The Debate on PoW and PoS Efficiency: Is it better to consume capital or to consume electricity?

This article is from Ethereum enthusiasts, originally titled: “Opinion | Beyond Power ------ PoS is Not the Savior,” written by KF.

Within and outside the cryptocurrency circle, more and more people are pinning their hopes on Proof-of-Stake (PoS) to contribute the advantages of cryptocurrencies while avoiding the energy-consuming attributes of Proof-of-Work (PoW). This wish will ultimately be in vain, as Paul Sztorc discussed this issue in his 2015 article “nothing is cheaper than proof of work” (Chinese translation). PoS merely obscures the costs but cannot eliminate them. Here, I want to explain why PoW is like democracy: it is the worst method, but it is the only one that works.

This article does not discuss whether the security of cryptocurrencies is “worth” such a high cost; it only focuses on whether PoS is an efficient method.

The best cryptocurrency should achieve security in the cheapest way

To reasonably compare PoW and PoS, we need to consider both costs (carbon footprint) and benefits (blockchain security). Therefore, our questions can be “Is the environmental cost of achieving a given level of security lower with Proof-of-Stake?” or “Given the cost, can Proof-of-Stake achieve higher security?” These two questions are logically equivalent. In this article, we will adopt the second formulation:

Axiom 1: The most environmentally friendly cryptocurrency should be the one that is most efficient in purchasing security.

Money is not magic. For money to circulate, a certain level of security is necessary, and there are differences in the efficiency of obtaining that security; the most efficient technology will also have the smallest environmental carbon footprint. This is a very useful perspective because it reminds us of a profound economic principle: marginal benefit equals marginal cost. If a blockchain spends X dollars on security, it means that miners will collectively spend X dollars to compete for that reward.

Axiom 2: If a cryptocurrency purchases security worth X dollars, its validators (to earn rewards) will collectively incur expenses worth X dollars.

This is an unbreakable economic principle about competition. Any miner who spends less than they earn will be defeated by those willing to spend more; conversely, any miner who spends more than they earn will eventually go bankrupt. A blockchain that spends X dollars on security will ultimately lead to miners spending X dollars to defend that blockchain. The proportions of different types of expenditures may vary, but the total expenditure will always be X dollars. This difference is akin to comparing a pound of cotton to a pound of iron.

Power is only part of the equation, not the whole

When people say “Bitcoin wastes energy,” they mean that electricity is needed to power the mining equipment that produces Bitcoin blocks; ironically, on one hand, this energy is clearly not the entirety of the energy used by the Bitcoin network, but only a part of it; on the other hand, this part is the most directly used to protect the blockchain, thus it is the least wasteful. To comprehensively measure Bitcoin's carbon footprint, we must also consider hardware costs (manufacturing and disposal costs) and normal business operation costs (such as renovating a decent office and flying around to attend Bitcoin conferences). The latter two types of costs are more ambiguous and harder to estimate. However, they are equally important to understand the overall impact of the system.

Axiom 3: To consider a system's carbon footprint, you must account for all aspects of validator expenditures.

Each of these three activities will incur waste. The Bitcoin network will inevitably experience orphan blocks, and the energy used to mine these orphan blocks is wasted. Some hardware is produced, but its operational efficiency is not perfect. Some operational expenses are necessary, while others are merely for a change of scenery to have tea. The more closely an expenditure is related to the actual work of validating the blockchain, the higher the efficiency of purchasing security it achieves. This is similar to how machines operate more efficiently when there are fewer degrees of freedom: adding more components leads to more useless work. The same principle applies to economic systems—more structures lead to lower efficiency.

For example, a business trip to build a relationship with a major client can be abstracted as improving overall business efficiency, thus helping with purchasing security—but its effect is very indirect, and its effectiveness in purchasing security may not be as good as spending the same amount on electricity. The more a cryptocurrency directly spends its security budget on electricity, the higher its security; the more it spends on business operations, the lower its security. The CEO of a mining company may be necessary for the company itself, but they contribute little to the network's security.

Axiom 4: The more an expenditure is related to the actual actions of validating the blockchain, the higher the efficiency of purchasing security.

Thus, sometimes you see people discussing ASICs, believing they harm a cryptocurrency's environmental impact, but in reality, it is quite the opposite. The specialization of mining equipment allows for greater computational power per unit of hardware, meaning that less of the blockchain's security budget is spent on hardware and more on electricity. This is good for efficiency—ASIC mining is fundamentally cleaner than traditional CPU mining because ASICs are more efficient (and thus cheaper) at converting energy into security.

