This is a revised transcript of the "The Future of Proof-of-Stake" discussion with Brian Crain, CEO of Chorus One in our community Telegram channel. Topics discussed include Brian's background and why he decided to focus on PoS, the concept of decentralization and how different protocols approach it, the connection between staking yields and lending protocols, the role of validators in governance, as well as stake diversification and validator economics:
Brian, how did your involvement in Proof-of-Stake start? What made you found a company focused on this part of the ecosystem?
I studied economics as an undergrad and did some graduate school focusing on economics and game theory too. While I immediately found Bitcoin amazing when I learned about it, it's security assumptions struck me as unsound. So PoS made lots of sense to me even when I read about the early work like Peercoin or Vitalik's article on PoS ideas for Ethereum.
In 2015 I joined a company called Monax (before Eris Industries). They were the first company to build on Tendermint. They built enterprise Ethereum applications. I became skeptical of the timeline of enterprise use cases, but when I left late 2016, I was very familiar with Tendermint and the Cosmos vision so I joined them.
I would love to get a kind of compare/contrast b/w what I see as the most interesting flavors of PoS: true pBFT-style DPoS like Cosmos (where the delegators can be penalized for misconduct by the delegates) and Shasper. Pros/cons of each. And whether you think Shasper will ultimately end up looking a lot like DPoS because many people will use third-party staking services instead of staking directly.
In my view, the biggest contrast in PoS approaches is the following:
- We have some networks / approaches that embrace validators as having a key role in controlling the networks. In governance and economic power. The Cosmos model is probably the dominant one there.
- There is another approach that tries to have "maximum decentralization" which is often expressed by having a huge number of nodes. There may also be attempts to minimize the power of validators in other ways.
We don't have a good clue about how either direction will work out. It's just so early and there is little data so far. But I do think the networks optimizing for the number of validators often disregard the natural dynamics of markets. As Gabe points out, just because you have 1000 nodes, doesn't mean a network is decentralized. If in the end, you have 3-5 providers that host the majority of them.
Another key thing: In my view the most important questions around PoS are around the economics and incentives. In a way, Tendermint has solved the consensus piece reasonably well. Maybe Solana and others will improve on the performance somewhat but Tendermint is already great. But the economics are a complete open field. And we have no idea about the best approaches. The great thing is that there are many experiments. Cosmos, Celo, Polkadot, etc. are all radically different.
PoS with delegation is a really interesting quickfix to some of the concerns in previous PoS iterations and consensus in general, however do you guys think this is PoS endgame?
I think PoS with delegation is probably here to stay. It just makes sense that you use cryptoassets as collateral to ensure security. And it also makes sense that as an owner of cryptoassets you want to have someone else do that work on your behalf. So I think PoS & delegation will still be prevalent >5 years from now.
What about Avalanche, they claim to build governance extending as much as deciding critical network parameters which directly changes them without a hardfork, and by permisionless membership.
The automated network upgrades isn't such a novel thing. We have Tezos and Polkadot doing that already. I think it makes sense on balance. The Cosmos hard-fork approach seems a bit more cumbersome.
And governance being able to change network parameters makes sense too. You have that in Cosmos too. I'm not actually familiar with the details of Ava's governance though.
Ava's claims are the absolute dream of decentralization, scalability, performance and governance. If you know the team do you think that they are capable in delivering something of this scale?
Yes, I know the team quite well. Emin Gün Sirer is fantastic and I'm a big fan of him. I had dinner with him twice in Seoul in July. He's come up with many amazing ideas like Bitcoin-NG, selfish mining & others. Kevin seems very sharp too.
I have high expectations for Ava. We have a podcast episode with Meher and Kevin from Ava coming out shortly, which might provide some more insights!
It's interesting that Polkadot's governance power will be unbundled from validation security. Thoughts on that?
Yes, it's an interesting approach. I don't have strong opinions on whether that is better or worse than the Cosmos approach. I'm excited both experiments will exist.
The benefit of Cosmos is that the validators already engage with the network and are often well suited to make the right governance decisions. Hard to say how Polkadot's governance will play out.
I've seen emerging views that financial use-cases will compete with staking rewards. Thoughts on that?
