Blockchain technology and “smart contracts” are often accused of hype. Yet multiple industries continue to search for the opportunities to unleash their inherent innovative and disruptive potential. Given the fundamental mechanics of insurance and the pervasive issues with contract management, one could argue that the industry truly needs the potential of the smart contract.

What is this Blockchain that Underpins the Concept of Smart Contracts?

Blockchain is a database, recording transactions, forming a “block” across a peer-to-peer network. Participants install the application locally and all the “nodes” hold a copy of the database, i.e. there is no central entity that holds control. The database is structured as a ledger of transactions into the blockchain, which is replicated in full by each participant’s computer, and therefore consists of blocks that hold “time-stamped” batches of valid entries. Each block includes the “hash” of the prior block, linking the blocks together. The linked blocks form a chain, with each additional block reinforcing those before it. Transactions are passed from user to user, or node to node, on a best-effort basis. If any data in any block in the chain is later altered, the hash of that block will no longer correspond to the later blocks’ hash of that “tampered” block, so the change will be clear to participants. The result is an indelible record operated on a distributed basis.

So What is a Smart Contract?

Smart contracts are computer protocols, embedding the contractual obligations of the contract into source code, which is compiled into executable computer code that can then be automatically executed per the terms and conditions of that contract. The distributed ledger architecture of blockchain is used by adding the code that makes up the smart contract into part of an entry to a blockchain application. One can define a smart contract as “code deployed to a distributed and replicated ledger, with an ability to control its own state and assets, with an ability to respond to the input of external information”. The trust aspect that is fundamentally built into the blockchain means that smart contracts can be entered into between two parties unknown to each other. The participants of the blockchain (which could be party to the smart contract, or have an interest in it) have access to the block within which it is contained. The relevant block can be public (with an ability for all to view) or accessed on a “permissioned” basis.

Smart contracts can be fully automated and self-enforcing. Once the computer code representing the contractual obligations is set, the contract will run its course and the terms will be executed impartially by the computer on the basis of the code and the external information that the contractual obligations are dependent on. This automation, combined with the lack of traditional trust-building costs, significantly decreases transaction costs, making such exchanges much more profitable.

However, drafting a contract that takes into account all possible contingencies and states all their desired outcomes is difficult, especially in the real world of commercial contracts – hence the need for courts and judges to help in cases where circumstances unforeseen by the parties arise. In such cases, it is important that the smart contract’s coded outcomes should not occur automatically. There is a drive to use artificial intelligence for the drafting, managing, and enforcement of smart contracts – however, it has to date been a challenge to embed the concepts required by key industries such as fairness and good faith. This means an interim state where smart contracts are open-ended and rely on input from non-automated actors.

The Issues with Semantic Information and Smart Contracts

Information can be divided into two types: syntactic and semantic. The former are the rules about the relationship between symbols, whereas the latter relates to the meaning attributed to such symbols (i.e. “intent”). Syntactic information can be parsed and measured (as famously shown by Shannon and Weaver in their seminal work A Mathematical Theory of Communication), and is open to mathematical proof.

Semantic information, however, is what we as humans attribute to the meaning of a symbol. Normalizing semantic information, such as natural language, making it processable, and representing its meaning are difficult tasks. Legal notions of “good faith” commonly used in contracts to deal with circumstances unforeseen by the contractual parties, for instance, are particularly complex, especially in the context of insurance contracts.

Given the smart contract automatic execution of the outcome upon the meeting of pre-set conditions, and the immutability through blockchain implementation, there is significant danger in an incorrect determination of a condition that is semantic in nature. Various solutions have been mooted, such as the concept of an “escape hatch”: a pre-programmed way of changing the terms of a smart contract.

Deterministic versus Non-Deterministic

There are two types of smart contract that need to be considered: deterministic and non-deterministic. The primary differentiator is whether the network would have sufficient information to determine an outcome, or whether outside information is required. The ideal is to decentralize such a service in order to reduce the potential issues of trust and improve transparency.

Where non-deterministic, there isn’t sufficient information on the distributed network itself to make the required decision – so an external source/party is required – an “oracle”. The parties to the smart contract will need to agree a trusted source as the oracle, to submit the outcome that is a key part of the condition for a payout. Accordingly, the reliability and trustworthiness of the source is key. Some have approached this issue by implementing carefully thought-out processes to reduce the potential for undesired behaviours, e.g. federating oracles or an arbitration process. Principal component analysis can be used to reduce the risk of fraud/manipulation to break out the sources of the data and make the computation harder to “game”.

