PoS Thought

1/ Coinbase adds automatic staking for XTZ, and less than a month later, Binance introduces 0% staking fees on Tezos. Stake wars are just getting started, and things are already starting to feel a liiittle uncomfortable. A thread on centralization in Proof of Stake (PoS):

2/ Running a validator well is hard & requires technical expertise — it’s definitely not for everyone, so we let users delegate their stake to other people running validators. But this is causing stake to gravitate toward a small number of publicly-known validators.

3/ When stake becomes concentrated, collusion becomes a real risk. “Trustless” networks are replaced with networks that only trust a few. True censorship-resistance arguably goes away.

4/ So why is stake concentrating? A big part of this is the inefficiency in delegation markets. On paper, validators look the same, so delegators crowd into a few validators with a large amount of stake, decent track record, and low fees. Power laws take over.

5/ Validators have 3 ways to differentiate: 1. Pre-existing reputation 2. Fees — offer higher return to delegators 3. Differentiated services (i.e. offer better staking experience or additional crypto services)

6/ Even in an efficient validator market, stake concentrates to a few large validators who utilize reputation and scale to drive smaller validators out of the market. Binance using unsustainably low fees to undercut the market is the latest example of this.

7/ What does this mean for PoS security? Well, we can’t rely on honest & uncoordinated majority assumptions to drive security — collusion must be assumed to be possible and probable.

8/ To have true security, the economics have to check out for any single entity: profit(honestValidation) > profit(attack)

9/ In the context of PoS, this means slashing must be used to impose a cost on misbehavior. Without slashing, validators have little “commitment” to the network. Networks w lower slashing may tend to be less secure, and it seems unlikely that a slashless PoS could be secure.

10/ What can be done to combat centralization in PoS? A) Social consensus around min. staking fee — ‘require’ validators charge a sustainable fee. B) Anti-correlation penalties — slashing that rises in proportion to the amount of stake that misbehaves in the same period.

11/ C) Incentivize delegating to smaller validators — ex. Polkadot’s Nominated PoS (NPoS) that distributes rewards to validators evenly regardless of stake. This helps small validators bootstrap delegation.

12/ These mechanisms may help discourage centralization but ultimately won’t stop it. Validators will find ways around these rules. (e.g. running multiple nodes, ignoring out-of-protocol ‘rules’)

13/ It’s looking increasingly like centralization in staking may be inevitable. If this is the case, bribery for chain reorganization (i.e. Binance hack) & censorship may always be within the realm of possibility.

14/ There may only be one way to truly secure these networks long-term — Zero-knowledge Proofs. Rather than relying on economic mechanisms to keep validators honest, ZKPs make it impossible for validators to perform malicious behaviors like producing an invalid block.

15/ With ZKPs, centralization in staking networks no longer becomes a huge issue, as the value a validator can extract from an attack is decreased significantly.

16/ Networks currently trust in economics; in the future, networks will trust in cryptography. Cryptoeconomics might be just be a bandage, a short term fix to help bootstrap the crypto ecosystem until cryptography can provide real security.

微信 WeChat 新浪微博 Twitter Telegram 搜索 link clock