Sure, that's true. If you can stake assets and make 8%, lending rates will have to compete with that.
And vice versa! Why stake if you can make better returns with financial products?
Absolutely, but it's not a big issue for PoS networks IMV. E.g. in Cosmos if the staking ratio falls lower, the staking rewards increase. This is an easy way that PoS networks can ensure enough is staked so they are secure. Since the networks control the money supply, they can always make sure returns are attractive enough vs other use cases so security is maintained.
We saw that most (75%) of Atom holders don’t diversify and only delegate to a single validator. Do you think diversifying is important and should be encouraged?
Good question! On some level, I think it's the problem of the individual. Let's say everyone only staked with only a single validator. If those were mostly different, you could still end up with a broad distribution. From that PoV, it's just a question of risk diversification of the individual. But, of course, if people distribute their stake (particularly among smaller validators) it will increase the decentralization of the network. So maybe it makes sense to try to encourage that.
But there are tradeoffs, e.g. imagine if we had 200 validators on Cosmos and everyone had just 0.5% market share. Does anybody have enough at stake to actually invest and spend time on the network? We would certainly have to shift attention elsewhere just from a business perspective. I don't think "more decentralization" is the right thing to aim for. A better target would be: "sufficient decentralization and more rewards for those who contribute the most to the network's success". Not obvious how we can accomplish that, but I think much of the discussion around increasing decentralization is somewhat misguided and misses the point.
Imagine on Cosmos instead of the current voting power distribution, all validators had 1% of stake delegated to them. Would the network be massively improved? Would things work better? Would it be more secure?
I'm not convinced that any of these would be true. So this striving for wider stake distribution seems fairly pointless to me.
What solutions do you think could encourage that?
I saw Jesse Livermore propose that a staker could earn more if they delegate to multiple validators. So that would be a simple way of doing so. But I feel mixed about it. It could obviously be gamed too, e.g. we could spin up Chorus One, Chorus Two, etc. Then people could split across them, earn more while really still only stake with one party.
So I'm not sure I'd be supportive of Jesse's idea. But haven't thought too deeply about it yet.
My experience so far is that delegators are not sufficiently aware of the importance of decentralization and it feels like the current incentives lead to less decentralization (commission fees war).
Yes, good point. I definitely think that the price competition is a mixed thing and on balance probably a negative for the chance of Cosmos' succeeding.
This is the key point. I think the answer here lies in a robust dPoS but an additional mechanism that links a reward to those engaged and aligned to the long term success of a chain.
Excellent point. Yes, maybe these just have to be separated. One could have part of the issuance go to people building the community and decouple that from commission. This could be a viable approach. Celo is doing something like that. We have a podcast with them coming out shortly! It's pretty cool. They do a lot of novel things when it comes to staking incentives and mechanisms.
How do you think about network security? Do you feel the threat of slashing is enough to encourage validators to build secure infrastructure and for delegators to delegate to secure validators? How much focus should a network put on secure validation infrastructure?
I think slashing penalties are totally sufficient for that. There is no need for the network top-down trying to enforce some ideas about security or hardware being used.
I've seen some networks trying to mandate server capacity etc. That makes no sense to me.
On the subject, the financial incentive for validators you mentioned. It will certainly lower the quality of their service and time spent on delivering excellence if they are less incentivized, but doesn't this also mean that theoretically they would be easier to corrupt, which could be the end of any decentralized ledger sense?
Good point. I think if you have 100 different validator setups, even if they are individually less secure, corrupting a majority will be extremely hard.
So from a security perspective, if you have many validators AND they are actually run by different organizations, it could be very secure even if they individually are amateur.More than about network security, I'm worried that weaker financial incentives for validators will create a more passive community. There could be less content, less governance involvement, less community building, etc.
I have to wrap up now. Thanks so much for stopping by!
You can find podcast episodes mentioned during this discussion within the coming weeks on your favorite podcasting platform by subscribing to the "Chorus One Podcast" (Links to episodes can be found at: https://chorusone.libsyn.com/).
We did a small poll during the this session to find out whether participants staked tokens, which a majority (>70%) did. Feel free to join our Telegram channel to participate in upcoming Q&A sessions with our podcast guests!