A common claim among bitcoin enthusiasts is that it is a “decentralized” method of making payments. Here are some notable outlets making this claim:
Investopedia: “Bitcoin offers the promise of lower transaction fees than traditional online payment mechanisms and is operated by a decentralized authority, unlike government-issued currencies.”
Business Insider: “Because bitcoin is decentralized, it’s not directly subject to market forces such as interest rates or currency debasement.”
This is in addition to the hundreds of other websites, both professional and amateur, that assert that bitcoin is a decentralized system.
But a new working paper from economists William J. Luther and Sean Stein Smith is casting doubt on this characterization of bitcoin. Luther and Smith offer a new taxonomy of the different methods of processing payments: centralized, decentralized, and distributed. The differences may seem superficial, but the implications can be significant.
Before we begin defining and understanding the different systems, it’s best to review some basic concepts.
First, let us define a medium of exchange as a good which is acquired in order to be exchanged for another good. Second, let us define money as the most commonly accepted medium of exchange. Third, we must distinguish between the production of money and the verification of an exchange: the production of money is creating new monetary units and adding them to the system; the verification of an exchange (a.k.a., clearing or processing) is determining whether sufficient funds exist to fulfill a transaction. We can go further and define trust as the belief that a transaction will be verified before it actually is.
With those preliminaries out of the way, we can now proceed to understand the three different systems of payment.
A centralized payment system has all transactions going through a third party for verification. The “third party”, in practice, may be several distinct parties networked as a series, with one verifying the work of another in sequence. The key feature is, however, that if the third party is unable or unwilling to process a payment, the two parties in a transaction will be out of luck.
Under a centralized system, there is effective a monopoly on verification. No one other than the centralized authority is permitted to produce money or verify transactions. As such, all other users must trust the judgment of the central authority.
A decentralized system is one where there are many independent verification parties. As an extreme example, a system where every transaction has no intermediaries (like a barter system) is a decentralized system. But decentralization exists on a spectrum: a situation where there are dozens of independent firms competing for the privilege of verifying a transaction is also a decentralized system.
This was the case in the era of private coin production. In the days of commodity monies, any private person with access to a mint could create their own money. Producers would compete on the aesthetic and value-to-weight ratios of their coins. The economist George Selgin documented perhaps the golden age of private coinage, England of the 18th century, in his book Good Money:
The commissioning and issuing of commercial coins, which had been the preserve of a few industrial and mining firms, was taken up by all sorts of small businessmen – grocers, drapers, silversmiths, malsters, and pretty much anyone whose dealings generated a need for small coin. Well, not just anyone: even small-scale token issuers were almost always persons of good standing in their communities, whose token issues were generally modest in comparison with their capital and command of credit.
Selgin (2011, p. 123)
Of course, kings and other nobles have been exercising virtual monopolies on money creation for millennia. So, what had happened to cause a break in the centralized system? According to Selgin, the Royal Mint in England stopped producing low-value copper coins as a cost-saving measure, despite the fact that there was great demand for them. Conveniently, there were many copper mines that were willing to sell their raw metal to those private individuals who would go on to mint coins. As metals were the commonly accepted medium of exchange (otherwise known as money), anyone with access to metals who could fashion them into something attractive for consumers could create his own money. And as Selgin (2011) documents, they were indeed attractive:
The other thing most eighteenth-century tokens had in common… was their extraordinary appearance. According to Francis Klingender… the tokens displayed a unique ‘combination of intellectual vigor, social consciousness, and imaginative design’ (Klingender , 46.) [Citation corrected.]
Selgin (2011, p. 133).
Under this system, the functions of producing money and verifying transactions are divorced. The mint buys the raw materials and produces the money. A person then makes a direct exchange with the producer for the money, with no third-party intervening. The person who now owns the money will get use it at a new transaction. The transacting parties themselves verify each transaction, while the producers focus on minting easy-to-verify coins. If the users of the coins lose their trust in a coin, they stop patronizing the producer and he or she goes out of business.
Since every step in this process involved only two transacting parties, these are all decentralized markets.
Eventually, of course, the central powers took notice of what was happening and took over the money production process again. The initial excuse is to ensure safety of the parties. Then overtime more duties are assumed by the regulator. This is called regulatory creep. In the 1900s, regulatory creep led to the advent of the modern central banking system with the systematized backing of fractional-reserve banking (FRB). Under an FRB regime, while the central bank and government mint create so-called “base money” (cash and ex nihilo virtual deposits into the accounts of chartered banks), it is the private banks that end up generating most of the media of exchange in the economy. Each bank does this by lending out the deposits of their checking account depositors.
