I have raised similar concerns in the past, about how rollups may not be enough, and we’ll need a variety of solutions.
Kudos to the author for concluding a validium and rollup hybrid is the solution that’s significantly superior to centralized sidechains/L1s, but with even lower fees — this is the “third option” that many often tend to ignore. This solution is called volition — Immutable X is set to be the first example, with zkSync 2.0 + zkPorter following later this year.
However, I want to address the wooly mammoth and the blue whale in the room.
1) Centralized chains do not have a sustainable economic model: Polygon and BSC can offer arbitrarily low fees because the chain isn’t yet running at capacity. Any arbitrarily low gas price will be accepted by validators as first price auction mechanism hasn’t kicked in. Indeed, we have seen as both chains have gotten closer to capacity, the gas prices have started to rise as some users engage in bidding. Of course, the solution has been to increase gas limits, but this is obviously not sustainable. Eventually, both chains will reach hard limits of keeping a distributed ledger in sync across multiple nodes, or the EVM/client (in both cases, Geth forks). Even if you improve the VM to be more parallelized and further centralize the network, state growth will hit unsustainable levels in the long term. Meanwhile, given your only selling point is low fees, there’ll be a complete imbalance between transaction fee revenues collected by the network, and very high block subsidies issued to validators to keep the network secure in the face of rising costs. These chains necessary have delegated-type proof-of-stake protocols which have high inflations, that can lead to several orders of magnitude difference between revenues and issuance. Case in point: Polygon PoS is currently collecting ~10,000 MATIC in transaction fees daily, while distributing over a million (please correct me if I’m wrong — seeing conflicting data online).
Now, the argument here would be, over the long term, these two values will converge as the network matures. But they cannot! Either the network will become even less secure, or the transaction fees have to go up. There’s no other way.
In the short to medium term, centralized high-TPS chains can be subsidized by speculators. However, this is not sustainable long term, and I believe these chains will either capitulate to increasing transaction fees, or implode. Indeed, Binance Smart Chain has already done this, with significantly higher gas prices now than were originally promised. Techniques like state expiry will help. You can also basically become a validium-esque L1, with zk-SNARKing the VM, and using a separate data availability chain with erasure coding and data availability sampling. But this is basically only drawing parity with a validium, while still being thousands of times less secure and decentralized than a validium that commits state root diffs and zk proofs to Ethereum.
I don’t see any path for high-TPS chains to survive. They’ll always lose to validiums, in every respect.
2) Data shards. The most disappointing part of the article is that it seems to completely neglect the other half of the puzzle — data sharding. This is as important as rollups, which is why I always call the solution “rollups + data shards”. Perhaps we need a catchier name for it.
With data shards, rollups can get to tens of thousands of TPS in the medium term (by 2023), but millions of TPS in the long term (2030s) as more shards are added and each shard is enhanced alongside with Moore’s and Nielsen’s laws — not to mention new techniques that may be invented in the future. zkPorter is a fantastic short-term solution, but with data shards, you’re going to get this sort of scalability with a rollup itself, without the need for any compromise. Indeed, the developers of zkPorter themselves acknowledge this:
The current consensus is Eth2 data sharding will arrive by the end of 2022 to provide an exponentially larger data availability layer without sacrificing decentralization. zkSync’s zkRollup technology combined with Eth2 data sharding is the endgame, hitting 100,000+ TPS without sacrifices on any of the 4 factors.
I do think volition and validium like solutions will continue to exist, but for a majority of meaningful, valuable transactions, I’m not worried that nobody will care about rollups long term.
Keywords: long term. It’s going to be a bumpy road, and people will be distracted by hyped and unsustainable short-term solutions, but rollups + data shards are the blockchain industry’s only viable route to critical adoption long term. Until a better solution emerges.