Proof of Stake: The Free Lunch
In this article, I’m going to delve into the topic of Proof-of-Stake on Ethereum and explore whether or not it is a viable alternative to Proof-of-Work, the consensus mechanism used in Bitcoin. As it turns out, Ethereum is broken. But how is it that cryptos such as Ethereum so consistently convince people that their product has substance and value? Even in the case of obvious scams? There’s a lot to it, but I think it boils down to one underlying issue: people consistently forget—or wilfully ignore—this fundamental truth to which all things adhere: there ain’t no such thing as a free lunch. This principle dictates economics, and it’s essentially just an acknowledgement of the fact that value doesn’t just appear from thin air; everything valuable has a cost of production. Crops don’t harvest themselves, and if someone is nice enough to spend his time and energy harvesting crops, he’s not going to give them away for free. That would mean using his energy and getting none in return, so in a way, the principle “there ain’t no such thing as a free lunch” is sort of a knock-on effect of the law of conservation of energy (energy cannot be created nor destroyed), so we can be pretty sure that it applies at all times, in all scenarios. Still, people are often wilfully blind to this principle because they want to believe in a “free lunch” so much. So, instead of spelling it out every time, I’ll just be referring to it as “our/the axiom” for the rest of this article.
Axiom: there ain’t no such thing as a free lunch.
Many will say that the value you add to the network by staking (for which you are paid in staking rewards) is a product of your participation in the consensus protocol by running near-costless validation scripts. If this were true, the network would ask for as many validators as possible since security would be a product of validators; more validators = more security. However, validator status is a $52,037 walled garden at the time of writing. As such, any value you produce by staking is not a result of running validation scripts; it’s the result of “putting your money where your mouth is” to increase the confidence that other nodes have in your intent to run the validation scripts properly. This is not inherently a bad thing, just something to note since it highlights the fundamental difference between PoW and PoS: PoW is pay-as-you-vote, PoS is pause-liquidity-as-you-vote (since you’re not actually parting with ETH, just temporarily pausing your ability to use it). It also highlights the dependent relationship between staked ETH and security: the staked ETH is the security.
Stakers are capable of making lucrative the practice of running some lightweight scripts, which incur no real cost or expenditure to run. This is true since the funds are only ‘at risk’ should the staker decide to act maliciously; assuming they plan to stay within the rules, they can retrieve their stake after profit has been made and continue to use that ETH freely. They could even choose to take funds in and out of staking contracts at will. When a user withdraws his stake, his public key is forbidden from ever participating in validation again, but he may simply send funds to a different address for which he also owns the private key, and stake the same funds again; this essentially reduces the cost of staking to almost zero, but there ain’t no such thing as a free lunch, so what is the cost of staking?
Should you decide to end your “dynasty” (stop staking), there is a four-month withdrawal period in which funds remain locked, and any violation of “the Casper commandments” (Ethereum’s rules for validator nodes) causes portions of their stake to be slashed (up to 100%, depending on the number of dishonest validators in the network at time of the violation). But remember, so long as you don’t plan on single-handedly deceiving the entire Ethereum network by breaking the rules, this risk only applies to you in so far as general protocol risk (accidental slashing/loss of an honest validator’s funds due to a protocol bug) and, of course, stakers don’t get paid to be the subject of protocol risk, as the rest of the network is also exposed, yet unpaid, so this cannot be factored into staking rewards (not to mention, risk is only desirable under the conditions of potential reward). The cost of staking is this four-month liquidity penalty, but who is this cost paid to?
Considering the fact that you’re trading current liquidity for what is essentially interest payments, it’s natural to note the resemblance to lending/debt markets. However, when you lend, somebody else can utilise the capital you give them today. Since capital today is more valuable than capital tomorrow, they pay you the sum of that discrepancy plus reimbursement for counterparty risk. In return, the borrower gets to use that money now, rather than later. In the case of staking, nobody gets to use the funds you stake, so who benefits from your funds being locked away, and how do they benefit? Since you ‘lend’ to no particular individual or group, the only remaining answer is that all network participants benefit… but how? We know it must benefit everyone since it benefits no particular individual or group (and it must benefit someone; otherwise, the free-lunch principle would be violated). So, we need to find ways it could provide value that’s distributed to all network participants. Locking up funds reduces the effective supply of ETH, which is a deflationary pressure, pumping up its price/nominal value. This is one way all network participants could benefit from your stake being locked up. So, since the value of your money is a function of time—call it ‘temporal value’—you are taking a chunk of temporal value out of your ETH in exchange for additional ETH (which also remains locked with originally staked ETH). This is, in essence, just shifting value around the network and across time.
