Written by: @ 0xdoug
Translation: Blockchain in Vernacular
Someone recently posted on Twitter that currently:
Fee Status: Total fees for top L2s are around $500,000, or 10% of total L1 fees. The median fee for L2s is between $0.001 and 1 cent, and data fees for purchasing blocks account for less than 0.3% of L2 revenue.
Fee Growth Assumption: If L2 median fees double to less than 2 cents, and DA costs rise to ~50% of total L2 revenue (170x increase from today), then daily block fees could reach $500k, 10% of L1 revenue.
User price sensitivity: L2 users may be very sensitive to price, especially in the long tail segment. (The long tail segment refers to a group with a small but large demand that has a potential impact on overall revenue)
DA contribution potential: Despite price sensitivity, DA may have the potential to become a significant portion of L1 revenue. For example, if the number of blocks in Pectra is increased from 3 to 6, all else being equal, block fees could account for 20% of L1 revenue.
This is a great analysis and the most compelling bullish argument for DA. But one thing I think will never happen is that DA’s revenue will approach 50% of L2 fees. The value accumulation of sorting will always be much higher than DA for structural economic reasons.
The essence of blockchain is selling block space. Since block space is not easily interchangeable between chains, they are almost monopolistic. Not all monopolies can make extra profits, the key is whether they can charge different prices to different consumers (price discrimination).
Without price discrimination, monopoly profits are almost the same as for ordinary commodities. Think of how airlines distinguish between price-insensitive business travelers and bargain-seeking consumers, or how the same SUV model is sold at very different prices to Volkswagen, Audi, and Lamborghini.
Priority fees are an extremely effective price discrimination mechanism on the blockchain, with the highest priority transactions paying fees several orders of magnitude higher than the median transaction.
Both L2 and Solana achieve a combination of high throughput and high revenue by using sorter priority as a means of price discrimination. Marginal transactions pay very low fees, allowing for extremely high TPS. But for price-insensitive transactions, their fees are higher, accounting for the majority of network revenue.
Below is the distribution of 5 randomly selected blocks from Base L2. This is a clear Pareto distribution, making price discrimination very effective.
The top 10% of transactions contribute 30% of revenue, while the bottom 10% contribute less than 1%.
The problem is that while the sorters are making a killing from this, the DA layer can’t participate because it has no ability to price discriminate. Those high-value arbitrage trades pay the same fee for Ethereum DA as the 1 wei junk trades because they settle in one batch.
Because the value of marginal transactions is very low, high TPS can only be achieved when the median transaction can be on-chain at close to zero cost. However, for the DA layer, each transaction pays almost the same fee. Therefore, the DA layer must either achieve high throughput or obtain high revenue, but not both.
This makes Rollups essentially impossible to scale without a revenue collapse on the Ethereum network. The roadmap for Rollups is inherently flawed because it gives away the valuable part of the network (sorting) and thinks it can make it back with the worthless part (DA).
I was initially bullish on the Rollup-related roadmap because I thought any rational person would recognize the economics of price discrimination and that it would operate in parallel with the expansion of L1.
High-value, price-insensitive users will use L1 because it provides long-term, secure finality, and traceability. L2 will focus on low-value users who are squeezed out by L1’s high fees. In this way, Ethereum will still receive significant rent from the sorters.
But Ethereum leadership has repeatedly stressed that L1 is effectively dead at the application layer and will never scale. As a result, users and developers have responded very rationally, and the L1 application ecosystem is now declining, and the Ethereum network's revenue is also decreasing.
If you believe that the long-term valuation proposition of ETH is as a monetary asset, this may still be acceptable. Distributing ETH into more hands in this way gives it long-term value as a currency. At the same time, by providing subsidies to L2 without accruing value to the base layer, this also helps achieve this goal.
But if you believe that the long-term valuation proposition for ETH is as a network stake in a widely used protocol (which I think is more likely than ETH as money), then you need value accrual.
Clearly, we screwed up here due to faulty economic assumptions.
