Blockchain, the need for trust, and moral hazard
One of the most obvious advantages of blockchain technology lies in its “trustless” feature. This feature means that the database cannot be changed or manipulated unilaterally by a single entity, so that the users of the database need not worry that someone has secretly manipulated the data to their benefit. In other words, there is no need to trust anyone to manage the data.
With this in mind, it has often been argued that a “real” blockchain use-case is one where there exists some form of distrust among the users. While this is an extremely important point, it is sometimes misunderstood: there is no need for actual mistrust, i.e. it is not necessary that the current users do not trust one another. What should matter here is only whether there is a so-called “moral hazard” problem, where diverging interests and hidden actions intersect.
For those of you unfamiliar with the concept of moral hazard, this problem arises when, after a contract is signed, one of the parties can take a hidden action, which is not verifiable by the other side. For example, the problem may arise when an employer hires an employee, but can’t observe whether the employee works hard or shirks. For the context of blockchain, the question is then whether data manipulation constitutes such a hidden action.
I would like to demonstrate this point using two examples.
First, I recently came across some suggestions to launch a decentralized dispute resolution mechanism. For the discussion here, I would like to neglect the benefits that may lie in such as setting (e.g. lower coordination costs) and focus solely on the matter of (dis)trust. In a legal dispute, there are of course several important documents that may be stored in a database: decisions, petitions, power of attorney etc. However, is there a real concern of manipulation here? Suppose that a litigant has submitted a document that is later manipulated, i.e. the platform controller or a malicious entity revises the text. When this happens, it seems obvious that a manipulation occurred (even without knowing who the manipulator is) as all documents originate from the parties anyway, and a copy exists somewhere. One can think about the following aspect as well: people use arbitration services, because the arbitration process is transparent: everyone knows which documents are submitted. Hence, there seems to be no real fear of data manipulation – so that there should be no mistrust. It of course makes no difference whether people actually trust the arbitrator or not, what matters is that trust is not an issue to begin with. This first example illustrates the case where blockchain is not necessary from a trust perspective.
Conversely, as a second example, consider a case where you arrive at a restaurant and the host makes a waiting list. Suppose that the host is your friend – you know him and trust that he is an honest person who is unlikely to tamper with the list. Someone proposes to store the waiting list using blockchain – is this a real use case, given that there is no trust problem? The answer is clearly yes: there is a theoretical concern that after guests register to the waiting list, the host will want to revise it retroactively and hide the changes from the guests. For example, suppose that a famous person arrives at the restaurant and does not feel like waiting – so he offers a 1000$ “bribe” to the host, in exchange for letting him in first. The host may then change the list in a way that is unobservable to you. This second example illustrates the case where blockchain may be helpful, even when there is no current trust problem.
Another helpful way to think about this is through a thought exercise in which the guest and hosts become old, and their children – who do not trust one another – now interact. For them, there may be a trust issue.
Hence, I find it important to think about the issue of trust as whether there is a moral hazard problem or not, not whether people actually trust each other.
As a side note, I would like to point out two additional insights from the economic analysis of law (a.k.a Law&Economics) here:
First, the so-called "Coase theorem" tells us that if the parties can negotiate and transaction costs are low, then they will always reach an efficient solution anyway. Hence, we may not need to worry much about when blockchain is adopted; the parties should know better whether its adoption is needed.
Second, as mentioned, a decentralized dispute resolution mechanism may have some benefits that have nothing to do with immutability of data. For instance, if society is interested in adopting majoritarian rules, assuming that this maximizes efficiency, then letting a group of experts (or laymen) vote on the desirable legal outcome – and then impose it as a binding precedent – may be very helpful.