Permissioned or permissionless? A digital currency can’t be both

In a short piece in the Federal Reserve Bank of New York’s Liberty Street Economics blog today Rod Garratt, Michael Lee, Brendan Malone, and Antoine Martin critique the use of the terms “account-based” and “token-based” in categorizing digital currency schemes. “Use of the token-based and account-based terminologies for digital payments is problematic because it does not create mutually exclusive categories,” they write.

Here’s their conclusion:

Classifications can be a powerful tool to organize and communicate ideas. The main allure of distinguishing between account-based and token-based is to highlight a defining feature of certain new, emerging forms of digital currency. [Read CBDCs.] But if a digital currency can be both token-based and account-based, then the classification loses its power to meaningfully distinguish between new and existing methods of digital payments. Furthermore, it may slow down progress in understanding intrinsic differences between the growing set of digital payment options and technologies. Future classifications could modify the definitions of the terms account-based and token-based to more clearly distinguish them. In the meantime, perhaps these terms should be retired to avoid further confusion.

To prove their point, they argue that a distinguishing feature of an account-based digital payments system is that they employ “a process for verifying the identity of the would-be payer” and, given that, the quintessential token-based digital currency, Bitcoin, is also account-based.

Bitcoin fits the definition of an account-based system. The account is a Bitcoin address, and the private key is the proof of identity needed to transact from that account. Every time a Bitcoin user wants to spend Bitcoin, that user must verify their identity by using their private key. It is not relevant whether the system requires users to reveal their true identity. Rather, what matters is whether a user must follow a process the system has developed for verifying the identity that they established within the system, whatever that may be. Analogously, a bank that wants to move funds through the Fedwire Funds Service has to comply with the Reserve Banks’ security procedures, which includes a set of access control features.

​This is, by definition, a semantic argument, but probably one worth having. I don’t share the authors’ concern that use of the terms is “problematic” or confusing, though I agree that the language used to talk about CBDC can probably be sharpened, so let me offer an alternative set of terms.

As I’ve noted before, when using the account-based and token-based terminology, the distinction that’s usually being communicated is between permissioned and permissionless systems. In a token-based system like Bitcoin, identity verification (if it can be called that) is achieved via public-key cryptography (i.e. math) which is available to everyone. Anyone can create a Bitcoin address at will and receive payment without first seeking anyone’s permission. In contrast, account-based systems employ a third-party (or parties) to conduct the identity verification, thus creating a point of permissioning. Use of such a system is at the pleasure of the identity-verifying third party.

I don’t think it matters much that commentators say “account-based” and “token-based” when what they more specifically mean is “permissioned” and “permissionless,” but to the extent one wants to be more precise, those are the words I would suggest.