Proof of Work vs Proof of Stake – What to Expect from Ethereum’s Protocol Shift
It’s early days, but Ethereum, essentially a decentralised, blockchain-based, world computer, is changing its consensus protocol from Proof of Work to Proof of Stake. Julie Ros speaks with a few crypto traders about what the key differences are between the protocols and what they think of Ethereum’s bold move.
Ethereum, the blockchain-based network that was proposed in 2013 and released in mid-2015 to provide a virtual “world computer” as the base layer for decentralised apps (DApps), has begun steps to move the methodology for confirming transactions from Proof of Work (PoW), whereby miners compete to unlock and upload blocks to the Ethereum blockchain, to a Proof of Stake (PoS) methodology, which establishes validators that stake an investment to participate.
Ethereum overlaid a PoS mechanism over the existing PoW methodology, with a view to eventually moving fully to a PoS model with a hard fork of the Ethereum blockchain. The PoS protocol, Casper the Friendly Finality Gadget (FFG), was introduced by co-founder Vitalik Buterin in late 2017, who stated: “We introduce Casper, a Proof of Stake-based finality system, which overlays an existing proof of work blockchain. Casper is a partial consensus mechanism combining proof of stake algorithm research and Byzantine fault tolerant consensus theory. We introduce our system, prove some desirable features, and show defenses against long range revisions and catastrophic crashes. The Casper overlay provides almost any proof of work chain with additional protections against block reversions.”
Meanwhile, Vlad Zamfir, who joined Ethereum in 2014 and is the lead researcher working on a Correct-by-Construction Casper (CBC) protocol, says in a blog post that the introduction of validators in the proof of stake model introduces a layer of security against bad actors by requiring validators to place “a deposit to play, and if you play nice you make a small return on your deposit, but if you play mean you lose your deposit”.
In a recent tweet storm about the state of Casper research and development, Buterin explained the two methodologies: “Note that Casper CBC and Casper FFG are *both* ‘overlays’ that need to be applied on top of some existing fork choice rule, though the abstractions work in different ways. In simplest terms, in Casper CBC the finality overlay adapts to the fork choice rule, whereas in Casper FFG the fork choice rule adapts to the finality overlay. Casper FFG (and CBC) both require the *entire* validator set to vote in every ‘epoch’ to finalise blocks, meaning there would be tens to hundreds of signatures coming in every second.”
Although Ethereum has begun to roll out the PoS protocol, a timeline for a hard fork and full PoS methodology is some way off, as Buterin concluded in his tweet storm: “What’s left now? On the FFG side, formal proofs, refinements to the specification, and ongoing progress on implementation (already started by >=3 teams!), with an eye to safe and speedy deployment. On the CBC side, much of the same. Onward and upward!”, which is probably about as close to estimates as the dev teams will get at the moment, according to a spokesperson at Ethereum.
In a blog by cryptocurrency derivatives trading platform, Bitcoin Mercantile Exchange (BitMEX) released in April, the “consensus by bet” methodology of PoS directly ties the consensus agents to an investment in the coin, theoretically aligning interests between investors and consensus agents. BitMEX notes that PoS is currently being used as a checkpointing process every 100 blocks, which only comes into play once the PoW miners have decided on a chain – and the PoS votes are not valid until there are at least 12 miner confirmations. While this PoS checkpointing provides an extra layer of assurance over PoW, if two-thirds majority of votes cannot be achieved, the chain continues on a PoW basis. BitMEX notes that the complexity of the PoS proposal is the main downside, although the current iteration is better than previous versions.
In a full PoS proposal, notes BitMEX, blocks are produced from a pool of block producers, a random number generator is used to select whose turn it is to produce a block and then the producer is given a time window in which they can produce a valid block.
Profit & Loss spoke to some of those active in the crypto space to get their views on the merits of PoW versus PoS in the lead up to a hard fork.
If it ain’t broke, why fix it?
While both PoW and PoS have their merits, Ethereum is notable for pushing the boundaries of what the technology can accomplish. But while some express a preference for the ‘don’t break things in an irreversible system’ approach, Ethereum is lauded by others for its ability to evolve, pointing to the hard fork it already underwent (which spawned Ethereum Classic, in an effort to return pilfered funds to investors hacked from a DApp on its network, The DAO).
Zane Tackett, head of OTC sales at cryptocurrency market maker, B2C2, based in Tokyo, notes that the move to PoS is in line with Ethereum’s ethos, which has been to “move fast and break things and we’ll figure out how to solve it”, as opposed to standard-bearer Bitcoin’s “move slowly and make sure that nothing breaks” approach.
Meanwhile, another industry source believes the decision is “insane”. “This is a huge technological experiment without real, solid proof evidence that this will work. It’s a massive risk to try to pull this off when you’ve got multiple billions of dollars on the line,” the source says. “It’s one thing to do this right at the start, it’s quite another to do this, years after launch, having corporations put up money for the Enterprise Ethereum Alliance to encourage industry use of Ethereum, with no guarantee that this is going to work.” This is a major transition from POW to POS, so the step-by-step transition to proof of stake must be done to gain the confidence of the investors without confusing them, or it could destroy the faith in the crypto ecosystem,” says Sonia Goklani, WSBA Tech and Product Working Group, adding that the hybrid models that evolve could generate different types of tokens that might pose regulatory and financial model evaluation challenges.
Looking beyond pushing the envelope of technology, one of the benefits that Tackett points to is that PoS enables ‘sharding’, something that is practically impossible in proof of work. “In proof of stake, you can have different validators working on shards of the blockchain at the same time,” he notes.
In Bitcoin’s Blockchain for example, all nodes store and process the components of all transactions, but with sharding, only a randomly selected subset of nodes validate and store information relevant to a transaction. In this way, nodes on the blockchain are not required to keep records of every transaction, which in theory should speed transaction times, as well as reduce the energy required to produce blocks and maintain the network.
While Ethereum’s current, hybrid PoS/PoW model may not be speeding up transaction times yet, since the PoS part of the equation is serving primarily as a check point of blocks at this stage, Layer 2 solutions could eventually increase the transaction throughput of Ethereum on a massive scale, sources say.
“Transaction speeds should increase at the cost of security and decentralisation of wealth,” says Michael Moro, CEO of Genesis Trading. “The trilemma between security, scalability, and decentralisation remains – increasing scalability through PoS trades off the security of PoW and could introduce some wealth decentralisation problems.”
Meanwhile, BitMEX Research tells Profit & Loss that: “Ethereum’s block time is already just 14 seconds, which is pretty fast. How to measure transaction speed is difficult though. Using both PoW and PoS systems, one gains a greater assurance of the transaction over time, and one is never really totally sure the transaction is irreversible. There is a concept in PoS systems of ‘finality’, which means the transaction is 100% confirmed, however that process is likely to take far longer than 14 seconds. In general, I would say PoS systems are likely to have a less steep curve when comparing time with assurance over the transaction.”
As PoS evolves, some sources expect that it could incentivise more people to help Ethereum grow, as it will allow much smaller players to become part of the network since the barrier to entry for becoming a participant would be lower. With BTC, you have to have a massive mining operation, while in ETH you could potentially only need to stake a small amount, like 10eth. “Currently, anyone that isn’t a large-scale, industrial miner has been pushed out of mining, but if everything goes as designed, a lower barrier to entry could prove to be one of the benefits,” says Tackett.
Moro adds: “If it works and can create more of a decentralised consensus mechanism in ether than the current model, I do think it’s additive to the cause, because if it creates better stability, investors may get behind it.”
Currently, there’s an advantage for miners residing in places with cheaper energy to quickly unlock transactions. Reducing the global demand for energy by miners has a two-fold benefit.