If you zoom out and look at the network as a whole, it is a self-contained, virtualised environment in which value distribution is just bouncing around, nothing going in, nothing coming out. Thus, the network in total becomes an atomised singular entity—more like a product than a network—and since no value is coming in or going out, there is no way to add real value to the network; all that can change is the price of the token—which is now purely determined by psychological forces of perception and faith. Let me reiterate: so long as Ethereum is Proof-of-Stake, its foundation is, and always will be, built with bricks made of nothing but faith. With Proof-of-Work, the value of the chain is not self-contained; it’s connected to the most fundamental physical law: energy cannot be created nor destroyed… the same law that our axiom “there ain’t no such thing as a free lunch” results from. For instance, you can add value to the Bitcoin network by putting more energy into it via mining, making it harder to exploit, thus making it more valuable. But if you stake ETH, you’re just temporarily redistributing its price to the rest of the ETH supply by pausing its contribution to the effective supply; no value is being added to the chain, only redistributed amongst the network.
Can value be added to Ethereum at all?
So far, I have been saying, “no value can be added to Ethereum” in the context of a discussion about the consensus mechanism proof-of-stake. Some people will argue that value can still be added to Ethereum by software development. So let’s take a look at software development and ask, “Can value be added to Ethereum via software development?”. First, we have to define exactly what Ethereum is and pinpoint its value proposition. To do that, we will first define what it is not. It’s not a cloud computing service; they already exist and are much more efficient/effective than Ethereum’s cloud computing capacities. It’s not an app platform either; the Ethereum team themselves will tell you that how it separates itself from app platforms is by being a “dapp” platform: decentralised applications. So, the value proposition is that it facilitates the development of decentralised applications. So it’s a facility for decentralisation, and its ‘decentralisation’ is strictly achieved by its consensus mechanism. Putting two and two together, we find that Ethereum’s entire value proposition—and the value proposition of every project built upon it—is its consensus mechanism. Just to emphasise, every project built upon Ethereum takes a service that already exists and tries to create a decentralised version of it. To do this, they have to sacrifice some efficiency and efficacy found in centralised alternatives; this makes ‘decentralisation’ their sole competitive advantage. Without this, they are all worthless. Now, what happens when you take Ethereum’s consensus mechanism and destroy it? That’s what they did, and they replaced it with Proof-of-Stake. Thus, the value proposition of Ethereum and every project ever built upon it has completely changed and must be subject to re-evaluation. It may have the same name, but it is now a completely different product, at every level of analysis.
Exposing security to price & volatility.
Now, the total staked Ethereum is about 13.8M. To gain foremost control authority over the Ethereum chain (>60%) would cost ~8.4M ETH… to cause serious problems would cost half that. There is nothing that could be done to stop it besides every single holder of ETH agreeing not to sell to ETH ever again in case they are selling to the attacker, at which point the entire network draws to a halt regardless since ETH can’t change hands. Because governance is granted on a stake-weighted basis and relies on economic retaliation upon stakers’ funds to incentivise honesty, the network’s security is limited to the market cap and directly pegged to the price of its token. If the value of the ETH goes down 50%, validators now have half as many reasons to act honestly, and autocratic control over Ethereum just went on sale for half-price. Since stake is the sole source of security, PoS pegs security with the token price on a one-to-one basis, posing an additional ‘death spiral’ risk where falling prices cause stake to be withdrawn and sold, meaning less security, and less security lowers confidence in the chain’s token, further lowering the price and causing a feedback loop. You can see how allowing the resilience of the chain to be directly determined by its perceived utility/price—which is essentially just public opinion—might be a bad idea. Bitcoin does not encounter this problem because it is pay-as-you-vote and get-paid-as-you-vote, which means earnings from securing the network are instantly liquid and are only earned if honesty has already been demonstrated. If the price of Bitcoin drops, the cost of dishonesty remains unchanged since it’s priced in energy, not BTC. Conversely, the cost of dishonesty on Ethereum is priced in ETH—hence my calling it an “atomised singular entity”. This proves to have some serious implications. Staked ETH is locked away, meaning there’s no escape for