In October 2021, I wrote a speculative post about the “transaction quality trilemma”. Since then, pretty much all chains have implemented a minimum fee in the $0.01-$0.50 range, but there are still some holdouts like Immutable X, Solana or Arbitrum Nova which gives us data. As you know, I don’t talk about scaling and infra anymore, and the stuff I wrote in 2021 is well understood and proven out, so I don’t need to write about them. I believe writing about applications, governance and diving deep into the fundamentals of blockchains for the benefit of application developers is much more crucial at a time where we have nigh infinite scaling on the horizon with one-click rollups and massive data layers. EigenDA claims 100 MB/s. With stateful compression, this is 5 million “TPS” over validiums/optimiums settling data on EigenDA alone. But that also means with an overabundance of scaling, transaction quality trilemma will once again become an important topic.
Once your transaction fee is substantially lower than $0.01, the chain and chain’s infrastructure becomes vulnerable to spam, DDoS and micro-MEV attacks. This is particularly true if the chain also has a financial ecosystem. Let’s consider two chains with fees in the sub-$0.01 range - Arbitrum Nova and Solana. For strategic reasons, Arbitrum Nova is promoted as more of a gaming and NFT chain, while Solana is more of a degen casino chain. Obviously, there’s more to both, but those are the primary usage as far as I can see. The latter comes with a financial ecosystem, and has led to rampant micro-MEV and spam transactions - what I had previously described as “low quality” transactions.
But first, micro-MEV. The lower your fees are, it exponentially opens up the possibility of low-value MEV opportunities. There’s a mad rush to capture these, which leads to many failed transactions. Even Solana and Arbitrum Nova do have a minimum fee, but other chains could experiment with even lower fees, and it could eventually lead to a scenario where >99% of transactions are failed, spam, and just worthless.
Now, some would say - what’s the problem? Let them spam the chain away. There are two big problems:
1) Sustainability and cost: The spam is socialised across the network, and history can pile up to petabytes in short order. Innovations like validity proofs can minimize the compute burden significantly, but it’s still a linear cost for the sequencer/builder, and the history remains the endgame bottleneck. By simply not processing worthless transactions, the marginal cost of legit transactions becomes cheaper - orders of magnitude so for chains where >90% transactions are spam, micro-MEV or failed. It’s easy to sweep these under the rug in the short-term, but in the long term they can become crippling weaknesses.
2) $0.01 is too expensive for some usecases: Some usecases need free transactions. The difference between $0.01 or $0.001 is negligible if you’re making a $100 DeFi transaction, but the difference between $0.001 and $0 is everything if you’re taking an action in an onchain game or a social network.
So, it’s clear, we need solutions for the transaction quality trilemma. That brings us to Immutable X & Sorare, which have the best solutions thus far. They have free transactions, but employ “web2-like” spam mitigation methods. The key is offering an alternative censorship resistant path which does cost. This way, you get the best of both worlds - free transactions for those who need it, but censorship resistance for the edge cases. Ideally, you would want free transactions with censorship resistance, but that’s much harder to solve.
The key to achieving this is - a) state isolation, and b) spam mitigation. a) is key, otherwise you end up with micro-MEV and unrelated spam. Theoretically, this can be doable within a chain (L2 or L1), but currently it’s best done in L2s, as seen with Immutable X and Sorare.
Now, the obvious drawback here is a lack of composability, but the solution could be restricted composability with other L2s/L1. It’s also worth considering that pretty much all conceivable usecases that require microtransactions with zero fees only need composability - social, gaming, $ streaming etc.
Spam mitigation is an open-ended question - and I feel there’s significant research and engineering pending to develop the best solutions. I suspect this will need to be bespoke solutions tailored to the specific application.
Lastly, for the chains oriented towards high-value financial transactions, they will be just fine with a ~$0.01 fee floor. This is good enough to include pretty much all valuable transactions, but at the same time having minimal spam or micro-MEV. I don’t know what the optimal figure is, but it’s probably in the $0.01-$0.10 range.