One is obviously a reduction on the drain of natural resources, but the other is that more people will be willing to participate in the network if the high power requirement can be reduced. “In proof of work, the people with access to the cheapest electricity often have the biggest advantage. If you remove that need for cheap electricity, it makes the whole system a bit more fair,” says Tackett.
“Proof of work is inherently an energy intensive process, while proof of stake is not. Proof of work is essentially a race to use as much energy as possible, the more energy one uses, the greater assurance there is over transactions. The high cost of buying stake is not supposed to reduce the number of validators. Firstly, anyone can validate transactions, whether they are stakers or not. Secondly, the number of stakers should be larger than the number of miners if the system works well,” says BitMEX Research.
With cheaper electricity and other improvements being developed to make it easier for more people to get involved, does an unlimited number of validators, as Ethereum is entertaining, raise questions around governance? Ethereum posited that it could potentially cut interest rates if there are too many validators, and if there are too few, it could increase incentives. But wouldn’t this type of central control over elements such as interest rates be anathema to the concept of the decentralised cryptocurrency ethos?
“At the end of the day, I think it’s still going to be controlled by market forces. If there are enough people that think the interest rate is attractive enough for them to tie up their funds and stake their coins, then we’ll see an influx, and if there’s not, there won’t be. The fact that all of this happens in pretty much an open market solves a lot of these problems in theory – we’ll have to wait and see if it works in practice,” BitMEX tells Profit & Loss. “My issue with this is the high level of complexity, such that it’s likely that incentives may not work out well or there could be bugs.”
“The fact that Ethereum doesn’t have a rigid interest, inflation, or any other monetary procedures set in stone is of course a massive risk around central governance. Part of what makes BTC so attractive is that 21 million figure and elliptic curve – block reward halving is indisputable at this point, if anyone, even Core, tried to fork and change that, it would simply never work. With Ethereum, the Core devs have this discretion and the current holders are trusting them to do something sensible around supply policy,” adds Moro.
Regardless of how it plays out, the devs behind Ethereum do have control, whereas in Bitcoin, the miners make decisions by consensus.
So what happens to the natural competition factor that PoW engenders as miners compete to unlock and upload blocks to the chain? Will sharding change this? B2C2’s Tackett notes that since there is a hybrid model operating, mining is still being done by proof of work, so it will be difficult to know before sharding begins with full implementation of PoS.
There are some complaints that mining generally has centralised into the hands of a half-dozen or so large miners and that proof of stake can help decentralise this control, but since a stake is roughly the ether equivalent of $1 million, won’t this keep a core few at the heart of the new system?
According to Moro: “The wealthy may get wealthier, and the question of fairness is different than the question of equity. A larger stake means a larger return (there’s no reliance on pools), and a larger voice in any on-chain governance. You can buy votes in this world. “But,” he adds, “the arms race on the Ethereum network is different than the one on the Bitcoin network for many reasons. The cost of a stake is probably not the dominant factor,” says Moro.
“Mining centralisation in bitcoin has been way overblown,” believes Tackett. “People mistake mining pools as controlling the hash rate, but the miners control the hash rate and they direct it to the pool. I do think there is an issue around the whole model of proof of stake in terms of centralisation with the rich getting richer, but Vitalik has said that the plan is that once sharding is enabled, they can go down to as low as 10 ether for staking, which could lead to more decentralisation as more people can partake in securing the network.”
“Proof of stake can solve some of the problems with proof of work miner centralisation, but it can also introduce new centralising forces,” BitMEX Research explains. “Ethereum is trying to introduce reward systems that encourage stakers to spread the stake – for example, by punishing larger stakers a higher proportion of their stake for mistakes than smaller stakers.” Miners are currently paid in native currency for successfully solving an equation (hashing) and having their block transaction uploaded to the chain. As the payment mechanism for mining essentially shifts to a transaction fee model, could this consolidate the setting of fee levels in the hands of those that can afford to stake?
