Ethereum co-founder Vitalik Buterin has posted an analysis on Reddit detailing three reasons why this has been the case:
1. The ice age is delayed.
When the London fork began, the ice age had just begun to take effect. The average block time before London was about 13.5 seconds, and the average block time after London fell back to its long-term normal level of about 13.1 seconds. This is about a 3% difference in block speed, which explains 3% of the 9% increase in gas usage on the chain.
2. The target value is 15 million and the maximum is 15 million.
Before the London upgrade, the minimum gas usage of the block was 15 million. Not all blocks used the entire 15 million. Even the most complete block producer would leave 0-20999 unused gas because of the remaining There is too little space to accommodate a single transaction. In addition, there will always be occasional block producers to create empty blocks. An analysis in April showed that about 2% of the blocks were empty. Assume that there is about 2-3% unused space before London. However, after London, 15 million is not the maximum value, but the target value. This means that if the average gas used (including empty blocks) is less than 15 million, the basic cost will be reduced until the average value returns to 15 million. So this again accounts for about 2-3%.
3. Mathematical flaws in basefee adjustment.
The EIP1559 formula is not perfect when the target is 50%. Obtaining data from the most recently observed time span, 51.5% are full blocks (hence, about 3% higher than the expected 50%).