Partnered With:

Why is single-chip L1 Blockchain a "dead end"?

Read this article in 27 Minutes

Once Optimistic Rollups is ready for its prime time, all game is up for a monolithic L1 blockchain network.

Ethereum researcher Polynya: Why the monolithic L1 blockchain is a "Dead end"?

Original author: Polynya

Original editor: South Wind, Unitimes

I've never written an article about "Optimistic Rollups vs. ZK-Rollups." They're all great. My article is all about arguing against extremely inefficient, unscalable, insecure, and inflexible monolithic L1 blockchains, and we've seen the ridiculous allocation of capital to these L1 networks. My sole purpose is to see blockchain expand to global ubiquity in a technically, economically and socially sustainable way. Not only are monolithic L1 blockchain networks unsustainable, they'll never actually deliver the throughput required -- and they're orders of magnitude off target. But that's not a problem anymoreWe need a modular executive layer. Data availability sampling to extend the blockchain. This has never been clearer. The sooner we recognize this and move to this approach, the better the blockchain industry will be. Yeah, that's just my opinion. As always, please treat this article as an amateur's stream of consciousness ramble, not a professional research article.

Monolithic L1 blockchains are (still) extremely inefficient

I've been writing reviews of Rollups since 2020 and blogored this time last year when the prevailing narrative was "Cardano/ other L1 smart contract platforms are coming and there will be a massive migration of users from Ethereum to other L1 networks overnight." "The L2 network is a temporary stopgap, the only way to scale is the L1 network." However, even with the mildest tests at the end of 2021, these narratives of alternative L1 networks collapse with alarming speed:

Binance Smart Chain (BSC)The system is more demanding due to state bloat. As a result, the nodes start to get out of sync, causing many problems. The problem here is that the system they specified was very low -- but they didn't tell people,System requirements grow exponentially with state inflation. Now they are making reckless network changes without adequate auditing or testing. Of course, switching to Erigon and having multiple chains will provide incremental benefits, but there are always serious limitations.

Solana Instead of making that mistake, they were very open to very high system requirements from the start. Now, I'm definitely not going to criticize Solana for its various failures and problems, because that's largely because it's oneEarly beta products. Bugs and problems due to missing features are always going to show up in beta projects, whether it's Rollups, DApps, or single-chip L1 networks, and I just hope the developers fix them. But the problem is that over time, as Solana matures, gets tested to eliminate bugs, and implements a cost market, it will need to provide an incremental throughput andA huge price to pay in terms of technological and economic sustainability. Solana is by nature unscalable. Optimistic Rollups is a much better solution than Solana and will mature quickly.

Polygon PoS Arguably the most adopted chain after Ethereum. Now, Polygon PoS is indeed a "commit chain" rather than an alternative L1 chain, but it's still a monolithic chain, andConstrained by the same inefficiencies as L1 networks. Polygon PoS reached its limit and was affected by the garbage trade, raising their minimum Gas prices. But even after that, there were a lot of projects spamming it, causing Gas prices to rise by more than $0.10. To be clear, this is a much better result than Solana or Cardano,In Solana or Cardano, 99 percent of transactions fail during congestion, and only the tiny MEV robot succeeds. To its credit, the Polygon team, unlike other monolithic blockchain projects, has been very open about acknowledging these limitations andPut all your energy into zK-rollup-- This will actually be highly scalable. Actions speak louder than words, and The Polygon team deserves credit for investing $1 billion in zero-knowledge technology.

When it comes to & have spentCardanoThey are also a very early beta product, like Solana, tooCost markets must be realized. Cardano's system requirements remain low. Recently, I've seen a growing interest in Rollup from the Cardano community, and it's great to see that! However,All of this is in vain until Cardano itself implements data availability sampling.

We see many other projects that have not lived up to their hype. Whenever the block space is saturated,Avalanche C-Chain I mean,This is a basic feature of a monolithic block chain. Subnets lead to fragmentation of security or decentralization and will be subject to the same severe limitations (i.e., soaring costs). As for Avalanche's "online Purning" : Let's see, but this seems to be taking Geth's "offline Purning" and making it trim at a higher frequency. This might be a nice complement to Geth, butIt absolutely does not address the fundamental limitations of state bloat. We've seen Harmony fail and so on. But I also want to highlight these projects that are building scalable solutions for the next generation:Data availability sampling for Ethereum, Tezos, Celestia and Polygon Avail; Certification of validity of Mina and Aleo; And there are many RollupsNow there seems to be a new Rollup every week! It is clear that we have entered the era of "modular architecture" -- few people are seriously building new single-piece L1 blockchains. In a pinch,"Prototype modularity" projects such as Polkadot and NEAR are medium-term solutions that, while not addressing many of these issues, remain sustainable and secure. If you don't care about sustainability and security, Dfinity/ICP is building interesting things.

