Vitalik Buterin calls Ouroboros an “asynchronously observable synchronous consensus” 

Vitalik Buterin calls Ouroboros an “asynchronously observable synchronous consensus” 

The altercation between Ethereum’s Co-founder Vitalik Buterin and Cardano’s Charles Hoskinson has been going on for quite some time now. The rebuttal took off after Buterin accused Cardano’s Ouroboros protocol last month. On 16th August, Buterin responded to Hoskinson’s latest rebuttal confirming that the debate is still on.

Buterin responded to two main points made by Charles Hoskinson in his previous rebuttal mentioned below:

“Synchronous protocols are typified by a round-based structure, an explicit hard-coded upper bound on network delay.” “Such a protocol cannot tolerate that any single honest player suddenly experiences network delays beyond the hard-coded bound.”

While Buterin agrees to Hoskinson’s first point, he disagrees with the second one. He said:

“In any 50% fault tolerant synchronous protocol, if 20% of nodes start to experience delays, then you still have 30% fault tolerance.”

Buterin also questions Hoskinson if he meant the following:

“if honest player H gets a high network delay, H themselves cannot get a reliable view of the consensus anymore. “

Buterin further argues that:

“This is definitely true for the “99% fault tolerant” type of algorithms, though not true for the “50% fault tolerant” type, including chain-based PoS, and even a simple overlay where at the end of each round each active participant publishes what they think the result of the consensus is and everyone adopts the majority result.”

Buterin believes that the above is not similar to a partially synchronous consensus. The partially synchronous consensus continues to work even if all nodes’ latency goes arbitrarily high, “as long as at some point it stops growing,” he said.

Buterin further accuses IOHK of making a mistake in their article mentioning that Ouroboros and the Casper are similar in certain ways. He said:

“So I think the IOHK blog article makes a mistake in implying that the finality guarantees of Ouroboros and the Casper family are similar, as the former is latency-dependent whereas the latter has been proven safe under asynchrony.”

Moreover, Buterin cited a part of the IOHK blog article arguing that it does not offer any security guarantees when Δ >= 0.693 * f:

“Ouroboros is analysed in the “partially synchronous” setting where messages are delivered to the majority of the parties executing the protocol within a time window upper bounded by a network delay Δ which is unknown to the parties.”

Buterin concluded his post by agreeing to Hoskinson. He stated that Ouroboros is stronger than the Byzantine consensus setting where all honest nodes need to have a bound on network latency.

However, he also said that it is definitely weaker than partial synchrony. Buterin cited examples of partial synchrony defined by Dwork, Lynch, Stockmeyer. Lastly calling Ouroboros an “asynchronously observable synchronous consensus.”

Share your thoughts, add a comment!

You must be logged in in order to place a comment.

Article comments

Loading...
No comments yet, be the first to comment this article