I like the idea of a xDai chain with a possibility to stake tokens. It is definitely a new concept and avoids the problem of a native coin for every chain. I am already looking forward to see if this is an appealing concept for validators.
General: It is not clear to me what the target audience is for the paper. It seems that the problem this chain is solving isn’t that clear, or it could be made clearer in the introduction.
General: The reference implementation starting on page 35 feels like a separate paper and is more of a repetition. IMO it should be split in two papers.
General: In contrast to POA Network, there seems to be no identity at risk for the validators. Is this not at all planned or just not for the POSDAO chain?
MAX_VALIDATORS … what do you expect to be the upper limit?
4.3: “report on malicious validators” … is that done manually by every validator or is there some form of automation?
4.4. It would be interesting to have a description / reference how the selections process for active candidate pools works.
4.5. Does the withdrawal of tokens during an active validator epoch have any effect on the payout at the end of the epoch?
5.1.3. “the larger the candidate’s pool,” … doesn’t this lead directly to a few very large candidate pools?
5.2. Is there no use for the current Governance DApp’s in POA Network?
5.2. Initialization parameters - “Number of Validators” … This gives the impression as if it is set in stone at the start of the network.
6.1.1. “Varying block rates based on network performance” … that is basically clear, but shouldn’t there be a minimum block time and in case of long network latency, it just increases until the sufficient amount of signatures is collected?
6.1.2. I would not call them “elected validators” … it is rather “randomly selected validators” - right?
6.2.2. “validator pool” is not defined and caused quite some misunderstanding for me. Maybe a “validator syndicate”?
6.2.3. How can a delegator find the candidate pools and select one? Is there the intention to measure “pool” reputation and generate some rating?
6.2.4. I didn’t see any mentioning of an additional pool for additional funds filled by inflation and voted upon, like we have now in POA and ARTIS. If the staking token has value, it might be possible to get something into a pool for e.g. development work.
7.5.1. How is an “active” pool defined? I assume you mean “active candidate pool”.
“banned pool” --> “banned candidate pool”.
“Running nodes in the cloud is not recommended (for security reasons)”, what do you mean with that?
7.5.2. “known individuals” - I assume this means that there are Terms of Service, which specify, who is operating the bridge.
7.5.3. “extremely expensive” is kind of unclear. What does this mean?
8.4. How can locked funds, not owned by the bridge operator, be used as collateral for Compound? It is like I use someone else’s house as a collateral for a loan
9.4. Constants … can they be adjusted through a hard fork / voting or not at all?
Summary: I could imagine that a clear target group should help to get a better directed paper. The implementation part is much clearer for me. The selection of constants for POSDAO xDai chain is not explained, where it is derived from. In comparison to POA it has lost some nice features like the additional pool, the identity part (I understand that is deliberate) and the voting mechanism. More critical for me is that there seems to me a high danger of concentration in a few massive pools.