I also want to be clear that there is a way forward for monolithic L1 blockchain:Upgrade to next generation technology. Avalanche, for example, could implement a "data availability sampling subnet" underneath that is validated by an entire set of validators and introduces a modular execution layer. Unless this is a priority on their roadmap, I will continue to push this narrative until this shift is widespread across the industry.

Now, not every chain has to be a Rollup or have some form of modular design.When security is not important, and when you want to implement something in the. Sovereign L1 chains still have a place when it comes to new features that are difficult or impossible to implement in Rollups.Sure, it's very much a niche market, but it's real.Cosmos Ecosystems do this very well,Polygon Edge Compelling solutions are also being built, although I would like to see these chains proven to be effective, and IBC (inter-chain communication) is evolving to validate the proof of validity."Bridge" across L1 blockchainSame thing. But even if it turns out to be a perfect L1 bridge in terms of validity, like =nil; Foundation is buildingMina < > Ethereum bridge", stillYou need to assume that this more vulnerable chain (i.e. Mina chain) will not be attacked; By contrast,Rollups "bridge"Provide complete security, even if the weaker Rollup chain is attacked, you can still inherit the security of the stronger chain.

Finally, timing is important. Currently, Optimistic Rollups aren't ready, so it still makes sense to use these non-beta monolithic L1 blockchains. It should be noted that both Optimistic Rollups and single L1 chains existVarying degrees of maturity/instability, so you have to evaluate it on a case-by-case basis. But I'm not going to talk about that in this article, my only interest isHow can blockchain scale on a large and sustainable scale in the long term. But & have spentOptimistic Rollups is maturing quicklyThere is a clear path to becoming a sustainable solution. It just takes engineering work -- there's a good chance that at least one Optimistic Rollup will be fully decentralized, achieve data compression, have a highly mobile "bridge" and scale up in a year's time.Once Optimistic Rollups is ready for its prime time, all game is up for a monolithic L1 blockchain network.

In the past, I have taken a very long-term view, so I have probably discussed ZK-Rollups in a more positive light.I always thought zK-Rollups would be the end of the game and thought all Optimistic Rollups Will either convert to ZK-ROLlups or use the validity proofs to replace its fraud proof system.But that could be three years away.

The conclusions of part I above are:The monolithic L1 blockchain is a dead endThe evidence on this point is overwhelming so I want to emphasize this point once and for all. The second part of this article, which follows, will focus primarily on Ethereum, because ethereum is where all the Rollups innovation is located, and ethereum has the most ambitious scalability roadmap.

Optimistic Rollups vs ZK-Rollups

Now, let's talk about Optimistic Rollups and ZK-Rollups. By the way, I recently learned about Obscuro (a TEE Rollup) and Urbit's Naive Rollup. The Rollups design space is a blank canvas, so there can be many types of Rollups! But here, I'm talking specifically about the basisFraud to prove (Optimistic Rollups) andProof of validity (ZK-RolluPS) security Rollups.

Transaction fee

Let's startApplication-specific Rollups Beginning: Obviously,Zk-rollups is ahead in this regard.Loopring, zkSync Such schemes are popular in the payment space, with fees for ERC-20 Token transfers around us $0.10. At the time of writing, a decentralized trading platform based on ZK-rollup technology & NBSP;ZigZag The fixed fee is $0.28, including Gas and transaction fees. At the same time, decentralized derivatives trading platform & NBSP;dYdX There is no Gas fee for users, but we can calculate the transaction fee for the platform: on high transaction volume, they pay ethereum $0.08 Gas fee for each transaction; On lighter trading days, it is around 10 cents.

Now, Optimistic Rollup is likely to be able to provide that range of fees -- in fact, the Worldcoin project plans to use.Hubble Erc-20 transfer fee (based on Optimistic Rollup's L2 network) currentlyBelow the $0.10 range. We move on to smart contracts Rollups -- why currently.Optimism and Arbitrum So expensive on the Internet? At the time of writing, according to the data of lefees.info, the Token exchange fee on Optimism is $0.85, and that of Arbitrum is $1.35, which is really too high!