Insurance Contracts: Opportunities for Smart Contracts

UK insurance contract law has its foundations in common law and was codified into the Marine Insurance Act 1906, designed to protect an insurance market from exploitation by customers who knew their business better than the insurers, giving the insurers draconian powers to avoid insurance claims at any sign of wrongdoing by the insured, imposing on them a higher standard of utmost good faith (uberrimaefidei).

In Prudential Insurance Co v IRC, Channell J held that the essential requirements of a contract of insurance were:

• The payment of a premium in return for the other party undertaking to pay a sum of money upon the happening of a specified event;
• The specified event being one that is adverse to the interest of the policy-holder, over which the policy-holder has no control (i.e. there is an element of fortuity); and
• The presence of an insurable interest (i.e. the insured has a legal or equitable interest in the subject matter of the insurance cover and would either be prejudiced by its loss or benefit from its safety).

The UK Insurance Act 2015, which came into force in August 2016, has created a different balance, representing a significant change to insurance contract law as it has stood for over 100 years. For business contracts, it forces cooperation between insured and insurer at the pre-contract stage. It does this by introducing a new obligation on an insured to make a fair presentation of a risk to the insurer. This greatly assists the use of smart contracts, encouraging a transparent discussion of the code conditions at its heart, based on this fair presentation of the underlying facts known.

With the explosion of data required in the assessment of a payout of an insurance policy being available from social media and the IoT (the Internet of Things), there is also the ability to objectively assess the occurrence of the event at the heart of the insurance policy. Provided the underlying data can be trusted and protected from fraud and manipulation, a major issue for insurance – fraudulent claims – can be significantly mitigated by the smart contract. Moreover, given the ease of the claims process for the insured due to the automatic payout, it is likely that customer satisfaction can also be improved.

Another challenge for the insurance industry is the control of costs relating to claims, for example, trying to limit the medical professional panel that must be used in the case of injury, or the garage used for repairs in respect of a car accident and the repairs to the vehicle. In the latter case, the condition in a smart contract could be met through a connected car, whose data clearly validates the occurrence of an accident. The smart contract code could limit any payout to specific designated third parties who are also on the relevant blockchain underpinning the smart contract.

Concern as to whether the insurance company will view an event as one caught within the coverage of the policy is also an area that the smart contract can assist with. A quick glance at a critical illness insurance policy shows a long list of exclusions that raise the concern of whether the policy provider may try unfairly (in the eyes of the policy-holder) to exclude an event. Frequently, the policy-holder is at an informational disadvantage to the insurer in terms of understanding the true coverage of the policy. In many ways, a smart contract would just codify the conditions and list of exclusions; however, the fact that the execution of the payout would be automatic, based on the data logic that has been coded (and agreed between the parties), means that it at least forces the removal of subjectivity that the insurer might use to its advantage (albeit with reputational risk).

The Next Steps to the Smart Contract Transformation of Insurance Policies

Bearing in mind the issues we have discussed, and given the breadth of the insurance sectors, it is clear that smart contracts are able to transform some areas of insurance more readily and sooner than others, typically where the conditions of the policy can be determined objectively and verifiably through data that has a low risk of being fraudulent.

The reliance on oracles cannot be underestimated, and, going forward, social networks and IoT will be the cheapest and most commonly used data feeds for various different insurance applications, making for the cheapest, most reliable, oracles for all types of future claims validation efforts. However, there are rightly significant concerns over the use of this data in relation to rights of data privacy and protection. How the industry manages this within the context of the smart contract will be a key determinant of its adoption.

But is it Still Insurance?

One of the essential requirements of an insurance contract is showing loss to the insured. Without it being present, some would argue that it is in fact a derivative contract, rather than one of insurance (and the regulatory requirements around the treatment therefore differ significantly). Accordingly, unless one can get comfortable on the objective determination of loss to the insured, there is a more fundamental question thrown at the heart of the insurance smart contract. That said, if it meets the desired commercial objectives of the contractual parties, should we really worry about the form (rather than the substance), assuming the regulation can keep pace with the transformative change potential of the smart contract?

Smart contracts certainly are on the way to helping the management of uncertainty better than the current traditional contractual forms and regulation will simply be forced to keep up as the potential of insurance smart contracts is realized.