sunnya97.com

Proof of Stake Debate | Sunny Aggarwal and Christopher Goes

The debate centers on the decentralization of Cosmos Hub's proof of stake consensus, highlighting the balance between governance, stakeholder interests, and the inherent challenges of achieving true decentralization.

Summary

In this lively debate, we explored the decentralization of the Cosmos Hub's proof of stake consensus, with one side arguing that its design allows for adaptability and flexibility, empowering stakeholders to adjust governance and parameters in response to real-world conditions. The conversation touched on the challenges posed by the current distribution of voting power, questioning whether large coin holders would genuinely support greater decentralization when it might not serve their interests. I raised concerns about the inherent path dependency of proof of stake systems and the potential for centralization, while my opponent emphasized the importance of built-in controls that could mitigate risks. We also grappled with the broader implications of decentralization in economic systems, debating the efficacy of blockchain versus traditional governance and institutions. Overall, the discussion highlighted the intricate balance between fostering decentralization and ensuring the security and functionality of the network.

Key Takeaways

  • The Cosmos Hub's proof of stake consensus is experimental, built with flexibility to adapt to external changes and stakeholder interests.
  • Current distribution of stake in the Cosmos network shows promising decentralization, but concerns remain about the influence of large coin holders.
  • Governance mechanisms are crucial for maintaining decentralization, allowing stakeholders to adjust rules and incentives as needed.
  • Instead of forcing decentralization, it's important to design systems that are secure and resilient against potential centralization threats.
  • The complexity of human institutions compared to coded systems highlights the potential of blockchain to create more accurate incentive structures.

Detailed Analysis

In this lively debate about the decentralization of the Cosmos Hub's proof of stake consensus, we delved into the complexities of governance, stakeholder incentives, and the potential pitfalls of centralization. I took the affirmative stance, emphasizing how the system is designed to adapt and respond flexibly to the unpredictable nature of its participants. My opponent, Chris, raised significant concerns about how power dynamics might skew the incentives in favor of large stakeholders, questioning whether a truly decentralized system could emerge under the current structure. This back-and-forth illuminated not just the technical aspects of the Cosmos Hub but also the philosophical underpinnings of decentralization itself.

One of the central themes that emerged was the inherent tension between decentralization and the economic realities of power laws. Chris argued that we might be fighting against a fundamental aspect of economics, suggesting that centralization could be an unavoidable outcome. This perspective resonates with broader trends in the cryptocurrency space, where we've often seen wealth concentration leading to governance issues. The challenge of ensuring that power doesn't consolidate in the hands of a few remains a significant hurdle not just for Cosmos, but for all blockchain projects aiming for decentralization.

The implications of our discussion are profound. If we can’t address the potential for centralization, we risk undermining the very ideals that blockchain technology promises. My argument was rooted in the idea that a well-designed governance system could empower stakeholders to act in the network's long-term interest, thus fostering a more decentralized environment. However, Chris's critique highlights a genuine concern: that those with the most substantial stakes may not have incentives to promote decentralization if it threatens their control.

A strength of the debate was the diversity of viewpoints, illustrating the complexity of the issues at hand. However, a limitation lies in the somewhat idealistic assumptions about stakeholder behavior. While I pointed to the built-in governance mechanisms, Chris's skepticism about the efficacy of those systems in the face of powerful interests brings a necessary realism that shouldn’t be overlooked. This dialogue underscores the need for continual refinement and adjustment of governance models to ensure they remain responsive to changing dynamics.

This video is particularly useful for blockchain developers, economists, and enthusiasts interested in governance models and decentralization. It serves as a reminder that while the technology itself may enable decentralization, the human factors at play can complicate the picture significantly. By engaging with these ideas, viewers can better appreciate the intricate balance between innovation and governance in the blockchain space, ultimately leading to more informed discussions about the future of decentralized systems.

