertent non-application.
Thus, smart contracts can be said to be "smart" to the extent they offer the efficiency of automated contractual performance and reduce the risk of human error and prospects of a dispute.
But barring quantum improvements in artificial intelligence, the utility of smart contracts is limited to situations where the satisfaction or non-satisfaction of a particular factual condition is objectively ascertainable through programmatic reference to an extrinsic data source (an "oracle"). In the context of commercial agreements, a computer can be programmed, say, to use an oracle to ascertain whether LIBOR has increased or decreased.
This is not to say that smart contracts are simple; indeed a smart contract could encompass a range of complex results based on multiple inputs. But a computer cannot (today, at least) be programmed to accurately ascertain, for example, whether or not a party to a merger agreement has used reasonable best efforts to obtain a regulatory approval. Viewed through this prism, smart contracts aren't actually smart; they are deterministic.
The promise of blockchain technology and smart contracts is, then, at root not about any kind of native intelligence or computers as lawyers.
Rather, it is about blockchain's potential to offer efficiency gains by greatly expanding the scope of contractual matters for which performance can be automated as the nexus among assets, service capabilities and the blockchain grows (especially with the advent of the Internet of Things).
However, many aspects of sophisticated commercial agreements are not susceptible to automation, including matters requiring subjective human judgment, the rendition of sophisticated or human-intensive services or the resolution of disputes.
Accordingly, while some relatively simple agreements could have essentially all of their performance automated (including by blockchain), for more sophisticated agreements only discrete elements are likely to be automated in the foreseeable future; thus, "smart contractual provisions" might be a more appropriate term.
By analogy, blockchain technology will for the lawyer be like power tools to the carpenter, not like the self-driving car to the chauffeur.
And given that computer code is as susceptible to error as prose and that legally binding agreements must be subject to judicial enforcement, as the use of smart contracts grows, in the context of sophisticated agreements providing for ongoing performance, it will be necessary to maintain safety valves (in the context of blockchain technology, perhaps in the form of private keys) for a court or arbitrator to be able override smart contractual provisions.
In other words, irrevocable automation with no possibility of intervention by a "smart human" would be perilous.
Are smart contracts legally recognizable?
Blockchain technology has increasingly become a point of state legislative focus (notably in Delaware), and there have been recent efforts to specifically authorize smart contracts in some states. Of particular note, in March 2017, Arizona enacted legislation legalizing smart contracts (defined in the law as "an event-driven program, with state, that runs on a distributed, decentralized, shared and replicated ledger and that can take custody over and instruct transfer of assets on that ledger").
Unlike the working definition proposed above, Arizona defines smart contracts with specific reference to blockchain technology, and limits the applicability to the custody and transfer of assets – a narrow definition, but one well suited to many burgeoning use cases of the technology.
And new blockchain-related legislation in Nevada was initially proposed including a definition of smart contracts as an "electronic record … which is verified by the use of a blockchain" (but the legislation as enacted dropped the definition).
Even without welcome state legislative initiatives, there are existing common law and statutory bases for a court to enforce a legally binding agreement – or component of a legally binding agreement – existing in electronic or code form, as long as that form is reducible to writing:
- Under longstanding common law contractual principles, extrinsic writings may be specifically incorporated into a legally binding agreement. Accordingly, insofar as smart contracts represent components of legally binding agreements, they are binding if specifically incorporated by reference into a written agreement. For example, in the case of a written loan agreement providing for a pledge of collateral where the pledge is reflected in a blockchain and would be automatically released on repayment of the loan, an agreement referring to the contemplated automatic release and including an excerpt from the applicable executable code in an annex should be enforceable by a court.
- Under the federal Electronic Signatures in Global and National Commerce Act (2000), a contract, signature or record is not considered unenforceable merely on the basis of being in an electronic format (but the record must remain capable of being reproduced for later reference). While the author is aware of no case specifically considering the question, there is no reason to conclude that an electronic contract in code and not prose would be unenforceable, so long as the parties, subject matter and terms are clearly articulated in a manner essentially translatable to English, like a foreign language, and there is evidence of mutual consent in deploying the code in question, with each party giving consideration.
- Under the Uniform Electronic Transactions Act (adopted by 47 states), transactions may be conducted by electronic means, such that the act gives legal recognition to electronic signatures, records and contracts.
- Under Article 9-105 of the Uniform Commercial Code (the law adopted by all 50 states governing secured transactions) a secured party has control of “electronic chattel paper” if a system for evidencing the transfer of interests reliably establishes the secured party as assignee (among other things, a single authoritative copy of the record must exist which is unique, identifiable and generally unalterable).
Terminology aside, new technology (and blockchain in particular) promises to have a far reaching impact on how legal agreements are evidenced and performed. But existing concepts of what constitutes a legally binding agreement will endure, as will the human element in negotiating and administering sophisticated commercial agreements.
And while thoughtful legislative initiatives are welcome, existing legal frameworks likely already provide a reasonably robust basis for enforcement.
Disclaimer: The views expressed are the author’s and do not necessarily represent the views of Wachtell, Lipton Rosen & Katz.