The answer is simple:These Optimistic Rollups themselves are not optimized, and most of the early dApps projects deployed on Optimistic Rollups are forked versions of existing projects on Ethereum and are not designed for Optimistic Rollups.More recently, optimised for Optimistic Rollups.Aave V3 Successfully reduced the cost by about 10 times!

But there are many more optimizations to come.Signature aggregation/compression will result in direct savings of approximately 1,000 Gas per transaction.Basic compression of calldata can be directly saved by 2.5 times, and more advanced solutions are coming.Optimistic Rollups transaction fees will easily reach the $0.01- $0.10 range as Optimistic Rollups and dApps deployed on them are optimized.

Wait, this was before The Surge phase of ethereum's roadmap, which dramatically increased Rollups scalability through sharding! If the late 2022 Ethereum Shanghai hard fork upgrade (which will be the first upgrade since the ethereum merger) goes well, the first step will most likely be the introduction of"Carry a BLOB transaction"(Blob-carrying transactions, a new format proposed by Vitalik),This will bring the cost of callData down to about zero, and Optimistic Rollups will actually become cheaper!

Today, ZK-Rollups have high fixed costs. Currently, zK-Rollups is cheaper because its call data cost exceeds this fixed cost; However, as the cost of calling data becomes negligible, transaction fees will be dominated entirely by the fixed cost of ZK-Rollups. While & have spentZk-rollups is much more expensive to run.

Obviously, there are a lot of different proof systems out there, and the costs vary, butCommon estimates range from $0.01 to $0.02.This will be one of the lowest costs for ZK-Rollups. However, if they want to,Optimistic Rollups are free to go lower: Since 99% of the cost of a mature Optimistic Rollups is the cost of calling data, they can access ~5200 TPS of blobspace at negligible cost.

By the way,Optimistic Rollups will also be cheaper than the Validiums scenario, which keeps its data availability down the chainBecause this lowest cost also applies to Validiums! I believe thatzkPorterTransaction costs (this is a Validium scheme) range from $0.01 to $0.03, while Optimistic Rollups can cost much less than that. Personally, I don't recommend using this scheme for transaction quality reasons, but in this case Optimistic Rollups will end up with a higher "profit" that can be redistributed to stakeholders or used for public goods financing, development, etc.

Eventually, of course, ~5200 TPS will saturate, and if there is a huge demand beyond that, call data costs will start to rise again. But by then,Danksharding Will roll out, first expanding this space to 125,000 TPS and then to millions of TPS within a few years. (By the way, the "TPS" number is meaningless when talking about Rollups, but it is useful for illustrative purposes.) In the long run, the cost of calling data will never be the bottleneck -- the bottleneck will be the Rollups themselves. To reiterate, the bottleneck will be Rollups itself, not Ethereum.

In the long run,As ZK-Rollups matures, we will have ASIC certifiers and the cost of ZK-Rollups will become insignificant. When it does, zK-Rollups' advantages -- most notably instant withdrawals (no waiting) -- will win out. For some use cases with highly compressed state increments (such as dYdX), they will soon be much cheaper than Optimistic Rollups.

In the end (Finality)

Speaking of which, a common misconception is that Optimistic Rollups has a 7-day knock. But in fact, Optimistic Rollups will achieve the same finality as L1 faster than ZK-Rollups. We've seen Optimistic Rollups submit a trade lot every 5 to 10 minutes, so that's a delay. These 7 days are to ensure that this finality is maintained, assuming that at least one entity will maintain this finality.As Optimistic Rollups expands, these transaction batches will be submitted more frequentlyAnd Optimistic Rollups can be submitted in each block when it reaches about 20 TPS, at which time the finality of Optimistic Rollups will equal the finality of L1. Due to the much higher fixed cost of ZK-Rollups, committing on each block will require more activity (> 100 PTS) to achieve. However, with this BLOB transaction EIP (Ethereum Improvement proposal) proposed by Vitallik, ZK-Rollups can be redesigned to submit some proof to bloB, so this may not be a problem.

