Breakpoint 2024: Debate: 100 Validators Are Sufficient to Run a Decentralized L1

In a groundbreaking debate at Breakpoint 2024, industry leaders clashed over a fundamental question for Solana's future: Is a network of just 100 validators sufficient to run a decentralized Layer 1 blockchain? The discussion revealed surprising insights about network security, economic incentives, and the future of blockchain infrastructure.

Summary

The debate, moderated by Dan Albert from the Solana Foundation, featured Brian Long, co-founder of Triton 1, arguing in favor of a 100-validator network, while Anatoly Yakovenko, CEO of Solana Labs, opposed this view. Brian Long presented a case for the inevitability and benefits of a more concentrated validator set, citing economic factors and potential performance improvements. He argued that economies of scale would naturally lead to a smaller number of more financially robust validators.

Anatoly Yakovenko, on the other hand, emphasized the critical importance of physical security and decentralization. He contended that a larger number of validators is necessary to ensure network resilience against partitions and to maintain both consistency and availability. Yakovenko drew parallels to internet infrastructure and the need for a widespread, physically secure network.

The debate touched on crucial aspects of blockchain technology, including economic incentives, network performance, security, and the practical challenges of maintaining a truly decentralized system. Both speakers presented compelling arguments, highlighting the complex trade-offs between efficiency, security, and decentralization in blockchain networks.

Key Points:

Economic Factors Favoring Fewer Validators

Brian Long argued that the economics of running a validator naturally lead to concentration. He explained that larger validators with more stake can generate more fee revenue, allowing them to share more rewards with stakeholders. This creates a feedback loop where stake tends to concentrate on larger nodes. Long presented an example where a validator with 2 million SOL stake could share significantly more rewards compared to a smaller validator with 200,000 SOL stake.

This economic reality, according to Long, makes the transition to a network with fewer, but more financially robust validators inevitable. He suggested that this concentration isn't necessarily negative, as it allows validators to invest in more advanced infrastructure, potentially leading to better network performance and user experience.

Performance and Infrastructure Improvements

Long painted a picture of a future where a smaller set of well-funded validators could invest in sophisticated hardware setups. This could include FPGA network congestion hardware, multi-GPU setups for advanced computation, and separate databases to manage larger account databases. He argued that this level of investment would lead to significant performance improvements, such as 90% of transactions landing in the same slot and confirmation times of less than one second.

Furthermore, Long suggested that financially healthy validators would be in a better position to expand geographically, potentially addressing the current lack of validators in the southern hemisphere. This could lead to a more globally distributed network, even with fewer total validators.

The Importance of Physical Security and Decentralization

Anatoly Yakovenko presented a contrasting view, emphasizing the critical importance of maximizing physical security through a large number of validators. He argued that economic security is meaningless without a foundation of robust physical security. Yakovenko drew parallels to the existing internet infrastructure, highlighting how the physical layer's reliability is crucial for the entire system's stability.

He explained the concept of network partitions and how a larger set of validators makes it easier to guarantee that honest nodes can always reach each other, even in the face of potential disruptions. Yakovenko stressed that the ability to resolve partitions quickly, often through human intervention and communication, is a key aspect of network security.

The Role of Institutions and Self-Verification

Yakovenko introduced an interesting perspective on how large institutions might interact with the network. He suggested that as the network grows and handles more value, institutions like banks might choose to run their own infrastructure for self-verification. This could involve running a large number of nodes (potentially 100 or more) to independently verify the network's state and detect any inconsistencies or attacks.

This approach would allow institutions to trust their own infrastructure rather than relying solely on the public network. Yakovenko argued that this model of institutional participation could contribute to the overall security and reliability of the network, even if the number of block-producing validators remains relatively small.

Facts + Figures

• A validator with 200,000 SOL delegated stake might roughly break even on transaction fees after voting costs
• A validator with 2 million SOL active stake could potentially share about 390 SOL per month with stakeholders
• The largest expense for validators currently is the voting cost, exceeding hardware costs
• The monthly operating cost for a validator is estimated to be between 20 and 30 SOL
• Network partitions in major internet infrastructure typically get resolved in about 4 hours
• Tendermint networks typically have up to 160 geographically distributed nodes
• The Cosmos network operates with about 20 validators and has never experienced a double-spend attack
• Anatoly Yakovenko suggested that a network might need between 500 and 1500 validators for optimal global coverage

Top quotes

• "Future of Solana with 100 validators is inevitable and we should embrace it." - Brian Long
• "We're building the world's fastest globally distributed state machine." - Brian Long
• "Economic security is a meme." - Anatoly Yakovenko
• "We need to maximize that security to have any sense of economic security on top of it." - Anatoly Yakovenko
• "The larger validators simply have more fee revenue available to share with their stakeholders." - Brian Long
• "Web three will feel like web two." - Brian Long
• "If there's no partitions, the thing that you're dealing with, you cannot test it and show that it's not decentralized." - Anatoly Yakovenko
• "The bigger the set of nodes, the easier it is to guarantee that honest nodes, as a minority of that set, can always have a minimum spanning tree that can always reach each other." - Anatoly Yakovenko

Questions Answered

What is the main argument for having only 100 validators on Solana?

Brian Long argues that economic factors will naturally lead to a concentration of validators. Larger validators can offer better rewards to stakeholders due to economies of scale, making it difficult for smaller validators to compete. This concentration could lead to better-funded validators that can invest in advanced infrastructure, potentially improving network performance and global distribution.

How does Anatoly Yakovenko argue against having only 100 validators?

Yakovenko contends that a larger number of validators is crucial for maximizing physical security, which he sees as the foundation for economic security. He argues that more validators make it easier to ensure network connectivity and resist partitions. Yakovenko draws parallels to internet infrastructure, emphasizing the importance of widespread physical presence and the ability to quickly resolve any issues.

How might large institutions interact with the Solana network in the future?

According to Yakovenko, as the network grows and handles more value, large institutions like banks might choose to run their own infrastructure for self-verification. This could involve operating a large number of nodes (potentially 100 or more) to independently verify the network's state and detect any inconsistencies or attacks. This approach would allow institutions to trust their own infrastructure rather than relying solely on the public network.

What are the potential performance improvements with a smaller set of well-funded validators?

Brian Long suggests that a smaller set of financially robust validators could invest in advanced hardware setups, including FPGA network congestion hardware and multi-GPU systems. This could lead to significant performance improvements, such as 90% of transactions landing in the same slot and confirmation times of less than one second. The overall user experience could become smoother, making "Web3 feel like Web2."

How does the cost of running a validator impact the debate?

The cost of running a validator, particularly the voting cost, is a significant factor in the debate. Brian Long points out that smaller validators struggle to cover their operating costs, which can be between 20 and 30 SOL per month. This economic reality favors larger validators who can benefit from economies of scale. However, Yakovenko argues that as the network grows and becomes more valuable, the cost of running validators might be absorbed by institutions as part of their infrastructure costs, similar to how they currently invest in other IT infrastructure.

What is the significance of network partitions in this debate?

Network partitions are a crucial concept in the debate. Yakovenko emphasizes that the ability to avoid or quickly resolve partitions is key to maintaining both consistency and availability in the network. He argues that a larger number of validators makes it easier to ensure that honest nodes can always communicate, even in the face of potential disruptions. This resilience against partitions is seen as a critical aspect of network security and decentralization.

How does the debate address the trade-off between decentralization and performance?

The debate highlights the tension between achieving high performance and maintaining decentralization. Brian Long argues that a more concentrated set of validators could lead to better performance and user experience, while Anatoly Yakovenko emphasizes that true decentralization requires a larger number of physically distributed nodes. The discussion reveals that finding the right balance between these competing priorities is a complex challenge for blockchain networks.