In-depth Analysis of the Three Dimensions of New Public Chain Competition: Multi-chain, Single-chain, and Special Chain
Author: Maco, W3.Hitchhiker
Preface
Based on the previous report comparing second-tier public chains and the latest Delphi report, I have recently gained new insights into the latest round of public chain competition. The summary from the last comparison (excluding cosmos/dot) is as follows:
In terms of comprehensive public chains and ecosystems: Sol > Avalanche > Near > Bsc > Polygon > Ftm
This time, I will only compare from an architectural perspective, including ETH / Cosmos / Dot, and the views are basically consistent with the last time.
From an architectural standpoint, the current mainstream public chains can be divided into three categories: Multi-chain, Single-chain, Special
Multi-chain
Adopting horizontal or vertical scaling, another distinction lies in whether they share state, with trade-offs between security and freedom.
- Eth: L1 + L2
Advantages: The pioneer of smart contract platforms; the most secure chain; the most innovations and real users, currently ranked first without a doubt.
Disadvantages: Expensive, slow, outdated architecture, large and hard to maneuver.
- Dot: Relay + Parallel
Advantages: More advanced architecture (xcmp, wasm); impressive Gavin Wood; secure; theoretically suitable for "lazy teams," as they do not need to worry about chain design (security & economic model).
Disadvantages: The fee model puts significant pressure on "lazy teams"; the ecosystem is just starting.
- Cosmos: Hub + Zone
Advantages: More advanced architecture (Ibc, CosmosSDK); aligns with the spirit of web3's freedom and openness; project teams can fully utilize the chain's performance for customized designs (security & economic model).
Disadvantages: Loose organizational structure, insufficient empowerment; everything needs to be designed by oneself, high development threshold; the ecosystem is just starting.
- Avax: Mainnet + Subnet
Advantages: Plenty of funds; integrated architecture (new consensus + overlapping validation network); seized the EVM opportunity; comprehensive ecosystem.
Disadvantages: No interactivity between subnets; new ideas lack consensus in bear markets.
- Polygon: POS + Hermez + …
Advantages: Plenty of funds; broad layout, advanced concepts (zk + POS + DA).
Disadvantages: Overall network concept is unknown; new ideas lack consensus in bear markets; Indian team.
In summary, it appears that everyone is moving towards a multi-chain direction, including ETH, which has also shifted to this path. Multi-chain and modularization are definitely the mainstream direction. Ultimately, the choice comes down to the "impossible triangle."
Category A: ETH, Dot, Polygon
Dot is actually very similar to ETH, both relying on the main chain to achieve state sharing and settlement while ensuring the security of sub-networks (L2, parallel chains). The price paid is that they must follow the "big brother," pay "protection fees," and lack autonomy. Polygon Supernet is likely similar, so they are categorized as Category A. If project teams choose Category A, then what they care most about is (main chain) security. How can Dot and Polygon compete with the seasoned veteran ETH? Polygon's multiple solutions and strong capital layout will also be greatly weakened by similar directions like BSC. The issue with ETH is that the current route is rollup-centric, but there are three problems:
The OP solution currently has mediocre performance, with improvements expected in the future, but implementation is uncertain.
The zk solution has been given too high expectations. Although various solutions are being developed rapidly, I remain skeptical about the full functionality being realized; furthermore, it is still uncertain whether the future of zk solutions will be in cooperation with or competition against ETH.
Numerous rollup solutions are fragmented, lacking native cross-chain and collaboration, with some ecosystems showing trends of migration/multi-chain ------ growing ambitions when large, seeking alternatives when small (dydx, Boba). This can also be seen from the multi-ecosystem deployments of DeFi blue chips.
Category B: Cosmos
Building a concept similar to "super city clusters," maximizing decentralization at the cost of some security, theoretically achieving greater decentralization at the ecosystem level (Dapps are not constrained by the chain + avoiding disasters caused by single projects) and more independence in performance (Dapps enjoy their own chains). Flexibility and autonomy are actually very important yet often overlooked traits of public chains:
In terms of performance, no matter how high the performance of a single chain, it can be overwhelmed by a strange NFT mint one day, affecting other activities on the chain.
In terms of security, the entire network has "segmented security": since no one can guarantee absolute security, it is crucial to minimize losses when incidents occur. On the other hand, for Dapps, a time-sensitive Dapp certainly does not want its users to face liquidation or other unexpected losses due to chain upgrades or other unforeseen circumstances.
In terms of innovation, autonomous Dapps can leverage the underlying characteristics to turn disadvantages into advantages. For example: MEV is a natural flaw in decentralized systems; previously, everyone focused on how to eliminate it, which is difficult, while Cosmos seeks ways to utilize it, which is much more clever. Instead of suppressing it, finding ways to alleviate it is also a good approach. For instance, if a road is blocked, should we blow it up or find a way to open another road to alleviate the congestion? There are many similar cases, such as zero gas fee transactions, LP assets as staking nodes, etc. Ultimately, these innovations leverage the sovereignty of Cosmos. Therefore, this bottom-up model encourages innovation, and innovation is what blockchain encourages the most.
In terms of economics, Dot's model is "Come, all of you use me, I'm the best, pay me," which even ETH dares not do. This form of being the only one in the early stages is hard to accept from the perspective of evolutionary processes, while Cosmos attracts people first for free (similar to web2 companies spending money to gain attention) and only slowly considers collecting "protection fees" when there are enough users (atom2.0). Thus, the premise of token empowerment is having a "good product that people use." Therefore, I believe that independence and openness are Cosmos's greatest advantages over Dot and even ETH.
Category C: Avax
Avax resembles an intersection of Category A and Category B, capable of building permissionless subnets while allowing for targeted customizations; at the same time, it "partially shares security," helping subnets to cold start while empowering the native token. The latest news: the former ETH layer 2 Boba has deployed to Avax (previously, Boba also tested on Moonbeam and Ftm, and this collaboration with Avax seems to focus on gaming, combining Boba's Hybrid Compute concept, with the official promotion of a game called EvoVerses). Now Avax has both subnets and rollups, and from architecture --- ecosystem --- capital, Avax resembles a diligent "good student," none of its metrics are top, but it excels in balance, and this type should not be overlooked.
Single-chain
Focus on parallel execution, but there are also those that combine both, such as Fuel.
Smart contracts represented by EVM are executed serially, and due to their inability to leverage the multi-threading capabilities of nodes, the network's processing efficiency is low, leading to long execution times and low TPS, while EVM is inherently unsuitable for parallel execution. There are two directions for achieving parallel execution:
Using the UTXO model: Bitcoin's account model is suitable for parallel processing but is not easy to implement for complex transactions, so it is mainly applicable in the payment field. However, Cardano and Findora have explored this area.
Changing the state model: The biggest challenge is identifying "independent transactions" and "related transactions," which can significantly improve processing performance by executing "independent transactions" in parallel.
- Solana
The essence of all innovation is: minimizing network communication costs + using high-performance computing nodes. The essence of parallelism ------ changing the state model: transactions need to declare their "relevance" in advance ------ which states the transaction needs to change, thus enabling effective parallel execution.
Disadvantages: Pseudo-decentralization; extreme exploitation of node performance leads to frequent downtimes due to single points of failure. Future focus should be on the effectiveness of the fee mechanism implementation.
- Aptos
Uses blocks STM caching and adopts "optimistic execution," not requiring transactions to declare their "relevance."
All transactions are executed in parallel without distinction; if related transactions are found after execution, they are judged and re-executed, and this process repeats. This method is more developer-friendly (no need to declare transaction relevance), but if all transactions in a block are related, the speed-up goal is not achieved. However, the official claims that even in such cases, it will not be less efficient than serial execution.
Parallel execution efficiency is improved; verification nodes change every epoch, unlike Sol, which is predictable and easy to attack.
On the user side, many web2 user-friendly measures have been implemented: transaction recognition, private key mixed custody.
- Sui
Similar to Aptos, but requires transactions to declare their "relevance" in advance.
- Linera
Technical details are less disclosed, seemingly more focused on the payment field.
- Fuel
Focuses on modularity, only doing the execution layer, abandoning consensus and DA layers, currently aiming to be ETH's L2.
Uses the UTXO model, making parallel execution easy.
The parallel execution solutions face two common problems:
The extent of efficiency improvement: Mainly depends on the proportion of "related transactions" in on-chain activities; the larger the proportion, the more limited the improvement. Based on ETH data, currently about 30% of transactions fall into this category, so estimating based on ETH's gas limit and block time, parallel execution can improve efficiency by 3-5x, of course, chains with better gas limits and block times can achieve even more (100X).
Centralization issues: Higher execution efficiency means greater pressure on hardware performance, leading to validator centralization. In response to this issue, Aptos and Sui have not made statements, while Fuel is actively interacting with ETH to address this problem.
In this grouping, apart from Sol, which has been validated by the market, the others are still in early stages, and the specific implementations and issues they bring largely need to be tested in practice. Until then, skepticism is warranted, and theoretically, Libra systems, like Sol, will encounter many pitfalls. Moreover, if capital pushes too hard, it risks becoming a "doomed project." Of course, Libra systems also have many innovations; the general feedback is that the language is user-friendly, theoretically safer and easier to use, and has developers + capital, so it needs to be monitored closely.
Fuel is quite special and requires in-depth research before drawing conclusions.
Currently, I only have a positive outlook on Solana in this category: the most prosperous ecosystem among non-EVM chains + the highest single-chain performance + the best breakout effect. These three "mosts" are enough to keep it highly regarded. From previous observations, it is clear that Sol is following a very distinct "non-Ethereum" path, with all actions aimed at weakening blockchain attributes, lowering user thresholds, and making it easier to produce blockbuster projects, but at the same time, observing developer trends to guard against being drained by Libra-like projects.
Special
Public chains in this category are difficult to simply classify as single-chain or multi-chain, with very prominent features and obvious shortcomings.
- Near
The main highlight is sharding; from the developer perspective in previous reports, it shows strong momentum.
Disadvantages: The overall development of the ecosystem is currently slow; the asynchronous contract calling sharding scheme increases system complexity, and the potential inefficiencies and security risks brought by optimistic validation are also worth noting. The future direction is uncertain.
- Ar
A new paradigm public chain built on storage, inherently providing low-cost "data availability." Under the iteration of the SCP paradigm, it can theoretically achieve ultra-high TPS.
Disadvantages: The security/decentralization of SCP may have limitations, possibly unsuitable for DeFi applications; market acceptance is questionable; infrastructure is still at a very basic stage, with many standards not unified.
Attention is needed around the construction and application of infrastructure based on the SCP paradigm.
- Bsc
Top-tier traffic + unlimited bullets player + innovative CEO; it is almost the second most prosperous ecosystem, riding the wave: EVM - multi-chain - zk-bab. Although the technological innovation is weak, the superpower of "having money" can still facilitate more "suitable" solutions.
Also, due to the presence of Binance, I am less optimistic about Polygon, which has a similar approach.
In this category, there are many other distinctive chains, and since they are still in early stages, I will not introduce them one by one:
Similarly storage-based public chain Filecoin.
zk public chain Mina.
Possibly suitable for social applications, ICP.
TL; DR
Of course, the factors not considered in the above analysis may also cause changes in the competitive landscape:
Capital push (Delphi suddenly supporting Cosmos, the cash flow of Polygon / Sol / Avax).
The rise of super Dapps.
Based on normal development projections, the ranking of public chains is as follows: ETH > COSMOS > SOL > AVAX > BSC > AR.
However, it is important to note that the development of public chains vs applications may not follow a sequential progression, with many factors leading to "leapfrog" development. Therefore, it is necessary to maintain an unbiased observation and experience of all applications/ecosystems.