I definitely agree that we need to find a better way to price blobs. However, I don’t think the domination of execution gas in L2 costs is necessarily an issue.
The decrease in blob demand due to high execution gas prices demonstrates negative cross-elasticity of demand, indicating that blob gas and execution gas are complementary goods. This joint demand relationship is straightforward to understand since both are required for blob-carrying transactions. Consider the analogy of electricity and appliances: rising electricity prices may reduce demand for electrical appliances. However, it would be absurd to argue that appliance prices should rise alongside electricity prices simply because high electricity costs dominate appliance usage expenses. This would only further reduce appliance demand. Coffee and milk provide another useful analogy.
I’m not suggesting that fee markets for blobs and gas should be completely independent. In most cases involving complementary goods, their prices are interconnected. For example, companies often employ strategies like pricing printers low and ink cartridges high to maximize profits, or similarly with gaming consoles and games. Therefore, I’d like to see whether bound markets for various resources could maximize Ethereum’s overall benefit. If not, such binding may simply complicate the pricing process for both resources, potentially causing significant problems.
In my opinion, regarding the proposed implementation changes, the execution gas price simply serves as a price dampening mechanism for blob gas prices to establish a price floor. The resulting price may appear functional in the future, but it will no longer accurately reflect the supply and demand dynamics for blobs alone.
There are several fundamental questions we need to answer:
Under the multidimensional gas roadmap, should fee markets for different resources be independent or bound together? Independent fee markets offer clear benefits: resource prices reflect only their own demand, and we can optimize usage of different resources separately. Bound fee markets, however, require further research.
If we choose bound fee markets, why would this design be superior to the current one-dimensional gas model? Could an optimally designed bound fee market essentially become a new form of one-dimensional pricing?
Good proposal to make the blob market effective. Furthermore, I’ve identified a potential issue: Economic Cliff at Target Threshold.
The mechanism creates a sharp incentive cliff: when blob_used < target, fees can decrease; when blob_used >= target, fees continue to increase when BLOB_BASE_COST * parent.base_fee_per_gas > GAS_PER_BLOB * get_base_fee_per_blob_gas(parent) is met. This encourages projects to coordinate keeping usage just below the target (e.g., 95%), even when real demand exceeds the target, which leads to systematic underutilization.
Would prioritizing ensuring blob base fee is bounded by execution cost better align incentives?
Yes it is straightforward to understand in retrospect, after this EIP explained it Concerning the analogy, it must be adjusted to reflect the store’s (Ethereum’s) actual circumstances:
The store currently uses a price update rule that shifts the appliance price by 12.5% per day (or every 12 seconds), depending on how many appliances that are sold. Changing this rule is not feasible in the near term, and it works well in the range where customers care about appliance prices. In a scenario where a high cost of electricity makes appliances less useful, the price falls and falls down to 1 wei. Customers stop caring about the appliance price long before that—only the price of electricity matters. Once demand for appliances picks up, the price rises too slowly, such that it for a long time does not influence demand. Customers must then instead tip the store’s employees to purchase an appliance. This relates to fee inelasticity discussed here in the EIP.
The store uses electric trucks for fetching the appliances from the main warehouse and for delivering each appliance to the customer. This is an operating expense that is not currently passed on to the customer. Adapting the price with the store’s variable electricity costs of delivering each appliance is reasonable. This relates to the KZG proof verification cost discussed here in the EIP.
The price of electricity fluctuates considerably. This motivates the dynamic solution to the problem at hand.
Regarding the question of accurately reflecting supply and demand, note the previous discussion about a composite fee market here.
I agree that there are many nuances concerning pricing in a multidimensional fee market, and I expect some additional innovation there. However, I would also like to share that the most pressing current issues are (1) dealing with subcalls within the EVM and phasing out gas introspection; (2) the weakening of FOCIL CR under independent base fees. I have recently published EIP-8046: FOCIL with ranked transactions (FOCILR), which potentially resolves concerns pertaining to FOCIL. Work on the EVM remains to be done.
Concerning the analogy, it must be adjusted to reflect the store’s (Ethereum’s) actual circumstances: