Law and Autonomous Systems Series: What is a Smart Contract?
The term ‘smart contract’ refers to computer code that is designed automatically to execute contractual duties upon the occurrence of a trigger event. Referring to a 1997 paper by Szabo, the simple example of a vending machine is often cited to explain the concept: upon insertion of a specific type of coin, the computer programme instructs the machine to release the good. Yet, there are different ways to understand the term ‘smart contract’.
The narrow understanding of smart contracts
The original understanding of smart contracts, to which also Szabo’s example relates, is the narrowest. A smart contract ‘excises human discretion from contract execution’ (Raskin, 2016). Unlike the performance of contracts generally, performance of a smart contract cannot be stopped, neither voluntarily by the parties (i.e. it can neither be breached nor amended), nor by a central entity, nor by a court or supervisory authority. Considering the example of the vending machine again, it actually does not seem to be a good example because it retains human discretion: the machine is still technically under the control of its owner.
This is where blockchain comes into play, even though smart contracts were originally conceived independently of the blockchain idea. The (older) concept of smart contracts will achieve its full potential only if combined with the (newer) invention of blockchain networks. This is because the certainty of execution is not absolute as long as human discretion can interfere with the process: the vending machine is technically still under the control of its owner. The record of a blockchain network on which a smart contract is stored is supposed to be absolutely immutable and its execution automatic – at least as long as the relevant blockchain is of the original, bitcoin-style type, as opposed to the modified use cases recently adopted for finance and other services, see here. Autonomy of execution is a direct consequence of the fact that blockchain networks operate without any central or trusted entity to balance the parties’ interests.
In other words, it is only in blockchain networks that there is truly no ex post review of contractual duties after contract formation. The only way to influence the execution of smart contracts is by programming them in such a way that they seek external input on the further execution (from a non-smart, human-controlled IT process into which they are embedded, or from an authority or court) at the occurrence of certain, predefined events. This is the point in time at which further execution can be aborted or otherwise influenced, however only on the basis of pre-programmed options.
The absolute certainty of performance makes contracting more efficient as the counterparty and settlement risks typically inherent in contracts are considerably reduced, if not eliminated. A simple example is the securities collateral kept in a blockchain network: if the debtor has not paid by a certain date, the smart contract autonomously transfers the securities to the creditor. Furthermore, the precision of the programming language is much greater than that of written human language; in particular, warranties and conditions can be formulated with much greater accuracy, and contracts can be treated and processed in data formats. Hence, it appears that smart contracts make transacting considerably less expensive owing to certainty of execution and the near-zero risk of litigation in court.
In financial markets, smart contracts could be used for a variety of functions. For instance, a bond held in a blockchain network might have a smart contract attached to it that automatically executes interest payments on the payment date, and the amount to be paid is determined on the basis of data retrieved from a predefined, reliable Internet source. A second example relates to the derivatives market. Parties might enter derivative contracts electronically; the relevant building blocks of that short programme would automatically be taken and assembled from an electronic contract library set up to this effect. The smart contract could be designed so as to automatically cater for due payments to be executed and to adjust collateral levels between the parties. Also, upon termination of the contract, the programme could autonomously calculate the termination amount due to be paid. Again, these payment amounts would be calculated based on reference data sourced from a predefined, reliable data provider.
Legal and regulatory issues
The main challenges of smart contracts (in the narrow sense) are rooted in the truly unstoppable execution of what has been previously agreed and programmed into the code.
First, such execution might produce effects akin to systemic dangers provoked by the phenomena of ‘herding’ or ‘flash crash’. Here, the autonomous and unstoppable execution of transactions and smart contracts in blockchain financial networks may aggravate a market movement that may lead to systemic instability and which is rooted in the uncoordinated but uniform behaviour of a significant number of market participants. Removing the human element entirely from decision-making eliminates the last vestiges of elasticity in the behaviour of parties, which does to some degree exist in wholesale financial markets due to the generally relational character of contracting prevailing in this environment. Eliminating elasticity may be advantageous from a market efficiency point of view in good times, but may also amplify market distortions in times of crisis. Smart contracts, in particular those hosted on blockchains, take the ‘immediateness’ of market reactions to an extreme and may combine it with a high degree of interdependency of the various processes involved.
Secondly, on a more general note, the use of smart contracts leads to a situation in which court decisions do not exert the same authority as in the traditional transactional context. Should a party claim that a smart contract was unenforceable, the court will be unable to order a rectification of the result, as the outcome cannot subsequently be changed without destroying the logic of smart contracts (i.e. their immutability). This situation leaves the relevant party with the only option of claiming damages from the other party, in kind (i.e. the court may order the initiation of a new, reverse transaction) or in money. However, as is generally the case, claiming damages will often frustrate the party whose interests were overridden, in particular if the other party has become insolvent in the meantime. Yet, as opposed to the ‘non-smart’ environment, where performance can still be corrected and transactions operationally reversed in certain cases, the smart contract environment offers damages as the only remedy, invariably subjecting the transferor to the risk of insolvency of the transferee.
Also, in relation to “regular” contracts, things may become more complicated for the parties: where one party claims that the contractual duties should be adapted in response to new circumstances not previously considered by the parties, there is no way of changing the contract, even in cases where both parties agree to the change. In more general terms, smart contracts significantly limit any kind of ex post review and/or adaptation of the contractual stipulations.
Combining smart contracts
Smart contracts can theoretically be combined and thus interact with one another in a decentralised and distributed structure, operating autonomously (i.e. without human intervention), once deployed by their programmers on the basis of the rules and mechanisms programmed into them. Such ‘decentralised autonomous organisations’ (DAOs) could even enter into new smart contracts with other market actors, creating a complex, evolving ecosystem of interacting agents linked by pre-determined, hard-wired and self-enforcing rules. They are not owned or controlled by any single person or corporation; yet they can interact with the market.
The most important DAO so far was created on the basis of smart contracts recorded and processed on the Ethereum network: ‘A humanless venture capital firm that would allow the investors to make all the decisions through smart contracts. There would be no leaders, no authorities. Only rules coded by humans, and executed by computer protocols’ (Wong and Karr). It raised a spectacular USD 150m of which 50m were subsequently diverted by a malicious node to a private Internet address, leading to the project being abandoned.
Still, despite this failure, similar projects may emerge in the future. By contrast, it is not yet clear whether and to what extent economic actors will develop an interest in such entirely autonomous, self-referential actors since, as for-profit organisations, they ultimately need to keep legal and economic ties with the device and exercise some control over it. In any case, the somewhat extreme concept of totally autonomous self-executing software shows that smart contracts stored on a blockchain network can operate in varying degrees of autonomy from humans and on a smaller or larger scale, providing input to one another in the form of reference data and triggering events, potentially across different blockchain networks. Obviously, the more intertwined smart contracts become, and the lower the degree of control by humans, the more difficult it will be to govern this phenomenon.
Automation, speed and AI
The broader understanding of what constitutes a smart contract typically includes elements of automation, speed, autonomous decision-making or even artificial intelligence, alone or in combination with the feature of unstoppable execution described above. One may agree or disagree with the various usages of the term ‘smart contract’ – however, the label itself is not so important. Where legal issues related to smart contracts are discussed, the clear identification of the functional results of the relevant mechanisms is crucial. In simple words: before embarking on any analysis of the legal framework underpinning the relevant phenomenon, there needs to be certainty regarding the precise effect of having a specific task performed by a machine. Is the process designed to –
- Perform repetitious actions a human could perform following a linear algorithm, in particular with a view to decreasing cost or increasing speed (robotic process automation);
- Process unstructured data and perform actions following complex algorithms, in particular with a view to outperform human decision making (intelligent process automation); or,
- Learn from previous decisions and their effects (artificial intelligence)?
For example, the automatic production of contractual arrangements, say derivative contracts, may include elements of the first, and, depending on the degree of the machine’s influence on the material terms of the contract, also the second category. While these processes are extremely complicated to design, they pose relatively insignificant problems in terms of the law. There is a human or a corporation behind the machine, controlling the outcome. Unwanted or illegal results of the process can be undone. There is a clear addressee for claims for damages, and the competent regulator has no difficulty enforcing the applicable rules. However, legal responsibility for the process needs to be clearly assigned. Where outperforming others in terms of speed and quality is at the centre of the process, as is the case in algorithmic trading, problems lie predominantly in the regulatory sphere, more precisely in the area of systemic stability. All other legal issues that may arise are easily attributed to the owner of the relevant machine.
The future?
Smart contracts offer unstoppable execution. This may bring down immediate transaction costs. However, the concept may come with costs of its own for the relevant parties, and even for society as a whole, depending on the circumstances. Therefore, I submit that we will see the emergence of ‘semi-smart’ contracts, offering automatic execution, however with an ‘emergency exit’ which can be used subject to certain pre-agreed and programmed conditions.
Philipp Paech is an Assistant Professor of Law at the London School of Economics.
Share
YOU MAY ALSO BE INTERESTED IN