So what are those 7 days for Optimistic Rollups? Because the challenge period is 7 days, there is one deal that does have a 7 day knock period, which is withdrawal from Optimistic Rollups to L1 chain. In some cases, such as cross-chain NFTs, this can be a challenge. However, we have seen many "Bridges" coming online in Optimistic Rollups, and as Optimistic Rollups matures, activity & Liquidity will increase,The seven-day challenge period will no longer be an issue. The bigger issue is that ZK-Rollups supports many new use cases and cross-rollup activities that are not possible in Optimistic Rollups.

For example, with the help of Danksharding, ZK-Rollups can call L1 synchronously, and I even speculated that if we implemented L1 pre-confirmation through crLists, some degree of composability between different ZK-Rollups would be possible! Optimistic Rollups is left out of these new and innovative scenarios. Appendix:While Optimistic Rollups can make private transactions, ZK-Rollups will be able to do so at a much lower cost.

throughput

Application-specific Rollups are fairly lean and scalable. If I remember correctly, StarkEx demonstrated 9000-18000 TPS back in mid-2020. For smart contract Rollups, however, things are more challenging. We have seen that StarkNet is only capable of throughput lower than dYdX or Immutable X, with throughput optimization as a top priority. Since both Optimism and Arbitrum are based on EVM clients, they have a) battle-tested code base and b) relatively optimized clients. A few days ago, I asked how far EVM could scale before parallelism was required, and Alexey Sharp said Erigon could scale up to 500M Gas/ SEC. Therefore, there is a lot of room for optimized EVM-based Optimistic Rollups, and more room can be provided through multi-threaded clients or multiple instance/recursive Rollups.

This is our point of contention: because Optimistic Rollups has an honest minority of assumptions, you can't push system requirements too hard. To be clear, Optimistic Rollups can be more secure than any monolithic L1 blockchain because this is an honest minority assumption rather than the honest majority assumption of monolithic L1s. But you still need to have a limit, which by default needs to be lower than zK-Rollups, because zK-Rollups can be easily verified by validation.

But Optimistic Rollups has a solutionStateless client. I saw this on Optimism's roadmap, but I can't remember where. It's easy to verify Optimistic Rollups with stateless clients, and it's easy to verify transactions with the right tools. In addition, Rollups are more aggressive than L1 by using state expiration because it is easy to reconstruct the Rollup state directly from L1. Once statelessness and high frequency state expiration are implemented, Rollups will have overcome the serious problem of state bloat! In addition, you have a new solution like Fuel V2, which is parallelized through a UTXO-like system that costs no transaction output.

Having said that, zK-Rollups is still a more elegant solution for achieving maximum network throughput through recursive validation. It's likely that as StarkNet itself matures, we'll see application-specific ZK-Rollups build huge throughput on top of it, and as a whole, StarkNet is far bigger than any Optimistic Rollups. The tradeoff, however, is atomic composability -- unless the zK-Rollups team wizard discovers that soon!

conclusion

In the end, it's all about the timeline.

For application-specific chains, ZK-Rollups is already the best solution for most situations. This should be obvious to anyone who has used dYdX, zkSync, Loopring, or Immutable. There will be some trade-offs, but all of this will be done by the end of the year (e.g. DYdX V4).

Smart contracts: The monolithic L1 chain has a year left to remain relevant.The writing is on the wall for these monolithic L1 chains and there is no alternative but to make an urgent switch to proof of fraud, proof of validity and DA (data availability) proof.

Optimistic Rollups will scale faster than ZK-Rollups. Much of Optimistic Rollups' code base is mature and will be ready for mass adoption once sequencing and upgradability are decentralized. By the end of 2022, Optimistic Rollups could have a sub cent Gas fee, be fully decentralized, have essentially the same security and finality as Ethereum L1, and have enough liquidity for quick withdrawals. andOptimistic Rollups will continue to improve with the implementation of Statelessness and state expiry in 2023.

Zk-rollups will continue to evolve, mature and tested, optimizing proof time, moving to GPU prover and eventually ASIC prover. Their novel virtual machine and sequencer nodes will also mature and expand over time.By the end of 2023, I expect ZK-Rollups to catch up and replace Optimistic Rollups. But Optimistic Rollups will continue to work until 2024/25, by which time I expect most Optimistic Rollupsor to become ZK-RolluPs, or at least fraud proof systems that replace them with proof of validity.

I have said many times that my time frame is 5-10 years. But Rollups and DAS (Data availability sampling) are moving so fast that I now think the end is in sight and will happen within 5 years.

Thanks to Reddit user U /proof_of_lake for proofreading.