"Black box" algorithmic contract: contract law and algorithmic contract
Oct 12, 2021 By Si Gyeongmin

We should recognize in the contract law that algorithms are not just tools, and we cannot overestimate that companies can predict the behavior of complex algorithms.


The "black box" algorithm contract already exists and is being used. DAOs are the clearest example of a "black box" algorithm contract. In the financial services industry, the "black box" algorithmic contract is the most common contract. However, the issue of enforceability still exists, and it is not clear at the same time whether it is possible to sue the court in this regard.


The "black box" algorithm contract is different from the ordinary contract. In the "black box" algorithm contract, both parties do not know the operation to be performed by the algorithm, and have never regarded reaching an agreement or pursuing an agreement only after gaining a profit as a contract. The two parties to the "black box" algorithmic contract are also unable to negotiate (such as "bargaining"). The basic assumption structure of contract law needs to be adjusted before the "black box" algorithmic contract can be applied.


Some principles and policies restrict algorithmic contracts. The "valuation principle" exposes the problem that the "black box" algorithm contract is valued higher than the market price due to user preference. "Uncertainty" is an important issue in the use of algorithm contracts, which will lead to an increase in transaction risks. "Liability pool" is necessary. Complicated and important tasks are completed by algorithms, so it is necessary to clarify the attribution of responsibilities. "Social welfare" is also an issue that algorithm contracts need to pay attention to, and the interests of disadvantaged groups cannot be infringed under the guise of the "objectivity" of algorithms.