Adjustable risk contracts
Context
The concept of risk-adjustment, used in healthcare and insurance to deal with problems of adverse selection and moral hazard, can be applied to all kinds of services. Services, as contracts, are joint ventures of risks between customers and service providers. But now and then they must change to factor in new realities, including those imposed by climate change; either customers, or providers, may struggle to keep their promises of demand, or supply. We can think of every service having a balance sheet of commitments, with promises of demand and supply on the two sides. The benefits, risks, and costs of keeping a promise, fill the rows and columns. A 'debit' on one side is reflected by a 'credit' on the other. This balance of assurances is fundamental to the mutualisation of risks that makes services work.
There are 'good risks' and 'bad risks'. Good risks result in benefits and costs that are shareable between customers and providers. If good risks fall below a threshold, the very concept of the service will shrink and collapse. Bad risks create unnecessary costs that parties impose on each other through terms and conditions. If bad risks rise above a threshold, the contract will fail or fall short of expectations. Risk adjustment applied to services, can thus be like cholesterol management, helping improve the ratio between good risks and bad risks, and by extension, between necessary and unnecessary costs. That will create room for the extraordinary costs of making services socially responsible and environmentally friendly. Service providers will face less pressure on profit margins. Or, in the case of governments and nonprofits, budget deficits. Customers are less likely to face the Hobson's choice of paying more or accepting less. Risk adjustment can thus reduce the economic stress on both sides.
The problem
But all that is easier said than done, because the 'contracts' behind most services do not lend themselves to that kind of dynamic risk adjustment. Why? Because of the tail wagging the dog. The densely worded text of terms and conditions (the tail) limits and restrains the ways in which demand and supply come together within windows of opportunity (the dog). We need to convert services into adjustable risk contracts (ARC) in which the promises of demand and supply are packetised into several, discrete and finite elements that we can individually move and modify. Then we have a better chance in a larger solution space, of finding the most optimal set of several, small adjustments that will reduce unnecessary costs and reduce pressure on both sides. Think of it as actuarial acupuncture. Risk-adjusted contracts compensate providers for the true costs of delivering a service, relying on others to underwrite the difference. Whereas ARC also account for the true costs of using the service and therefore promote an adaptive, smart, and self-compensating dynamic.
The idea of adjustable risk contracts came out of a research project funded by the Dutch Ministry of Defense. The government was looking to develop new ideas for strengthening its relationships with suppliers while being good stewards of taxpayer money. The project was driven by the belief that smart and flexible contracts, based on mutual understanding, cooperation, and trust, can better adapt to changing geopolitical conditions and economic realities. Through special purpose adjustable risk contracts (SPARC), organisations recognise, relate and rely on each other – recognise in each other the potential for demand or supply, relate that potential to a specific purpose, and rely on each other to show up within windows of opportunity.
You can think of a SPARC to be a cross between a special purpose vehicle and a smart contract, written in a language and format that make it easier for humans and machines to collaborate. The rich text of a SPARC is editable like software code. The authoring and editing, and the associated reinforced learning, are directed toward an optimality: Customers get higher value at a lower price while providers enjoy a higher profit but without cutting costs in ways that increase the risks of operational failures, or negative impacts on ecosystems and environments.
What follows is the speculative design of a system for creating, updating, and modifying SPARC, featuring SOFI and RUMI. Think of it as Spotify for Analysts; Casetext for Consultants.
SOFI
Sofi is an ‘artificially intelligent system' that predicts problems in demand and supply, analyses the implications, and proposes changes to contracts and agreements. The changes are acceptable to customers and suppliers because they put everyone in better positions than before, with promises of demand and supply risk-adjusted to new realities. Mutual attraction is higher than before; threats of competitive and regulatory action are lower. Sofi notarises the new version and prepares advisory notes for implementing the changes. But how does Sofi know when to act and what to do? As a trusted system, Sofi is aware of the goals, priorities and concerns of each enterprise. It has privileged access to documents and discussions, allowing it to monitor developments. Its neural networks are like Wall Street analysts, keeping an eye on portfolios of promises and looking for abnormalities and patterns. So, when all indications are that there is going to be a problem, Sofi opens a case, notifies the principals and, following standing instructions, starts looking into the SPARCs involved. It focuses on promises that are most likely to be broken, either by the enterprise or its customers or suppliers.