This is also why people misunderstand the environmental impact of cryptocurrencies like Proof-of-Space (e.g., Chia) (which use memory rather than computational power to validate the blockchain). Using hard drives as a scarce resource for generating security does not lower costs; it merely reallocates the blockchain's security budget from electricity to hardware. If you only look at electricity consumption to measure a cryptocurrency's carbon emissions, you might think we have made significant progress—but if you also consider hardware expenditures, you will find it is a significant loss of efficiency.

Currently, there are many idle hard drives in the world, so Chia effectively receives some subsidies. But don’t expect this to last forever. Once a cryptocurrency successfully monetizes computer memory, what has happened with GPUs will repeat in the relevant hardware field. More importantly, ultimately this system will be worse than an equivalent PoW-based system because the security budget is spent on developing hardware, which is less efficient than directly spending the security budget on electricity.

Tom's Hardware has reported on the skyrocketing price of Chia and the shortage of large-capacity mechanical hard drives. But the more severe issue may be the shortage of solid-state drives, as you also need solid-state drives to create Bram Cohen BinggoTM cards that can fill large hard drives.

Conclusion 1: The most environmentally friendly cryptocurrency is the one that spends its security budget on the matters most closely related to validation work.

Locking capital also incurs real costs

Chia and other blockchains using Proof-of-Space are relatively easy to compare with PoW because you can easily imagine old hard drives filling landfills, making it easy to believe they “really” impact the environment. However, lost capital also has a real impact on the environment. Because that money could be spent on carbon capture technology, carbon efficiency research, and other matters. People “feel” that they are free (without consumption) simply because humans are not good at thinking about the time value of money. If capital were truly free, miners could obtain it without any cost.

Sometimes you will also see people argue that Proof-of-Stake may have waste from an individual perspective, but from a societal perspective, it is a form of savings, as Dan Robinson said:

Suppose your religion requires you to prove your sincerity through your sacrifice. You can throw a cow into a volcano or give it to a stranger. Both actions cost you the same (thus satisfying the religious requirement), but the social cost of the first is higher.

The reality is quite the opposite. Locking capital is not wasteful for you personally because the staking algorithm of PoS compensates you—this capital is not thrown into a volcano; it earns you returns at the expense of other ETH holders. Locking capital is entirely different from giving it to a stranger because there is no stranger taking over these funds. Validators do not “distribute their capital evenly among everyone”; in fact, everyone pays equally for the validators’ work (in the form of newly mined ETH). Moreover, thanks to Axiom 2, we know that the payments received by validators are exactly equal to the cost of those funds.

On the other hand, locking capital is a waste from a societal perspective because that capital can no longer be used to build factories, fund research, and do everything beneficial to society. Think about the 2008 financial crisis—no factories were destroyed, no houses collapsed, and there was no loss of any tangible assets. The consequences were merely capital loss—but this is still a very heavy social cost. Capital is a special form of accumulated information, condensing information about the optimal allocation of resources. Losing information is not the same as losing tangible things, but it is still a loss, and perhaps even worse.

So, keeping Axiom 1 and Axiom 2 in mind, a fair comparison of PoS and PoW should look like this:

Proof-of-Stake does not (and cannot) eliminate miners' expenditures; it merely converts the portion of PoW system expenditures on electricity into capital expenditures. The externalities of locked capital compared to the externalities of electricity consumption are complex and subtle issues—but the unfortunate reality is that PoS system supporters often pretend that electricity is the only cost that needs to be paid for the system to operate:

This intentionally misleading comparison is taken from the Ethereum Foundation blog.

Disappointingly, the Ethereum Foundation is well aware that “average energy consumption per transaction” is a completely meaningless metric, and using this data in such a way is intentionally deceptive. The key is not whether the PoS system uses less energy—indeed, it does—but that is not the whole picture. The meaningful discussion is: Is it better to consume capital or electricity? Any theory supporting PoS without discussing capital costs is fundamentally flawed.

Efficiency Comparison: Staking vs. Mining

In the PoW/PoS debate, PoS supporters often have an advantage: there has never been a fully decentralized Proof-of-Stake system. This also means that PoS supporters can fantasize about how efficient and elegant the final solution could be. Known mining algorithms are not perfectly efficient (for example, there is the issue of “selfish mining”), and there is no reason to believe that PoS systems can be flawless—however, until the solution is fully formed, PoS supporters do not have to consider the specific strategies validators might use to deceive the system.

