原文标题:《 Why has the chain capacity increased by ~9% after London? Three answers... 》
原文作者: Vitalik Buterin
Yangz, Babbitt
Note: Yesterday, Ethereum founder Vitalik Buterin posted a post titled "Why did the chain capacity increase by about 9% after the London upgrade?" , the following is the full text compilation.
Looking at a chart of ethereum's daily Gas usage, we can see that the average daily Gas usage has increased from about 92 billion to about 100 billion: a 9% increase.

So why does this happen?
In my opinion, this situation can be broadly explained by three different reasons :(1) ice age delays, (2) the blocks before the London upgrade were not filled, and (3) the basic fee adjustment formula was not perfect.
The London fork delayed the ice age, which was just beginning to take effect when the London fork began. Before the London upgrade, the average block time was around 13.5 seconds, while after the London upgrade, the average block time returned to its long-term normal level of around 13.1 seconds.

That's a 3% difference in block speed and 3% of the 9% increase in on-chain Gas usage.
Before the London upgrade, the largest block Gas usage was 15M. But not all blocks use the entire 15M: even the most functional block producers leave 0-20999 Gas unused because the remaining space is too small to accommodate a single transaction. In addition, there are always some block producers that occasionally produce empty blocks. An analysis in April suggested that about 2% of the blocks were empty. Overall, we can assume that the unused space in London before the upgrade is about 2-3%. However, with the London upgrade, 1500M is not the maximum but the target. This means that if the average Gas usage, including empty blocks, is less than 15M, the base fee will decrease until the average is back to 15M.
This, in turn, accounts for about 2-3% of the causes of this phenomenon.
Because of the complex relationship between arithmetic and geometric means, the EIP 1559 formula is not perfect for 50% utilization. A 0% complete block reduces the base fee by 12.5% (multiplied by 7/8), while a 100% complete block increases the base fee by 12.5% (multiplied by 9/8). So, what happens if you have a 0% full block, and then a 100% full block? As a result, the base fee will be multiplied by 63/64. So in order to keep the base cost constant, you actually need an average utilization rate of a little over 50%.
How much higher depends on what the volatility is. The theoretical minimum volatility deviation is zero: 50% of the blocks are complete, and the base fee remains the same in each block. The theoretical maximum deviation is that 53.13% of the blocks are full and 46.87% are empty (the ratio of full to empty is 1.133, the same as -log(7/8)/log(9/8));
In this case, the base fee will remain the same when the average block is 53.13% full. Actual usage appears to be in the middle of these two extremes: 51.5% of the data from the most recently observed time period is full blocks (hence, about 3% higher than the 50% expected).
Recent data analysis also broadly confirms the latter two numbers (look closely at the blue line before and after the London conference, and at the 15M level on the chart).

One possible way to solve this math problem is to make the base cost adjustment mechanism more explicitly exponential:https://ethresear.ch/t/make-eip-1559-more-like-an-amm-curve/9082. This will establish a hard constant factor that the base cost can be calculated as a direct function of the total "excess" Gas usage (thus, for any level of excess Gas usage, the base cost will have to eventually approach infinity).
But for now, Ethereum users can rejoice in the unintentional 6% capacity increase from the London upgrade (while client developers need to continue working on stateless/state expiration issues to ensure we can safely handle state growth).
The original link
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia