>Since they don’t need to expend resources to stake, they can simply increase their overall staking amount as they earn ongoing coins from staking rewards, and exponentially grow their influence on the network over time, forever. Network dominance tends to lead to more network dominance, in other words.
So, literally like reinvesting profits in proof of work?
In PoW centralization forces are way, way worse for two reasons.
The first is because there are infinite economies of scale: as any mining operation grows, costs per hashrate drops. The starting point right now is a person mining at home, paying residential electricity prices and using air cooling, with an asic bought from an external supplier at a high markup. The end point is a mining company that uses its own power plants and its own chips, with custom made power supplies and cooling that potentially reuses the heat in some way.
The second is that mining is vulnerable to forced exclusion that can increase revenue per hash by orders of magnitude: one that's consensus based, and another economically based which can force a consensus-based exclusion.
Controlling 51% of hashrate means that every other miner can be ignored, almost doubling revenue per hashrate. However, it's actually worse - because as competitors close, the dominant entity can reduce its active hashrate to a level just above remaining competitors, until nobody else mines, at which point mining can reduce to a very low level, just enough to keep random small miners from winning. In the case of an attack full hashrate is utilized, ensuring it fails. By the virtue of nearly ~100% of revenue being profit, the entity can amass and unassailable amount of idle mining hardware.
Economic exclusion means that it's possible to intentionally mine at zero profit to drive difficulty too high, ensuring others turn off. It's a strategy for a biggest miner that doesn't have majority of mining hardware, but by the virtue of being the largest entity has the lowest unit costs.
Of course, such an entity would have absolute censorship power.
It's the lowest energy point of the system - because while it's in principle possible to replace such an entity, it can be only be done by an even bigger entity, only changing the name at the top, but still retaining full centralization.
Those design flaws ensure PoW is certain to centralize, given time. The process is already well advanced - it started with individuals mining on their home pcs. Now miners are buying power plants.
PoS has next to zero economies of scale (just the negligible fixed cost of a computer used for staking), and there's no potential of forced economic exclusion - running cost are negligible, so while more stakers mean profit rate goes down, it can never force a loss. Additionally, while it's not implemented yet (at least I'm not aware of anything), the fact it's a semi-synchronous protocol also allows for automated censorship detection - with appropriate punishments. That's because active nodes would be able to see correct messages that should have been included, but aren't. This is fundamentally impossible in PoW, as it's a fully asynchronous consensus system.
Socially, PoS is as decentralized as coin distribution - which for some networks is a joke - especially Solana which only sold 1.6% of its initial supply to the public - probably the most centralized L1 coin in existence, controlled by just few VCs. Ethereum sold 83.33% of its initial supply (72M) in a public kycless sale, since then 42M eth have been mined. Supply is very well distributed among thousands of mostly tech nerds that bought before the current boom. As recently as in March 2020 one eth could be bought for $90. This is the most perfect population achievable for resistance to government interference.
In contrast, for already mentioned reasons, mining is already controlled by big companies.
Censorship is a more realistic risk from regulations forcing miners to censor. With PoW, there's no defense except moving to a different country - because it's impossible to hide warehouses with industrial power connections, not to mention power plants. In reality, as long as regulations don't kill the network outright ('just' an OFAC blacklist, freezing coins from ransomware, theft, other criminal acts) - and these regulations are enforced in places with cheapest available electricity - miners would comply. Once those miners attained a sufficient majority, laws would force them to orphan non-compliant blocks.
PoS staking is trivial to hide - it's just a computer with a network connection. The whole network could be made anonymous by utilizing onion or garlic routing. Regulations would be impossible to enforce.
In conclusion, PoS is superior to PoW in every single fundamental factor mentioned in the article, especially censorship resistance.
I don't agree that complexity itself is a factor for or against: complex things are certainly harder to build, but once finished and proven to work that becomes irrelevant - only the functional differences remain.
There's only one real advantage of PoW over PoS: PoW is a coin distribution method, while PoS isn't. However, 90% of all bitcoins are already mined, so that argument is already very weak and becomes weaker with each day.
At the end, these discussions are for nerds - the average person may just get the impression that 'person X is saying PoS is less secure' and be wary for a while, but given enough time of working well, ~nobody is going to care about the anti-PoS position.
Additionally, belief in PoW is economically self-punishing: mining costs billions per year, ensuring that PoS supporters, as a group, are getting relatively wealthier to PoW supporters. A cost difference of double-digit billions per year is very hard to sustain over longer periods. It also means that a PoS network can continue working with next to zero buyers - while below some level PoW becomes hopelessly insecure (see eg. 51% attacks on ETC) - making PoS much more resilient to economic shocks.
This is a great post but I have one disagreement about proof of work. At the point that one entity controls 51% reliably and consistently, presumably users abandon the cryptocurrency altogether. So any discussion of what the steady-state looks like after that seems pointless.
If so, the profit from a 51% attack would have to be taken very quickly during an initial double-spend, which limits the incentive to do it. Unfortunately, returns to scale, as you mentioned, still can mean that mining can trend toward an oligopoly. And that could have pretty bad effects if the biggest miners collude in ways that users don't like, but that aren't bad enough to drive them off the currency completely.
>At the point that one entity controls 51% reliably and consistently, presumably users abandon the cryptocurrency altogether.
Why? There are many people that would even welcome a fully government compliant mining entity. The 'business types' like Michael Saylor. He co-founded 'Bitcoin Mining Council' which is a perfect foundation for a mining cartel. Another cofounding member is Marathon - which is already censoring bitcoin blocks according to OFAC sanctions list.