The more opportunities there are to manipulate the system, the looser the relationship between the security budget and actual validation becomes. Because “no one has ever tried a true Proof-of-Stake,” we first do not know how much of the security budget can be converted into locked capital, nor do we know how efficient that locked capital is in generating security. By definition, PoW is tightly coupled with validation, but PoS, as of the time of writing, has not been fully defined.

Observation 1: PoW is tightly coupled with security. But it is unclear to what extent PoS can achieve this.

Externalities: Staking vs. Mining

Assuming we all agree that, from the perspective of security efficiency, locking capital and mining are equally efficient, then the only remaining question is “externalities.” This is also where I believe rational discussions can still have differing views on the trade-offs involved.

In the short term, PoW systems must compete with other projects for energy use, but in the long term, they will incentivize the development and production of cheap energy, which will gradually lower electricity costs and benefit everyone (whether for financial purposes or environmental protection). Cheap electricity will also spur new technologies and more production.

Proof-of-Stake must compete with other projects for capital in the short term, but it cannot incentivize the creation of more capital, so it will gradually increase the cost of capital. Higher capital costs mean fewer projects—fewer factories and less research.

Whether you prefer the externalities of PoW or PoS may depend on whether you believe it is better to solve the environmental crisis through investment in technological development or by reducing current consumption. To save the planet, should we create cheaper renewable energy, or should we make everything more expensive?

Observation 2: Proof-of-Work encourages future investment, while Proof-of-Stake stifles exploration.

PoS may not even work

As we mentioned earlier, we still do not know whether a sufficiently decentralized Proof-of-Stake system can be realized. Existing cryptocurrencies that claim to be PoS systems largely rely on certain centralized coordinators or “checkpoints.” Since Ethereum launched in 2015, they have been working to transition to a fully decentralized PoS system, and to date, they have only released Phase 0, which merely supports validators staking… Once you lock your money in, you cannot get it out or spend it.

There is much debate about whether PoS systems can achieve sufficient security (see “Long-range attacks”、“nothing-at-stake”、“Work crosses time / stake does the opposite”). These are open research questions, not items on a to-do list for implementation. We do not know how long it will take to find a good solution, or even if a good solution exists. Consider this 2017 post:

PoS may “run,” but the way it operates is pathological. PoW favors those who can obtain cheap electricity and rewards them with capital. While not entirely equal, it does not fall into a positive feedback loop. PoS, on the other hand, favors those who can obtain capital at a low cost and rewards them with capital, creating a cycle. The wealthy become wealthier, and the richer one is, the easier it becomes to become even richer.

If Proof-of-Stake “can start,” but the end result is that all power is concentrated in a small number of super-wealthy institutions that can permanently control the system, then it cannot solve the problems that cryptocurrencies aim to address. We might as well continue using the current system composed of central banks and those “insiders benefiting from the Cantillon effect.” We hope cryptocurrencies can break this system, not just put new wine in old bottles.

Observation 3: Proof-of-Work already exists and has proven its ability to continue to grow. But Proof-of-Stake remains an unresolved issue.

Conclusion

• Axiom 1: The most environmentally friendly cryptocurrency should be the one that is most efficient in purchasing security.
• Axiom 2: If a cryptocurrency purchases security worth X dollars, its validators (to earn rewards) will collectively incur expenses worth X dollars.
• Axiom 3: To consider a system's carbon footprint, you must account for all aspects of validator expenditures.
• Axiom 4: The more an expenditure is related to the actual actions of validating the blockchain, the higher the efficiency of purchasing security.
• Conclusion 1: The most environmentally friendly cryptocurrency is the one that spends its security budget on the matters most closely related to validation work.
• Observation 1: PoW is tightly coupled with security. But it is unclear to what extent PoS can achieve this.
• Observation 2: Proof-of-Work encourages future investment, while Proof-of-Stake stifles exploration.
• Observation 3: Proof-of-Work already exists and has proven its ability to continue to grow. But Proof-of-Stake remains an unresolved issue.

Regrettably, after detailed examination, we find that most of PoS's “advantages” are merely superficial, and there are many significant public challenges. At least for now, it remains a utopian fantasy rather than a realistic choice. While PoW has its drawbacks, it is still the best-known solution for building a network without a trusted centralized authority (in fact, it is the only solution).

Footnote

Application-Specific Integrated Circuits (ASICs): Customized computers that can only perform a limited type of computation and cannot do anything else (ASICs developed specifically for certain algorithms are therefore very efficient for mining).

If we realistically consider human nature, the only way to reduce consumption globally is to improve technology or reduce the population. Any philosophy advocating direct “reduction of consumption” is either very naive or very frightening.