Conclusion

Ethereum's London hardfork introduced EIP-1559 economics to the network on the 5th of April 2021. Part of this upgrade included an elegant congestion management mechanism which allows the network to flex block sizes in response to high transaction load. We saw immediate evidence of this mechanism in action following the hardfork, and evidence that it was working as designed with both large and small blocks witnessed. The EIP-1559 upgrade is a defining moment in Ethereum's history, one which will shape the economics and useability of the system well into the future.

Addendum - Burn, burn, burn 🔥🔥🔥

A quick addition - this wouldn't be an EIP-1559 dashboard if we didn't show something about fees being burned. The graph below shows the fees by block in ETH for the 8 hours after London. Let it burn.🔥

Distribution of Block Size

With the average block size hovering around 15M gwei and the max being close to 30M, it's worth having a look at the distribution. The graph below shows the block size distribution in the 8 hours after the London hardfork. We see plenty of blocks at 25M-30M gwei, but also a large number of smaller blocks as well. It will be fascinating to see how the impact of EIP-1559 on the Ethereum network evolves over a longer time period.

Impact on Average Block Size

While we saw the maximum block size increase, we also saw the average increase too. The degree of magnitude of this was much smaller however, as the fee increase mechanism above prevented sustained high increases in block size. The graph below shows the average block size per hour around the network fork time. There were even a couple of hours where the block size was lower than 15m:

Block Size - Why does it matter?

The Ethereum network, in its current form, is limited in transaction throughput by two key parameters:

• How quickly blocks are generated - the block time. This is typically 12-13 seconds.
• How many transactions are in each block. This is measured by the gas used by the transactions in the block, as different transaction types can use different quantities of gas depending on their complexity.

To expand network throughput, you can change one or the other of these parameters. Changing block time requires more powerful hardware and more bandwidth to keep up, which conflicts with the goal of keeping Ethereum distributed by being able to run full nodes on commodity hardware. Changing the block gas limit, however, is something can and does happen in Ethereum.

The block size is a tradeoff - the bigger the block, the higher the transaction throughput. That's the upside. The downside is that bigger blocks take longer to propagate through the network, meaning a higher risk of stale blocks known as uncles. Uncles occur when two miners solve the same block at roughly the same time, but one makes it through the network faster than the other one. The winning miner gets the full block reward (2 ether + tx fees prior to EIP-1559), and the losing miner gets a consolation prize of 1.75 ether. Miners currently set the block gas limit based on the uncle risk vs opportunity for higher revenue from tx fees. This happens amongst many miners, and the resulting game outcome is a fairly robust equilbrium point for the network.

As an example of this process happened around the Berlin hardfork on 15th April 2021. This hardfork included EIP-2929 which made for some efficiencies in the network. As a result of these efficiencies, the miners increased the gas limit from 12.5m per block to 15m per block. This is shown in the graph below:

EIP-1559 Gas Limit Changes

EIP-1559 was the only upgrade included in the London hardfork, which occurred on 5th August 2021. This EIP radically changed the economics of Ethereum, introducing a fee burn mechanism which may eventually result in net ether issuance becoming negative (deflationary). EIP-1559 also introduced a number of gas management features, to help resolve the user experience issues of stuck transactions & uncertainty around gas fees for a transaction.

These gas management improvements included the ability for the network to flex the block gas limit by a factor of 2. The ability for the miners to set the gas limit remains, but now, the block size can respond to network congestion to accomodate more transactions. This gives a higher certainty for transactions making it onto the blockchain during periods of congestion, which leads to a better user experience and potentially lower gas costs. This flex factor is known in the protocol as the elasticity multiplier.

So why doesn't the gas limit double all the time? Given that this would result in a higher uncle risk, the network includes a mechanism to increase the network fees (basefee) when the gas limit is flexed above the baseline amount (currently 15M gwei). This makes transactions exponentially more expensive over time which helps to push down demand and stabilise the block size back to the target limit. A more thorough explanation of this is available at https://insights.deribit.com/market-research/analysis-of-eip-1559/.

Gas Limit Elasticity in Action

The London hardfork occurred on the 5th of August 2021. We immediately saw changes in the block size on the Ethereum network at this time. The graph below shows, in hourly slices, what the maximum block size was on the network in the hours leading up to and just after the EIP-1559 implementation. The previous limit of 15m was immediately smashed, with the full capacity of the elastity multiplier used in the first hour.