A Long Chain of Forks

Software upgrades in distributed systems like blockchain protocols can be more complex than upgrading your browser at. The core question is what happens to those who do not follow the upgrade and stay on the old version. If you do not upgrade your web browser for 5 years, it may start having problems with newer websites, but that’s about it. As the underlying TCP/IP and HTTP protocols powering the internet have not changed, everybody continues to be on the “same” internet, just with older or newer browsers. In blockchain lingo, these are “soft forks” (Table 1). If the underlying protocol changes as well, then the situation may change. Then upgraded and non-upgraded clients may find themselves on different “internets” that are no compatible with each other anymore – which is called “hard forking” in the blockchain world. While the first case is not a big issue, the latter is something that requires a high degree of consensus to protect the chain from permanently splitting in two. At the same time, it is the final option if the developer community cannot agree on a common direction for the network: take a copy and go your own way.

With these complexities in mind, the long chain of forks that Ethereum, the platform that pioneered smart contracts and decentralized applications, underwent since its first official release in 2015, is impressive. Most of the hard forks spanning six years across four development phases were planned and uncontentious. Illustration 1 gives an overview including the key changes of each hard fork (EIP are “Ethereum Improvement Proposals”).

While the Ethereum community overall managed to avoid long, paralyzing disputes akin to Bitcoin’s Blocksize Wars, there was fork, that stirred a lot of debate and discontent. It was when the leading Ethereum developers initiated a hard fork to rectify the “DAO Hack”, the siphoning of $60 million of funds invested in the first decentralized autonomous organization on Ethereum, The DAO. Since that contentious decision, there exists Ethereum Classic that kept the original Ethereum consensus rules, while today’s Ethereum chain recompensated people who lost Ether to the DAO Hack.

Now this long chain of hard forks on Ethereum 1 is slowly drawing to a close in 2022, as the community is working on the final phase towards The Merge, the merging of the proof-of-work Ethereum 1 chain into the proof-of-stake Ethereum 2.0 Beacon chain.

Therefore, the focus of many EIP changes during the Serenity phase has been to improve Ethereum’s scalability, be it through repricing of gas cost for certain operations (opcodes), improvements to make Layer-2 solutions possible or more performant, and t prepare the switch to proof-of-stake.

A rollup-centric scalability roadmap

The problem Ethereum aims to tackle with the Ethereum 2.0 upgrade, is plaguing all blockchain-based platforms: to be synchronized and get “in consensus”, distributed networks require to send information forth and back between nodes. This coordination requires two things: first, bandwidth – hence debates about different block size designs, like 1MB in Bitcoin and a “gas limit” for blocks in Ethereum. Second, coordination requires time for the information to propagate across the network – hence debates about different block time designs, like 10min in Bitcoin and ca. 13sec in Ethereum. There exists no best solution. Different projects optimize for different outcomes, but all are bound by the Blockchain Trilemma, the blockchain reformulation of the CAP theorem from Computer Science developed in the 80s. It states that only two out of three goals can be achieved simultaneously in blockchain systems: decentralization, security, or scalability.

As blockchain proponents have a hard time sacrificing decentralization or security, much research and development over the last years went into finding solutions to scalability that would minimize the effects on the former two properties. In October 2020, Vitalik Buterin, founder and master-mind of the Ethereum project, concluded that “the Ethereum ecosystem is likely to be all-in on rollups (plus some plasma and channels) as a scaling strategy for the near and mid-term future.”

When researching blockchain scalability, one distinguishes between on-chain (Layer-1) approaches that modify the consensus protocol and off-chain (Layer-2) approaches that require no modifications on the base layer but may use base layer security to operate (Illustration 2).

Rollups are Layer-2 scaling techniques that bundle (“roll up”) many transactions together and only send a proof of that bundle to the Layer-1, thus releasing the load on Layer-1. Depending on the proof logic, one distinguishes between optimistic rollups and zero knowledge rollups. Optimistic rollups are “optimistic” that the transactions are correct and only use fraud proofs when needed. Zk-Rollups on the other side offer validity proofs for every transaction bundle.

Rollups continue to be an area of active research and experiments and projects are active across the entire space. We expect the next year to provide more clarity in which direction the Ethereum project will head on the second layer. But first, the project has an immediate challenge on Layer-1.

Mastering the Merge amidst competing chains

The Ethereum community is facing two challenges that have been growing over 2021 and can be expected to intensify in 2022. The first is in-house: the switch from Proof-of-Work to Proof-of-Stake. The Merge aims to transform the Ethereum 1 chain into an execution shard and then connect it to the Beacon chain, which is already live and running since 1 December 2020 (deposit contract live since 4 November 2020). The second is coming from the markets: several other smart contract platforms have gone live and challenge Ethereum on its weak spot, transaction fees and transaction throughput.