Sofi then implements a procedure called sequencing. First, it converts each SPARC into a daisy chain of promises. It then artificially lengthens the chains by including what-if promises and the interests of stakeholders who aren't explicitly included in the SPARC but can make a difference. That includes ESG investors, governments and financial institutions who are willing to underwrite risk adjustments that promote profitability and sustainability. From a problem-solving perspective, the artificially long chains give Sofi a wider range of possibilities for balancing and distributing risks. Sofi makes small, speculative changes to a few selected promises and simulates a trading session. Each trade consists of small adjustments to the underlying commitments and considerations. Sofi looks for the smallest possible changes that lead to large improvements and iteratively resolves the gaps and conflicts that caused worry.
Sofi is able to do its work because of a system of accounting through which we can see every service is a joint venture with its own balance sheet of mutual assurances. In a given case or sequence, Sofi may go through hundreds of rows and columns to see 'who is promising what' (value), 'why and how' (assurance), and 'when and where' (window). The 'double-entry bookkeeping' is in words, not numbers. All entries are in the form of declarative statements. It's like functional programming for lawyers and accountants. The statements are the inputs and outputs of risk calculations. They define the commitments and considerations that are the structural units of SPARC. Each statement opens into a document that provides the reasoning and justification behind the words. The rich text of the document is annotated with facts and figures that support the statement — data, analysis and insight, routinely updated with real-time information, as and when there are changes in the markets in which an enterprise operates. For example, a container ship getting stuck in the Suez Canal, would trigger updates and flag several statements. That is also how Sofi knows there could soon be a problem somewhere in the SPARC chain that ship is a part of.
RUMI
Sofi personifies layers of human and artificial intelligence interacting with each other. Collaborative dialogue between humans and machines in the system is the powerful dynamic that allows the fast exploration of a wide range of possibilities, along several paths at once. The collaborations result in the authoring and editing of statements. Each statement, apart being an expression of meaning and intent, is proof of work, attached with data, analysis, and insight. The thinking that goes into each statement — the joint effort of humans and machines – is analogous to the computational effort that goes into cryptocurrency. Therefore, statements written for one case can be valuable in others, if they can be effortlessly included and inserted like software code. The 'business objects' are published to domain-specific libraries and frameworks; catalogued and indexed. They become reusable modules of intelligence (RUMI), packed with data, analysis and insight, on why-to and how-to promise certain types of demand or supply, in certain geographies, markets, and regulatory landscapes.
The entire system recursively creates new knowledge. When working on a case, Sofi always checks to see if there are statements it can reuse. Just like judges and lawyers look up case law for judicial precedents, Sofi searches RUMI libraries for 'commercial precedents'. For example, when helping a municipality reduce the total costs of street cleaning while also 'reducing carbon', Sofi looks into similar cases in the same regime or landscape. This habit not only saves time and effort but also bolsters a case with the solid arguments others have successfully made. When Sofi opens RUMI files, it may get full access to the data, analysis, insight, except whatever is kept hidden for reasons of confidentiality. Such access gives Sofi the confidence in analysing a problem, proposing changes, and resolving gaps and conflicts. Or, it might get limited access, and that means the system will have to generate new case-specific content and fill in the blanks. The level of access depends on who owns the files.
Management consulting, legal and accounting firms, are major content creators in the ecosystem. Not just the big, well-known firms, but also smaller, lesser-known ones who then get discovered like new artists on Spotify. Some RUMI files are exclusively available to the firm's clients and hidden from public search. But a majority are available for others to reuse and remix, through various licensing and subscription options. The RUMI are thus an additional source of revenue for firms big and small, but without the costs of putting consulting boots on the ground. Firms that supply data, analysis, and insight for the annotations, earn their own royalties and reputations. Academic institutions and think tanks also participate for their own natural purposes. Some RUMI receive high ratings and become more popular than others because. Some even become gold standards, like endpoint definitions in clinical trials.
What you have just read is of course a fictional narrative from an experimental project, originally written around the time ChatGPT was released. Solutions like Copilot, Harvey and Casetext will change the way we write software and prepare for cases in court. There isn't yet a 'Spotify for Analysts; Casetext for Consultants'. But it is only a matter of time.