The Potential Economic Problems of Proof of Stake
Proof-of-stake (PoS) protocols are all the rage at the moment with most new blockchain projects, as well as Ethereum, using or envisaging to use them instead of proof-of-work (PoW) or other alternatives such as Byzantine Fault Tolerant (BFT) voting.
There are at least four types of possible PoS approaches. Some projects like EOS, Lisk and Tron have turned to the arguably simplest PoS solution, namely delegated PoS (DPoS), where a fixed set of privileged nodes validate blocks at each point in time. However, this type of PoS is arguably the least interesting and promising, as the associated governance and centralization issues will likely preclude it from becoming the basis for mass adoption. Staking can also be used as an element of hybrid consensus mechanisms such as the one developed by Thundercore which we recently covered. Meanwhile, the Algorand team is working on what they call ”pure PoS” where every node that has holdings of the network token will be able to participate in consensus.
However, it may well be argued that the only genuine PoS is the bonded PoS (BPoS) approach taken, among others, by Ethereum 2.0 (Serenity), RChain and Casper Labs. The reason for this claim is that in the three other cases, staking, if it even actually takes place, plays a subordinate role in the consensus. Only in the BPoS case, does the number of coins committed by the validators tend to determine their probability of approving a given block and, thus, their stream of revenues.
A lot of articles have been written and debates had by blockchain thought leaders about the complicated technical aspects of making BPoS work right, the possibility of ensuring liveness and finality, avoid frequent forks and so on. Also extensively covered have been the issues of incentive design for the ultimate arrangements, how the misbehaving stakeholders should be punished and so on. However, BPoS appears to raise important economic issues that go well beyond narrow incentive design, and lately, with the advent of staking as a service, people have started reflecting upon those.
In the history of money or means of exchange in general (I will use the term money or currency for both), this approach potentially represents a revolutionary step. Never before were monetary units directly necessary in order to secure transfers and holdings of monetary units of the same kind. Of course, one needs to pay a small fee to make a bank transfer or maintain a bank account but one can also just hold or give someone else cash directly. More importantly, even with the increasing virtualization of money, there is no tight connection between how people use money in a given period and the overall fees for banking services, and vice versa. It may not be easy to unpack this idea just yet but the two potential major problems this leads to will, hopefully, make it clear
But first, a few words about how BPoS works at a high level. Owners of nodes wishing to validate new blocks are supposed to contribute a certain amount of network tokens to the bonding smart contract where they will be held as a sort of guarantee of diligent behaviour. If a stakeholder engages in one of a set of well-defined wrongdoings, her stake is going to be deleted, or “slashed” in the industry jargon. In return for not using their tokens for other purposes, the staking nodes are entitled to block rewards created with the blocks that they published, and transaction fees, just as miners in the PoW context.
Ben Davenport’s zero holding equilibrium argument
To get back to the problems the BPoS approach of using money to secure money creates, the first important issue has recently been raised in an insightful Medium article by the cofounder of Bitgo Ben Davenport. At bottom, Davenport asks a seemingly simple question. If, in theory, almost anyone can stake, will anyone hold the BPoS-based network’s tokens when this means that their holdings will tend to be diluted by block rewards? The only new money that enters the system comes from block rewards, which means that over time, staking nodes will tend to accumulate more and more tokens, while the non-staking ones will together hold an unchanged amount.
Davenport believes that since token holders are not stupid, eventually, no one will hold network tokens for any prolonged period of time and everyone who is not planning to immediately use tokens for some purchases will tend to stake them. However, in this case, no one essentially gets any return on staking and block rewards create pure inflation. Which implies that once this happens staking may just break down.
While seemingly powerful, this objection to BPoS has at least two flaws. First, a widely used means of exchange (which none of the blockchain network tokens currently is) derives its value from use in exchanges. This means that the only way stakeholders can hope to profit from their gains is through making purchases before the relevant prices reflect the increased amount of money chasing the goods they want to buy. This is not much different from the way first receivers of new money created by central banks profit from inflation today, which economists know as the Cantillon effect. Thus, the stakeholders’ share of the total supply will at least not tend to grow as much as Davenport believes.
The second issue is uncertainty. Many people cannot precisely predict what exact amounts of money they will need to spend in the near future. At the same time, the bonded tokens cannot be promptly withdrawn if the need arises since the whole point of bonding is freezing the staked tokens for a certain period. Thus, a lot of people will still tend to hold substantial amounts of network tokens even though they could, in theory, stake them.
The feedback loop between currency adoption and stake
Davenport’s case against BPoS, thus, appears to be weaker than it seems but the tight link between staking and the maintenance of monetary integrity creates another potential challenge.
A few preliminary observations need to be made. While most current economists tend to consider the economy in terms of equilibria, economic phenomena are essentially dynamic. With regard to a nascent currency, this means that we cannot talk merely about some final stable state. As an example of just such a discussion, consider a recent Github discussion among the Ethereum 2.0 team members on staking and issuance. They clearly focus on how much ETH should be staked in the long run and not on the process of getting there.
However, a nascent currency will first be used to settle only a small part of exchanges, then, if it is successful, it will extend to even more exchanges, and so on. If the volume of exchanges made in such a currency grows faster than the number of its units, the prices denominated in it and its exchange rates will grow.
In the case of a BPoS-based network’s token, one of the consequences is that the network will become more and more attractive for attackers or misbehaving staking nodes. If the initial amount of total stake is relatively low, this will mean that additional staking will be required. But additional staking will reduce the number of tokens in circulation, which, given that the staked tokens are frozen for some time, will make the circulating tokens more valuable. Which will make the network even more attractive for attacks and necessitate the staking of even more tokens to boost the protection, and so on.
This may create a lot of volatility and uncertainty for the nascent BPoS-based network. Even if the added purchasing power volatility does not prove to be dramatic, it is unclear how the process of frequently increasing the total money staked can be made smooth. The reason is that changing the total money staked will dilute the stake of the current bonded validators and thus their expected returns. Which means that they might resist the increases in total money staked.
A potential solution may be to require staking of a sufficiently large part of the network token supply from the beginning. However, it may be tricky to achieve this since potential bonded validators may expect a major network token price boost upon the launch and be reluctant to commit too much. Perhaps, the way the problem could be addressed is through setting block rewards at a sufficiently high level. This approach, though, has its own downside as it raises the expected inflation. It remains to be seen whether a credible promise could be made to reduce the block reward in the future, once the network stabilizes around a certain transaction volume.
***
Regardless of whether the potential challenges of BPoS discussed in this article will prove truly significant, rolling out BPoS is set to be an intriguing economic experiment. Never before was a currency directly used to secure the system of holdings and settlements involving itself.
