Ethereum’s London Upgrade
Jul 20, 2021
Ethereum’s long awaited “London” hard fork is now scheduled for August 4. The upgrade includes five EIPs – adjusting the fee market on Ethereum (EIP-1559 and EIP-3198) and effectively removing “gas tokens” (EIP-3529), among other auxiliary/preparatory changes (EIP-3554 and EIP-3541). While some of these upgrades are purely technical, especially EIP-1559 has been widely discussed in the community and will have an effect on the future net issuance of ETH by the protocol.
EIP-1559: Changing the ETH fee market
EIP-1559 was created in 2019 as inefficiencies in the current fee model had become obvious. With every transaction on Ethereum, the user includes a gas limit (depending on the specific transaction, a “normal” ETH send uses 21’000 gas) and a gas price per unit of gas, typically denominated in GWei (or one billionth of an ETH). However, gas prices are even more volatile than the USD price of cryptocurrencies, and therefore, transactions can get stuck during periods of high demand if the gas price is set too low. Users and wallets have to guess what gas price to pay based on historical prices, which leads to bidding wars and regularly overpaying for transactions.
EIP-1559 changes this dynamic: Instead, a BaseFee per gas will be set by the protocol each block, depending on how full the last blocks were. The user simply pays the BaseFee, plus a “tip” for the miners/validators for priority transactions, to ensure that the transaction will quickly be included in the chain. If the blocks are full for an extended period of time (>5 min), then the BaseFee will quickly rise to account for the increased block space demand.
Illustration 1: After 40 full blocks (ca. 8.5 minutes), the BaseFee would sit at ca. 100 GWei and continue to increase quickly unless demand for block space softens.
The BaseFee will not go to miners, but instead be burned – this is done to prevent some attacks on the network and economic abstraction (ETH will always be needed to transact on Ethereum), but it also changes the supply dynamics of ETH. Burning part of the transaction fees paid on Ethereum might add deflationary pressure on the total supply if demand remains high.
Illustration 2: Cumulative transaction fees paid on Ethereum to miners since January 2018 currently stand at ca. 3.8 million ETH – part of future fees would be burned post-EIP1559.
At 3.8 million ETH in cumulative transaction fees paid over the past three years, the amount of ETH that would have been burned if EIP-1559 were already live during that period is not negligible for the overall supply (which stands at 116.8 million ETH).
Illustration 3: During the periods of extremely high activity in the “DeFi summer” of 2020 and in early 2021, the annualized supply inflation would have been remarkably lower had EIP-1559 already been active. 30-Day simple moving average of annualized supply inflation shown, 50% fee burn assumed.
Currently, the annualized supply inflation stands at around 4.2%. In a scenario with EIP-1559 and half of the total fees being burned (the rest going as “tips” to miners for high priority transactions), this number would change to 3.8% at the moment with a relatively non-congested chain, or would have dropped as low as 2% during the heights of “DeFi summer” in 2020 (a temporary “halving”). As such, it will be interesting to see how these dynamics play out once live on the mainnet.
EIP-1559 is also already live on Filecoin – alleviating some of the concerns around its application in practice.
EIP-3529: Removing gas refunds
EIP-3529 will remove or reduce the gas refunds for certain operations (e.g. “selfdestruct”ing a smart contract). Previously, these operations enabled the creation of “gas tokens” – tokens that can be created to store gas when gas prices are low, and be redeemed when gas prices are high. These tokens basically wrote worthless code to the blockchain in a smart contract, and then selfdestructed (refunding gas) when needed, cluttering the chain. After the creation of the proposal in April 2021 by Vitalik Buterin and Martin Swende, gas prices almost immediately dropped to a two-month low, as people stopped minting gas tokens in anticipation of this change.
EIP-3554: Delaying the “Difficulty Bomb”
This is a relatively simple change updating some parameters in the difficulty setting for Ethereum mining. Ethereum core developers have implemented a “difficulty bomb”, which upon triggering starts to exponentially increase the difficulty for mining. If left running without manually defusing or delaying the difficulty bomb, at some point Ethereum would reach the “Ice Age”, where no mining is possible anymore. The purpose of this is to bring some urgency to protocol upgrades and eventually switch to proof-of-stake.
Illustration 4: In the past, the “difficulty bomb” started going off three times before it was delayed further with a hard fork. This is visualized through the increasing time between blocks.
Conclusion
The London upgrade for Ethereum is certainly one of the most anticipated hard forks. It will bring various improvements to the user experience of Ethereum, but also establish a link between protocol activity and supply of the native currency.