The Merge. The ambitious goal is to swap the core consensus mechanism from Proof-of-Work to Proof-of-Stake while introducing sharding – all without disrupting operations. The word “merge” is a bit of a misnomer in our view. Instead of merging one chain into another as the word suggests, the original chain and is split up. The different components of Ethereum 1 are dissected into a consensus chain (Beacon), an execution chain (former Eth1), and multiple data shards. Therefore, one needs to decide where storing of accounts, transactions, blocks, etc. happens as well as where and how computing of finality, consensus, fork choice, etc. happens (Illustration 3).

Developers expect several improvements: (1) higher security as slashing discourages chain reorgs and thus push “block finality”; (2) PoS removes the high energy consumption of PoW (by up to 99% as estimated); (3) setting the stage to introduce data shards later to increase the available blockspace; (4) Priority fees will go to validator controlled address on Execution layer instead of to miners, making ETH liquid again and allowing validators to withdraw stakes; (5) ETH supply issuance will drop from currently ca. 3.5% to ca. 0%.

As is obvious, the migration is a very complex and thus risky undertaking. Aside from the inherent design and software complexity of the Merge, the Ethereum chain holds tokens and manages smart contracts in excess of half a trillion USD. Thousands of users, developers, and businesses around the world rely on the functioning of Ethereum 24/7. Pulling this migration off without disrupting the functioning of the Ethereum network is a nerve-wrecking process that according, to the latest information, is expected to be done in the middle of 2022. We will see whether the developer community will be successful and on time in summer 2022.

Competition. The second challenge is the fast and vocal competition in the market segment of smart contract platforms. The competition can be grouped in projects that are compatible to Ethereum and its virtual machine (EVM) and those that are not. The EVM is the unit that executes smart contract code, written for example in the Solidity programming language. A platform that is EVM-compatible offers developers the advantage to not have to rewrite smart contracts again using a different language and therefore not undergo new code audits. In addition, the existing ecosystem of decentralized apps and users can be directly targeted. The market entry barriers are significantly reduced in this scenario. On the other side, platforms that decide against this compatibility feature, have all freedoms in implementing their vision for an improved smart contract platform: consensus protocol, fee mechanism, and other technical design decisions (Illustration 4).

Inflation and utility tokens. Ethereum still looks good with these metrics: the platform with the highest market cap is also the one with the lowest inflation rate (under 1%), while some of the competitors exhibit 2-digit inflation rates that make fiat currencies look good. Does inflation matter at all? In contrast to a pure monetary coin like Bitcoin, which has an ever-decreasing inflation rates built-in, the question which inflation rate is adequate for a smart contract chain is less obvious

Low inflation may increase the token price (in fiat terms) over time, which is what investors seek. High inflation rates on the other side, may just mean that many tokens are created that users and developers can take to run smart contracts cheaply. The less a “gas” token is used as money, the better it is suited to power smart contracts and vice versa (aka utility token). It will be interesting to see how the differing interests between developers, users and investors will play out in the longer run – especially on the Ethereum platform that changes the economics quite radically.

Emergence of “ultrasound money”

Two developments impact the circulating supply of ETH on the path to Ethereum 2.0. One is caused by the switch to Proof-of-Stake (PoS), the other one was deliberately introduced to change the fee dynamics.

Proof-of-Stake is powered by validators who stake tokens (a bit like in poker) instead of spending energy for producing the next block and keeping the network secure. It’s a bit like poker: You put on a stake and if you lose (i.e., do not adhere to the consensus rules), your stake is slashed. How large will the effect be? Due to the various changes, the future monetary policy aka token supply for Ether is very hard to anticipate without modeling it.

No mining takes place in Proof-of-Stake. Instead, special nodes called validators, coordinate to process transactions and block creation by locking a minimum amount of Ether. This stake can be foregone if they behave malicious and manipulate the block creation process to their advantage. The protection lies in slashing of their stake and not in investing energy to create blocks.

What are the economic effects of staking on the supply? Most obvious, the circulating supply is reduced by the amount staked.

From December 2020 to November 2021, the amount of Ether staked for Proof-of-Stake on the Beacon Chain grew from 656 160 ETH to 8 391 388 ETH (+1280%), which permanently removes ca. 7% of the total circulating supply of ETH.

The question of supply dynamics is further complicated by the fact that the dynamics of the (“gas”) fee market in Ethereum have been modified significantly by activating EIP-1559 in the London hard fork. Before EIP-1559, the fee market was an auction in which users bid to get their transactions included by miners and all fees went to the miners. This change introduced a new mechanism of calculating fees per transaction and per block with a particular twist. Now, transaction fees consist of two parts: a mandatory base fee and an optional miner tip. While the miner tip continues to go to the miner, the base fee is burned. Since activation, the mechanism removed a total of 1,087,615 ETH from the supply through fee burning, reducing the net issuance by 507,000 ETH (68%) in the period August-November 2021.