“I think transaction fees will naturally reach equilibrium due to market forces. If somebody makes their minimum so high that there’s a high demand for validating/mining lower fee transactions, there’s a strong incentive for somebody to come in and lower their minimum so they can collect those fees,” says Tackett. “I think the fact that people can come in and participate openly solves that issue.”
“Solving blocks (or finding the hash of a certain value), is a proof of work process. A pure proof of stake system may not require ‘solving blocks’, adds BitMEX. “Naturally, staking pools will form, so you can be a validator with a small stake by joining a pool, like mining pools today. You get your share of the total that the pool achieves,” says Moro. “While concentration of mining is always a concern, it’s not obvious why transaction fees might lead to centralisation more than other aspects of the protocol,” says Moro.
Since one of the most important aspects of blockchain technology is safety, is PoS safer than PoW? “Absolutely not,” says Tackett, “I think it has some very novel approaches, like the two-thirds factor regarding malicious actors as opposed to the 51% of Bitcoin; however, it’s all hot air until we see it in practice. Bitcoin’s proof of work model has been proved in theory and in practice time and time again, so from a safety perspective, I think it’s very hard to beat. I think that’s a very large reason why bitcoin is king – because everything worked out of the box. Everything it said it would do, it has done from the beginning and there haven’t been any big issues with the security model, while early in Ethereum’s history was The DAO, which caused the fork, because the system couldn’t risk one player holding that much Ethereum. That risk hasn’t really gone away. There are exchanges that could end up holding a ton of ether, so if somebody hacks one of these places, you would end up with a similar issue again, whereas with bitcoin, you don’t really have that from a safety perspective.”
BitMEX agrees: “I am not sure if proof of stake will be safer. Proof of stake cannot avoid 51% attacks. The argument is that a 51% stake attack costs more than a 51% mining attack. Actually 33% of stakers is enough to attack, as proof of stake requires 66.7% to work and move the chain forward.”
“Proof of stake is strictly less secure than proof of work in my opinion. One of the key differences between proof of work and proof of stake is the cost of attack. With proof of stake, you can damage the network and face penalty later. With proof of work, the financial cost must be borne before damaging the chain. Proof of work is a tried and tested, secure consensus mechanism. It may not be the best, but it is fundamentally more secure than proof of stake presently,” says Moro.
Since the validators are the ones with the most at stake, there is a view that they will have a natural incentive to make sure the network operates as designed. If a validator is considered a bad actor, “slashing” or forfeiting their staked coins, is the means of punishment. “Mistakes are costly in crypto. When somebody stakes, they know the coins are at risk of being slashed,” explains Tackett, and in an immutable system, there’s no do-overs…the ether is basically destroyed and the supply removed from the ecosystem.
Sitting as it does, at the heart of numerous DApps, will those entities need to adapt for the hard fork?
“Yes, I think proof of stake systems have fundamentally different dynamics to proof of work systems and therefore some DApps may need to adapt. However, many may disagree with this and many DApps have not been battle tested in proof of work systems yet,” BitMEX Research says.
“DApps are less concerned with the consensus mechanism of the base layer as they are with the Layer 2 solutions. DApps will move to their own side chains over time, putting limited trust in the developers by periodically pushing key data to the main layer such that if a side chain was compromised, all the data is easily transferrable to a new DApp contract. The base layer should have high throughput but even with PoS, it’s nowhere near what’s required to run a meaningful DApp, hence the necessity of side chain solutions, like Plasma,” says Moro.
AirSwap, a decentralised global marketplace for Ethereum tokens (ERC20), indicates the move to PoS will just be part of the technology’s evolutionary process. “Our smart contract is fairly simple and will remain intact even as Ethereum evolves,” says Don Mosites, AirSwap’s co-founder.
At the end of the day, although blockchain has been around nearly a decade now, these are still early days for this technology – so watch this space.