Since each bank makes the decision to expand the supply of money via loans of demand deposits, the system functions as a decentralized network of money producers. However, the verification of transactions is still centralized. Thus, the current monetary regime is a blended centralized-decentralized system.
That is, until the existence of bitcoin. Bitcoin is neither a centralized nor a decentralized system. Instead, it is a distributed system.
A distributed monetary system distributes the role of verifying transactions to everyone on the network. This is distinct from a decentralized system, where the power of creating and verifying money is split up among many people. A distributed system has a two-fold approach to verification: first, the entire network shares a ledger that documents every transaction that has ever happened; second, the network has a shared protocol or procedure for verifying an update to the ledger via some kind of consensus rule.
The principles of distributed monetary systems come from the world of distributed computing. The major innovation of bitcoin was that it was the first to recognize how distributing trust among the entire network can be an attractive method of transacting. In effect, bitcoin is a trustless payment network, as buyer and seller no longer have to trust each other or a third party. Instead, only trust in the faithful execution of the automatic protocol is required.
This removal of interpersonal trust is a giant achievement. Trade axiomatically makes us richer. But in order to trade, we must trust the person we are trading with. Historically, trade has been mostly within tribes, among family members or other closely-knit individuals where trust was high. Trust was the only way to exchange.
After intertribal and cross-regional trade was discovered, and exchanging with strangers became a common occurrence, the necessity of money became apparent: it is difficult to know what others want, but everyone will want money. Direct exchange gave way to indirect exchange, albeit decentralized. While money superficially looks like an inefficiency, it facilitated a much smoother market. A decentralized market does not require a fully trusting society, but just enough trust that you find it unlikely to be taken advantage of.
The early markets were decentralized. Buyers and sellers now only needed to trust each other in an exchange. Over time, as tribes settled down to form cities, kingdoms, and empires, centralization of trade in the form of government fiat money, government mints, and so on, became the norm. Trust from the counterparty was replaced with trust in the centralized authority. However, centralized authorities have tendencies to restrict trade on a whim, in the name of security, nationalism, or other infamous ends.
Bitcoin offers a way forward. There is no longer a need to rely on a mercurial central authority to verify transactions and create money, nor do you need high trust. By extending the scope of the market, many more exchanges can happen, creating more opportunities for innovation, cost-reductions, and other benefits of making a connection with other people.
Luther and Smith (2020, pp.14-25) do point out, however, there is more to bitcoin than its distributed payment network. Firstly, it is the governance of the protocol itself. Here, changes to the protocol must be arrived at via popular consensus among the developers and miners. We can describe the decision-making process as broadly decentralized, despite the fact that some mining pools are more influential than others. Secondly, there is also the issue of exchanges and e-wallets. As they are third parties that verify transactions, they are centralizing forces in the bitcoin space.
The original bitcoin whitepaper does not mention “decentralization” at all. Instead, the system was called a “peer-to-peer distributed time-stamp server.” Why did a distributed system become mislabeled as a decentralized one? Perhaps because many have an intuition that the current regime is highly centralized; and since bitcoin is not centralized, it must be the opposite: decentralized. However, there is a third way of organization: distribution.
A distributed system is not the same thing as a decentralized system. Decentralized systems require high trust between buyer and seller; meanwhile distributed systems require all peers on the network to share and communicate with each other constantly. As such, distributed systems can offer less privacy than a decentralized system, wherein all exchanges are only between buyer and seller. Furthermore, distributing among a large number of participants the responsibility to store every transaction between every participant on the system, may prove to be more costly (in terms of storage costs, energy costs, and time to approve a transaction) than a centralized system where only one entity is responsible.
On the other hand, decentralized systems can be cumbersome, costly, and the degree of trust required to fulfill them can serve as a hinderance to trade. A move towards decentralization would mean more trust is needed to engage in an exchange, in addition to regulatory creep. By reducing the level of trust required to verify a transaction, bitcoin has the potential to open up trade among more strangers who otherwise wouldn’t trust each other.
Luther, William and Sean Stein Smith (2020) “Is Bitcoin a Decentralized Payment Mechanism?” Social Science Research Network.
Klingender, Francis Donald (1943) “Eighteenth Century Pence and Ha’Pence.” Architectural Review 93: 41-46.
Selgin, George. (2011) Good Money. Independent Institute.