stakers if the price of ETH drops; they just have to watch their reason to stay honest slowly shrink. Even if we try to “steel man” the argument for Proof-of-Stake, and assume the price of ETH always increases, there’s another problem: there is no way of being certain that 60% of validators aren’t controlled by one single entity. Because Ethereum is pseudonymous, a single entity could slowly accumulate total authority over the project in a Sybil-style attack—one of the primary issues Bitcoin fixed—where it’s impossible to determine if votes are coming from many people or one person in control of many validators that hold a cumulative >60% of staked ETH. The only way of knowing for sure is if that person divulges information about their staked holdings, which individuals will likely not, but sometimes companies actually do. Here’s what we know about the distribution of authority across the Ethereum network as of recently:
*Due to stake-weighted governance, this is a chart of Ethereum’s control authority distribution*
Note the 23% “unlabelled” group of validators. As of right now, it’s not possible for there to be some unknown supermajority owner of >60% stake because >30% is already controlled by one entity, and the second-largest known entity has 15%. Still, assuming this centralisation decreases, which is the best-case scenario, the threat of the Sybil-style attack becomes very real since we don’t know who owns a given ETH account. One person could appear as lots of accounts in which the sum of staked balances amount to >60%, unknowingly controlling the entire network, and could even begin slashing the stake from honest validator’s accounts—and there’s no way of verifying whether or not this centralisation is ever occurring. So, it’s a facility for decentralisation with no way of verifying the decentralisation of the facility. Hmm. The magnitude and quantity of these risks essentially renders mute any claims about the supposed ‘decentralisation’ offerings of Ethereum, which is the one thing that must not fail if the project, and all its child projects, are to succeed, for without this key factor, Ethereum is nothing but a slow, inefficient database, and all projects built on it are nothing but uncompetitive versions of pre-existing digital products and services, with none of the associated benefits of decentralisation. To be more curt, it’s a pile of trash with no redeeming qualities. From this conclusion, we can circle back to our original question to ask, “Can value be added to Ethereum via software development?” and it becomes clear that it cannot because the sole reason you would want to build atop Ethereum turns out to be untrue, so any development done thereon out is just role play. It’d be like renovating the sleeping quarters on the Titanic and saying it’s now a better boat. It’s not. It’s just a shipwreck with beds.
So, where do the staking rewards come from?
Considering our axiom, the value that stakers earn is not coming from the production of actual value (if no value can be added to the network, staking cannot add value to the network) but rather some form of lag between real and nominal/perceived value production (price). “Real” means the actual value produced, and “nominal” means the perceived value according to the market, which is subject to fallacy and error; nominal value = price. Such an example of this is the perceived value of newly printed money when it is first used, and the market has not yet adjusted the price of all fungible currency units, i.e., the lag between currency printing and the decrease in purchasing power of that currency. This is why interest rates which don’t consider inflation is called “nominal rates”, and those that do are called “real rates”. This leads me nicely to my next point.
People criticise bitcoin by saying it can be stopped by governments. With proof of stake, if governments want to take over Ethereum, they can simply do so by trading something with a near-zero marginal cost of production (fiat currency) to purchase a majority share of Ethereum’s stake. From this point onwards, they will have ultimate control over the state of the chain and direction of the project with negligible maintenance cost. They could simply sabotage it if they wanted, or worse. So long as they don’t sell their ETH to the rest of the network, they can do little to nothing to stop them since security is derived from ‘stake’ and the government already has it all. They don’t own it for profit, so they’re not selling it back to honest validators, no matter how high they bid.
There’s a lot wrong with Proof-of-Stake, but to sum it up, it’s a tautological catastrophe that uses circular logic to present a facade of “value for nothing” by over-simplifying the multi-dimensional problem of decentralised consensus that Bitcoin and its Proof-of-Work-based consensus mechanism solved. There ain’t no such thing as a free lunch, and decentralised consensus is one hell of a lunch, so we won’t get it for free.
Casper FFG & transaction finality: https://arxiv.org/abs/1710.09437
Ethereum’s Proof-of-Stake: https://ethereum.org/en/developers/docs/consensus-mechanisms/pos/