Opinion: Why L2s Won't Drive Ethereum into Deflation?

Original source: BREAD X account

Author: BREAD, Crypto KOL

Compiled by: Felix, PANews

There is a prevailing view in the crypto community that L2s will cause Ethereum to fall back into deflation. This statement is incorrect, and there are examples to prove it.

This article is a report on L2s, blobs, and why the Ultrasound Money meme will fade away without mainnet users.

(Note: A meme related to Ethereum that emphasizes its potential to become deflationary in the long term. The basic concept is that if the supply of gold or Bitcoin is capped and considered "Sound Money," then Ethereum's supply is decreasing and should be regarded as "Ultrasound Money.")

Key Points:

• L2s will only make ETH deflationary when both blob and regular fee markets are saturated.
• L2s are largely unconstrained by the specific batching rhythm of L1.
• The combination of the above factors means that L2s will continuously coordinate with each other to avoid creating a high-fee environment for themselves.

Background

First, let's briefly review the state of Ethereum after Dencun (the March upgrade made L2s about 10 times cheaper for users).

Dencun introduced the concept of blob space, which is an expansion of block space on Ethereum designed to allow L2s to publish their batch data.

This new area has several notable features:

• "Blobspace" currently limits a maximum of 6 individual blobs per block.
• It is a fee market independent of regular block space but uses similar mechanisms.

Blockspace: If the utilization of the current block >50%, the base fee for the next block increases.

Blobspace: If there are 4 or more blobs (i.e., >50%) in the current block, the base fee for the next block increases by about 12%.

Note: L2s can decide for themselves whether to use Blockspace or Blobspace.

Given the above, we can predict some behaviors in the expected flow of these L2s that will lead to destruction:

• L2s will first saturate blobspace until it is no longer idle (3 blobs/block).
• Once this level is reached, they will begin to calculate the cost/reputation savings of publishing and choose between the following two options:

• a.) Use calldata
• b.) Reduce publishing frequency / decentralize / coordinate to cool down the fee market

That is to say, even though top L2s are already earning millions in profits each month (with profit margins close to 100%), L2s will continuously adjust their behavior to avoid creating a high-fee environment for themselves.

ETH vampire death loop:

• Extend low-fee environment
• Wait for Ethereum scaling, reset the market
• Extend low-fee environment
• Wait for Ethereum scaling, reset the market…

So when does ETH plan to scale?

The earliest would be next year's Pectra update, which could expand blob capacity by about 2-3 times.

However, the market does not provide many such opportunities, but there are two examples of L2s deliberately modifying their behavior to save costs and avoid fees (as any business or rational actor would do).

Example 1: Blobscriptions Craze

In late March, blobs experienced "price discovery" during the blobscription phase; nonetheless, L2s and builders averaged their blobs at a non-price-increasing level of 3 blobs/block:

Data source: Dune

They were not obligated to exceed the fee growth level, so they did not.

Example 2: LayerZero Airdrop

On June 20, the ZRO airdrop led to a significant increase in transaction volume on Arbitrum, resulting in an influx into the blob market.

In short:

• More blobs from ARB
• Higher blob costs for all L2s
• Due to the lack of proper infrastructure to switch to regular block space, L2s destroyed about $800,000 worth of ETH.

If you want to learn more, you can refer to the author's other tweet.

The biggest takeaway from this event is the response of various teams to the high-fee environment:

Scroll is a zk rollup with no publishing obligations, completely halting publication.

When looking at Scroll, it is found that they are indeed faster than normal, but their response is much slower than OP Mainnet. Interestingly, when base fees soared and* blobdata became more expensive than calldata, Scroll completely stopped publishing blobs. This also explains why* Scroll has the lowest fees compared to other L2s.

Taiko is a rollup with strict rhythms, slowing down batching speed.

Since Taiko is based on Rollup (i.e., once their transactions are confirmed on L1, they are confirmed on* L2), they must continuously publish information to keep the chain running. This led them to overpay fees of* 25+ ETH on L1. However, when the* blob base fees soared, even Taiko slowed down its batching speed by 30-50%, indicating that they do have some price sensitivity at extremely high prices.

Both market participants took reasonable measures to reduce management costs.

In the future, they will certainly adopt automated processes to avoid wasting hundreds of thousands of dollars before switching.

All L2s will.

What Can Be Done?

When the author set out to write this report, they simply wanted to calculate "the need for 4 more foundational L2s to generate ETH Ultrasound Money™️ again," but the more they dug, the more they realized that this goal could never be achieved.

L2s will continuously adjust their behavior to avoid high costs. They are businesses, and of course, they will do so.

So given the above, what should be done? Make the mainnet "cool" again to attract users and builders.

A balance must be struck between scaling through L2s and keeping high-level users on the mainnet, rather than indiscriminately pushing them into one of a dozen different structured ecosystems, which have minimal economic returns for Ethereum.

Progress can be made by adjusting messaging:

• Fill capacity
• Scale to suitable levels
• Tell users and applications to "get lost" (skip this step)
• Scale through L2s

Measures like EIP-7623 can also be implemented to prevent L1 user block space from being used as a fallback space for L2 cost savings.

Isn't it better to rely on L2s as participants, with the mainnet as pure settlement? Given the current incentives, that is not the case.