Transcript

Speakers: A, B, C
**A** (0:00): So let's get started. The idea is like a quick. It's two minutes. So we have up first to give you an example of what it is, we have Sunny and Chris. Basically they have two minutes to go back and forth on a topic and one takes the side of an affirmative and one takes the side of a negative. And then, then we'll ask the audience, which you agree with, who won the. Who won the debate? **B** (0:33): Okay, so the first question is. **A** (0:35): So the first question is, will the Cosmos Hub's proof of stake consensus be decentralized enough? You should explain what side you're taking. **B** (0:47): Yes. So to clarify, Sunny and I are representing different sides of this debate. To give you some perspective on what we think are the best arguments, they may not necessarily be our personal opinion. Cosmos Hub proof of state consensus is the first BFT proof of state consensus that has ever launched into the wild and is necessarily experimental. But when constructing that consensus and that state machine algorithm to allocate state between validators and delegators, we have put in many ways for the system to adapt to an unpredictable external world. All we can do is write an internal system. We can write in logic and code to govern how validators vote, to govern who gets what kind of voting power, to govern who can delegate, and to govern when they can change the delegation. But we can write that logic in a way that permits a lot of flexibility on the people who actually hold the stake. The stakeholders of the system can act in its best interests, understanding the world outside in a way that we can never predict when designing it. So that's why we've built in a lot of different controls, a lot of levers that stakeholders can tweak to adjust the parameters of the system according to what they see happening. They can redelegate to different validators. They can, you know, validators who equivocate will be slashed. Possible faults to put the protocol at risk. We punish. Delegators can see this, it's transparent. We've allowed for a high amount of liquidity of stake within the systems so you can instantly redelegate your stake to another validator if you decide the validator you delegated to isn't doing a good job or is likely to put the system at risk. The system can react very quickly and everyone using the system can pay attention and ensure that this happens. Furthermore, we built in a governance mechanism because we anticipate that no matter what kind of flexibility we built into the initial protocol, it would still need to change. That governance system will enable people who hold stake, people who have the best, you know, the future value of the Cosmos Hub as their objective function will allow those people to decide what ought to be done and allow those people to change the rules and change the incentives to facilitate the decentralized network topology that will be necessary to make the Cosmos Hub a useful system. **C** (3:00): So why would a, you know, I'm just going to go into the back and forth. Okay, so you know, why would governance like, you know the right now in the system is the coin holder based governance. And it's like if they're the, if the largest coin holders are the ones who are like, you know, benefiting from centralization, I don't see why that would help them. Why they would want to increase the decentralization of the system when they profit from it. Irregular. **B** (3:28): Perhaps some people will have short term interests, but many delegators and validators will necessarily have stake locked up. They can't simply outshort their positions and vote against the best interest of the cosmos. Moreover, there's always the threat of a fork. If I submit a proposal to maybe privilege my interests of a validator over the best long term interests of the network, someone else can set up a credible threat to split the network and that will force my proposal to fail. **C** (3:57): How decentralized do you think it has to be? How much is enough? The current distribution of stake and voting power, of validation, is that fine? **B** (4:05): It's hard to tell. The current GENIE coefficient of Cosmos voting power is about 0.79, which is already less. Bitcoin mining pools are about half, but the Ethereum mining pools are 0.85. So the Cosmos Hub is already more decentralized. **C** (4:22): Yeah, but like the Bitcoin mining pool is a way bigger industry than Ethereum mining and like, you know, that's. Why would they use it? Because partially because they use ASICs on. **B** (4:31): My Bitcoin mining pools have had years to become decentralized. Yeah, millions and millions of people have tried their hand at Bitcoin. **C** (4:37): Okay, but that's because a thousand people proof of work allows new people to enter the system. Proof of stake is inherently a state dependent like a path dependent system. It's all any future state will have to have been gone through through the initial distribution. What evidence do you have that the current state atom holders with like the large whales are actually going to exit their positions and like help decentralize them now? But the problem right now isn't like, you know, we can't solve it with tiny little things like you know, oh, making easier for redelegation or Whatnot, because currently the largest atom holders, like, you know, polychain, Doki, etc, most of their estate is from self bond, these large item holders that together through their self bonds already make up 33% attack. **B** (5:21): It's silly to draw conclusion from what the state distribution has done in a week when transfers are disabled and you have to burn nine atoms just to. **C** (5:26): Sell Dokia, are you going to sell any of your atoms as soon as transfers are enabled? **B** (5:30): No. **C** (5:30): Okay. They're the second largest animal though. **B** (5:36): Dokia sends pictures every morning on Twitter with eggs baked in the shape of atoms. That devotion is admirable, but it's not necessarily representative of the rational interests of. **C** (5:48): You think policy is going to sell? **B** (5:50): Absolutely. **C** (5:51): Okay, let's see. **B** (5:54): What. And even, even if they don't, this is why we've built other controls in the system. This is why the community fund receives 2% of inflation fees and can be directed by validators. Because the stakeholders of the network, who are not just the validators, the delegators who want Cosmos to be valuable in the long term, can elect to make it easier for new people to join. It's true that if a cartel of large stakers emerges and they simply maintain control of the network, it won't be decentralized enough. But that won't happen because if someone sees that they will redelegate, they will threaten to fork. They will force the stake district to act in the long term. **C** (6:30): So I guess essentially my point is we should be focusing less on like forcing decentralization and more on learning how to like. No, I think the universe has power laws that we're never going to be able to find fight. Instead we should be learning on how to like, focusing on how to make systems that are like secure in the face of centralization. You know, if you have a cartel of three ballads, how do we make it so like they should distrust each other and make it like. So it's hard for them to cartelize. How do we make it so it's like extremely unprofitable for even a large whale to like, you know, create a single valid with as much stage as possible. So I think we gotta focus in the mindset of like, we accept centralization is the reality of the world and instead we have to learn how to like, put checks and balances in the system rather than trying to force the decentralization of the state distribution. **B** (7:15): You give up too early. Power laws are not a fundamental property of physics. Last I checked, they're not part of the standard model, they're part of quantum field theory. They're a contingent fact about how the universe of, say, validators has been organized in many blockchains so far. **C** (7:30): But that doesn't mean it's not a fundamental pout thing of the fundamental thing of economics. **B** (7:36): Right, but we have so much more design space. It's true that proof of work has tended to be more decentralized maybe than a database. Your arguments would apply equally to writing everything in SQL instead. **C** (7:48): Okay, what fundamentally is a proof of stake blockchain giving us different than like, you know, we've had democracy for hundreds of years. We've had like, you know, digital databases for hundreds of years. Like, no. Why, if anything, we see the world distribution of capital in the non crypto world centralizing. Why do you think people, how do you think suddenly putting things on a blockchain is going to make it any better? **B** (8:11): Because the human institutions are fundamentally limited in the complexity and the accuracy of the incentive systems they can implement relative to code. Because humans have transactions. Humans are corrupt. **C** (8:22): What do you mean? We have code. You don't need a blockchain to run code. **A** (8:27): Okay, hands up. **C** (8:54): And that has not solved capital distribution. **A** (8:56): Okay, so if based on that, would you take Chris's side or Sunny, who would take Chris's side? **B** (9:08): Come on. **A** (9:16): Okay, we don't have to keep going if we want to do the champagne and cake, but if we want to keep going, we can.