There is a common view among bitcoin advocates, which also prevails to some extent among economists working in the Austrian tradition, that decentralization is always good and more-decentralized solutions are always better. In both forms, this view starts with a reasonable premise. Bitcoin revolutionized payments by enabling digital transactions to take place between strangers without a trusted third party serving as a centralized clearinghouse. Ludwig von Mises argued that economic calculation was impossible under central planning, while Friedrich Hayek explained that markets relied on decentralized knowledge of a particular time and place. It does not follow in either case, however, that decentralization is always good and more-decentralized solutions are always better. I call that view the decentralization fetish. We should reject the decentralization fetish and, instead, think carefully about the optimal degree of (de)centralization, which might vary across applications.
Let’s start with Hayek. He seems to have been more concerned with decentralization as a property of the institutional environment rather than as a description of the outcomes. Consider the meaning of competition. Competition, in Hayek’s view, is not marked by a large number of atomistic firms––what economists might call price-takers. Competition is about rivalry. Perhaps there are only two firms. Perhaps there is only one firm and the threat of entry from a second firm. It is tempting to look at such situations and conclude that they are pretty centralized. But the institutional environment can be decentralized, even if relatively centralized outcomes result.
We can say much the same about bitcoin and other blockchain applications. It’s great to have the option of decentralization. But it doesn’t follow that this option should be exercised in every state of the world. Sometimes more centralized outcomes are preferable and, as such, might persist in an otherwise decentralized institutional environment.
Before thinking about bitcoin, consider the extent to which monies are essential—or, useful for making transactions—in general. In a provocatively titled article, Nobuhiro Kiyotaki and John Moore maintain that “Evil is the Root of All Money.” The basic idea is that we don’t need money if we can trust each other enough to engage in reciprocal gift-exchange (i.e., producing for others when others desire our production). Money, or some record-keeping device, is useful because it enables exchange to take place when trust is hard to come by.
That money is useful is beyond dispute. We would not have highly-specialized economies—and, correspondingly, the high standards of living we enjoy today—without money. It does not follow, however, that money is always useful. Monies are useful in situations where trust is limited. But sometimes it is better to trust than to use money.
Long-term intimate relationships serve to illustrate. At its core, a long-term intimate relationship is a bundle of mutual exchanges carried out over a long period of time. We don’t typically use money in those transactions. My wife doesn’t charge me for dinner. She doesn’t pay me when I mow the lawn. We rely on trust and reciprocity, which enables gift exchange, because it is more convenient than using money in ongoing, high-trust relationships.
Does this deny that money is useful? Of course not. We can recognize that money is useful without concluding that it is sufficiently useful in all situations to warrant use in all situations. Sometimes we use money. Sometimes we rely on other institutional mechanisms to make transactions.
The decentralization fetish is especially common among those interested in bitcoin. Many bitcoin proponents rightfully recognize that it makes centralized clearinghouses, like banks, unnecessary for processing transactions. They then jump from unnecessary to undesirable. Some go so far as to claim that bitcoin is good because banks are bad.
That bitcoin makes it unnecessary to rely on centralized clearinghouses, like banks, does not mean banks are undesirable. Banks require trust to process payments between their customers. And that trust might be broken. But banks also engage in financial intermediation, channeling savings into valuable investment projects. Historically, banks have used their payment-processing facilities to acquire funds for intermediation. If we look around the world, places with deeper financial systems––i.e., where there is more financial intermediation––tend to have higher incomes and grow more rapidly.
It is good to be able to make payments without needing to trust a third party. It is also good to engage in financial intermediation. The relevant question isn’t whether we should use bitcoin or banks, but when to use bitcoin and when to use banks. It is not, in other words, an all or nothing proposition. For some transactions, the cost of foregone financial intermediation will be worth bearing in order to realize the benefits of finality and financial privacy made possible by bitcoin. For other transactions, the benefits of bitcoin will not warrant the requisite costs.
One might push back by noting that, with decentralized finance (DeFi), we can use the blockchain technology to engage in financial intermediation without trusting a bank or other third party, as well. And that’s true! But, once again, can does not imply should.
It is very difficult to identify all of the potential problems that might arise over the course of a contract and specify precisely how the parties to the contract must behave in each scenario. The cost of writing a contract increases with the degree of specificity. In some cases, the benefits of smart contracts are just too small to warrant the costs; we’d be better served by writing a traditional contract and delegating a third party to resolve any disputes that arise.
Might the trusted third-party adjudicator violate our trust? Sure. But, for some transactions, we might prefer to bear that risk rather than incur the costs of DeFi.
It is good to have decentralized options. But we must not merely assume that decentralization is always good and more-decentralized solutions are always better. Instead, we should recognize that it depends on the unique benefits of decentralization and the costs of realizing those benefits relative to more-centralized alternatives. We should reject the decentralization fetish in favor of clear explanations that recognize the relevant tradeoffs.