Some old idealistic bitcoin would maybe sell, but that would be the extent of a negative market reaction.
Ironically it's turning out that banning mining in China greatly accelerated centralization, because cartelization of Chinese and non-Chinese miners would be hard to achieve. Contrary to China, America thinks it has global jurisdiction, so it's possible it may try to enforce future mining laws even on foreign miners. While a miner company in eg. Kazakhstan won't be closed outright, it would mean that owners of that company would risk extradition to America if they traveled to most Western and America-allied countries. Easier to just comply. Electricity in Europe is too expensive for it to matter, so mining in America naturally dominates.
I don't actually expect that to happen, because I think bitcoin will become irrelevant long before those slow trends manifest in censorship enforced by orphaning non-compliant blocks, but in the absence of that, it looks virtually certain that the mechanisms I described would manifest as an American government-sanctioned entity that every miner, globally, has to join and comply with all rules or get its blocks orphaned.
So, literally like reinvesting profits in proof of work? In PoW centralization forces are way, way worse for two reasons.
The first is because there are infinite economies of scale: as any mining operation grows, costs per hashrate drops. The starting point right now is a person mining at home, paying residential electricity prices and using air cooling, with an asic bought from an external supplier at a high markup. The end point is a mining company that uses its own power plants and its own chips, with custom made power supplies and cooling that potentially reuses the heat in some way.
The second is that mining is vulnerable to forced exclusion that can increase revenue per hash by orders of magnitude: one that's consensus based, and another economically based which can force a consensus-based exclusion. Controlling 51% of hashrate means that every other miner can be ignored, almost doubling revenue per hashrate. However, it's actually worse - because as competitors close, the dominant entity can reduce its active hashrate to a level just above remaining competitors, until nobody else mines, at which point mining can reduce to a very low level, just enough to keep random small miners from winning. In the case of an attack full hashrate is utilized, ensuring it fails. By the virtue of nearly ~100% of revenue being profit, the entity can amass and unassailable amount of idle mining hardware.
Economic exclusion means that it's possible to intentionally mine at zero profit to drive difficulty too high, ensuring others turn off. It's a strategy for a biggest miner that doesn't have majority of mining hardware, but by the virtue of being the largest entity has the lowest unit costs.
Of course, such an entity would have absolute censorship power.
It's the lowest energy point of the system - because while it's in principle possible to replace such an entity, it can be only be done by an even bigger entity, only changing the name at the top, but still retaining full centralization.
Those design flaws ensure PoW is certain to centralize, given time. The process is already well advanced - it started with individuals mining on their home pcs. Now miners are buying power plants.
PoS has next to zero economies of scale (just the negligible fixed cost of a computer used for staking), and there's no potential of forced economic exclusion - running cost are negligible, so while more stakers mean profit rate goes down, it can never force a loss. Additionally, while it's not implemented yet (at least I'm not aware of anything), the fact it's a semi-synchronous protocol also allows for automated censorship detection - with appropriate punishments. That's because active nodes would be able to see correct messages that should have been included, but aren't. This is fundamentally impossible in PoW, as it's a fully asynchronous consensus system.
Socially, PoS is as decentralized as coin distribution - which for some networks is a joke - especially Solana which only sold 1.6% of its initial supply to the public - probably the most centralized L1 coin in existence, controlled by just few VCs. Ethereum sold 83.33% of its initial supply (72M) in a public kycless sale, since then 42M eth have been mined. Supply is very well distributed among thousands of mostly tech nerds that bought before the current boom. As recently as in March 2020 one eth could be bought for $90. This is the most perfect population achievable for resistance to government interference.
In contrast, for already mentioned reasons, mining is already controlled by big companies.
Censorship is a more realistic risk from regulations forcing miners to censor. With PoW, there's no defense except moving to a different country - because it's impossible to hide warehouses with industrial power connections, not to mention power plants. In reality, as long as regulations don't kill the network outright ('just' an OFAC blacklist, freezing coins from ransomware, theft, other criminal acts) - and these regulations are enforced in places with cheapest available electricity - miners would comply. Once those miners attained a sufficient majority, laws would force them to orphan non-compliant blocks.
PoS staking is trivial to hide - it's just a computer with a network connection. The whole network could be made anonymous by utilizing onion or garlic routing. Regulations would be impossible to enforce.
In conclusion, PoS is superior to PoW in every single fundamental factor mentioned in the article, especially censorship resistance.
I don't agree that complexity itself is a factor for or against: complex things are certainly harder to build, but once finished and proven to work that becomes irrelevant - only the functional differences remain.
There's only one real advantage of PoW over PoS: PoW is a coin distribution method, while PoS isn't. However, 90% of all bitcoins are already mined, so that argument is already very weak and becomes weaker with each day.
At the end, these discussions are for nerds - the average person may just get the impression that 'person X is saying PoS is less secure' and be wary for a while, but given enough time of working well, ~nobody is going to care about the anti-PoS position.
Additionally, belief in PoW is economically self-punishing: mining costs billions per year, ensuring that PoS supporters, as a group, are getting relatively wealthier to PoW supporters. A cost difference of double-digit billions per year is very hard to sustain over longer periods. It also means that a PoS network can continue working with next to zero buyers - while below some level PoW becomes hopelessly insecure (see eg. 51% attacks on ETC) - making PoS much more resilient to economic shocks.