Blockchain Killer: The "Impossible Triangle" Troubling Transaction Quality
Original Title: "Transaction quality trilemma - blockchain killers"
Original Author: polynya
In October 2021, I wrote a speculative article about the "transaction quality trilemma." Since then, almost all chains have implemented minimum fees ranging from $0.01 to $0.50, but there are still some chains like Immutable X, Solana, or Arbitrum Nova that are reluctant to comply, which has given us some data.
As you know, I no longer talk about scaling and infrastructure; what I wrote in 2021 has been deeply understood and proven, so I don't need to write about it again. I believe that as we have almost unlimited scaling, one-click rollups, and massive data layers in sight, it is more critical for application developers to delve into the fundamentals of blockchain and write about applications and governance. EigenDA claims to process 100 MB per second. With state compression, that translates to 5 million "TPS" on EigenDA, settling data only on Validiums/Optimiums. But this also means that with excessive scaling, the transaction quality trilemma will again become an important topic.
Once your transaction fees drop significantly below $0.01, chains and their infrastructure become vulnerable to spam, DDoS, and tiny MEV attacks, especially if the chain has a financial ecosystem. Let's consider two chains with fees below $0.01 - Arbitrum Nova and Solana. For strategic reasons, Arbitrum Nova is more promoted as a gaming and NFT chain, while Solana is more like a frenzied casino chain. Clearly, both have more functionalities, but in my view, these are their primary uses. The latter comes with a financial ecosystem, leading to a large number of tiny MEV and spam transactions - what I previously described as "low-quality" transactions.
But first, let's talk about tiny MEV. The lower your fees, the more likely low-value MEV opportunities arise. People rush to seize these opportunities, resulting in many transaction failures. Even though Solana and Arbitrum Nova have minimum fees, other chains might attempt even lower fees, ultimately leading to over 99% of transactions failing, being spam, or having no value.
Now, some might ask, what's the problem? Let them spam attack the chain. There are two major issues:
Sustainability and Cost: Spam is shared across the entire network, and historical data can accumulate to several PB in a short time. Innovations like validity proofs can significantly reduce computational burdens, but for sequencers/builders, it remains a linear cost, and historical data is still the ultimate bottleneck. By simply not processing worthless transactions, the marginal cost of legitimate transactions becomes cheaper - for those chains where >90% of transactions are spam, tiny MEV, or failures, costs are reduced by several orders of magnitude. It's easy to set these issues aside in the short term, but in the long run, they could become fatal weaknesses.
$0.01 is too expensive for some use cases: Some use cases require free transactions. If you are making a $100 DeFi transaction, the difference between $0.01 and $0.001 is negligible, but if you are taking actions in on-chain games or social networks, the difference between $0.001 and $0 is everything.
Therefore, it is clear that we need solutions to address the transaction quality trilemma. This brings us to Immutable X and Sorare, which are the best solutions so far. They offer free transactions but adopt a "Web2-style" spam mitigation approach. The key is to provide an alternative censorship-resistant path, which comes at a cost. This way, you can have the best of both worlds - providing free transactions for those in need while offering censorship resistance for edge cases. Ideally, you might want to provide free transactions while being censorship-resistant, but that is harder to solve.
The keys to achieving this are:
a) State Isolation
b) Spam Mitigation.
a) is crucial; otherwise, you will end up facing tiny MEV and irrelevant spam. Theoretically, this can be done within a single chain (L2 or L1), but currently, it is best implemented on L2, as shown by Immutable X and Sorare.
The obvious downside here is the lack of composability, but the solution might be limited composability with other L2/L1s. It is also worth considering that almost all imaginable use cases requiring zero-fee microtransactions only need composability - social, gaming, etc.
Spam mitigation is an open question - I feel that a lot of research and engineering work is still needed to develop the best solutions. I suspect this will require tailored solutions to fit specific applications.
Finally, for chains oriented towards high-value financial transactions, they can accept a fee floor around $0.01. This is enough to include almost all valuable transactions while minimizing spam or tiny MEV. I don't know what the optimal number is, but it might be in the range of $0.01 to $0.10.