Combined, staking and fee burning have a dampening effect on the supply of ETH. Some argue that with these changes, Ethereum is moving into a deflationary monetary policy that is stronger than Bitcoin’s inflation reduction over time, leading to “ultrasound money” in the end.

Discussion

Given the long history of exploits in the mainly Ethereum-based DeFi space and the approx. $1.5b loss of funds in 2020 and 2021 are testament to the fact that the crypto space is technically very demanding. While the number of web3 developers may still be limited, the number of computer science experts able to reliably analyze cryptographic protocols is vanishingly small.

In that context, Ethereum is headed for a fundamental transformation that turns the system upside down. The complexities and risks of dissecting the consensus components and spreading them across multiple chains are mind-blowing. A few aspects to keep in mind:

Complexity. The migration to Ethereum 2.0 is a complex undertaking with several moving parts that change simultaneously and are not easy to anticipate in their consequences: Merging to PoS, mapping Ethereum 1 into an execution shard, complex fee market dynamics, split into execution and multiple data shards, etc. How the cryptoeconomic dynamics will play out longer term is very hard to assess even with simulation.

Economics. Tightening the ETH supply because of staking may be good for investors, but it is less beneficial for those (many more) people who want to develop, run, and use smart contracts on Ethereum because it makes much more expensive measured in USD or other fiat currencies. Cheap gas is good for driving, not for investing in the oil industry.

The new fee mechanism exacerbates this trend. This strong limitation of the supply towards “ultrasound money” is a double-edged sword for a smart contract platform whose ambition is to be a world computer rather than a world currency.

Ethereum’s primary focus has always been to be the “world computer”, executing smart contracts and powering decentralized applications. However, as the recent price developments have shown, people also like to just invest in Ether as a “currency”, not worrying about developers and smart contract users per se. The interests do not seem to be fully aligned and we will see how the changing dynamics will work out in 2022 and beyond.

Finally, burning fees that users had to first earn/buy while validators receive newly created Ether as staking rewards is vaguely reminiscent of the Cantillon effect in fiat money systems: the ones “closest” to the money printer profit most of newly created coins while those at the end of the money trail have to deal with the reduction in value the market realized in the process. It will be important to watch the evolving monetary policy of Ethereum 2.0 as it is deliberately kept flexible.

Competition. All these changes promise a bright future for Ethereum, but the biggest short-term issue that is dragging users and developers to other platforms is not immediately addressed: high to very high transaction fees to run smart contracts for DeFi, NFTs, and other booming markets. Ethereum may be on the right track but may still not win the race against time.

Proof-of-Stake. While the mainstream opinion sees Proof-of-Stake winning the energy consumption debate, the more technical debate of whether Proof-of-Stake is as secure and immutable as Proof-of-Work is not so easy to decide.

Different points of criticism have been voiced in the past and we recommend consulting the cited sources for details. (1) Voskuil argues for the need for a source of security external to the blockchain because otherwise overcoming censorship is impossible should a censor reach the majority stake because it cannot be unseated anymore. (2) BitMEX’s “Guide to PoS” from 2018 already mentions the “Nothing at Stake” problem and the long range consensus attack. “Nothing at stake” means that in the case of two simultaneous blocks, a staker can use the same tokens to stake on both blocks, thus increasing rewards without additional risk to the detriment of convergence of the network to one single next block. In other words, stakers can easily change their mind and back different (fork) chains, while in PoW that would cost a lot of energy, so miners must make a decision and stick to it. Second, the long-range consensus attack problem is the problem that attackers could get hold of an “old” private key that held a large amount of tokens in the past. Simply using it to rewrite history in their favor is easy as there is no “energy wall” preventing catching up with the main chain. The solution, setting regular checkpoints, requires nodes to be online 24/7 or trusting other nodes for syncing when getting back online. (3) Nguyen is worried about Ethereum’s resilience in dealing with worst-case scenarios and gives a summary of PoS critiques, which we lack the space to dive into here.

Altogether, a few years into the Proof-of-Stake debate, the final verdict on the security merits is still pending. More research and debate are needed in our eyes to come to a clear understanding, analysis, and decision. The fact that the two largest permissionless blockchain protocols disagree on something as fundamental as the security models of the underlying consensus protocol should be a matter of concern to the crypto community.

Conclusion

Ethereum has embarked on an ambitious journey by performing the largest transformation of a blockchain project ever attempted. The stakes are high (no pun intended) as Ethereum is by far the largest smart contract platform in the DeFi space, with the most users, developers, funding, and buy-in. The switch to PoS and the separation into dedicated chains/shards is hands down the most ambitious attempt of Layer-1 scalability and will catapult Ethereum into another league if successful. At the same time, the staggering complexity and fundamental concerns raised by some experts cast doubt on the viability of the mega project.

Time will tell. Twelve months from now, we should have a much clearer picture of where Ethereum is going, and which will be the smart contract platform of choice for the crypto community.