Loading video...
Video Failed to Load
Quai's new consensus mechanism, Proof-of-Entropy-Minima (PoEM), is a small modification to Nakamoto Consensus that opens the door to infinite multi-chain coordination Let's learn how👇 1/
25,922 views • 3 years ago •via X (Twitter)
16 Comments

First, it's important to understand Nakamoto Consensus (NC). Nakamoto Consensus is the most common implementation of Proof-of-Work, and is used in Bitcoin. 2/

Nakamoto Consensus uses a difficulty threshold to determine whether a block has performed enough "work" to be considered valid. Importantly, any blocks that exceed this difficulty threshold are treated as equally valid. 3/

Proof-of-Entropy-Minima (PoEM) uses an entropy removal threshold that is similar to Nakamoto Consensus' difficulty threshold. Blocks that remove enough entropy to exceed this threshold are considered valid options for nodes to select. 4/

A major difference between PoEM and NC is that PoEM does not treat all blocks that exceed the entropy removal threshold as equally valid. Under PoEM, a block can remove more entropy than the threshold, and have that additional entropy removal be counted in the block weight. 5/

But why does this matter? Bitcoin seems to be operating fine, right? Yes, but sometimes an event occurs that interrupts consensus. Every ~300 Bitcoin blocks, something called an orphan block is produced. 6/

An orphan block is a valid block that is not included in the canonical blockchain, and is essentially "thrown out." These blocks occur when two valid blocks are proposed near-simultaneously, and nodes are forced to select one over the other. 7/

In Bitcoin, when two valid blocks are proposed in this manner, nodes have no preference between them. Because Nakamoto Consensus treats all blocks that exceed the difficulty threshold as equal, there is no mechanism to determine which block nodes should prefer. 8/

When this occurs, nodes will simply prefer whichever block they saw first. Oftentimes, this will lead to ~50% of the network mining one of the proposed blocks, with the remainder of the network mining the other proposed block. 9/

The only way this consensus fork can be resolved is by another block being produced on top of one of the proposed blocks. Once this next block is produced, all miners return to mine the longest/heaviest chain, orphaning the other proposed block. 10/

This contention in consensus, while a minor inconvenience for BTC miners/nodes, has massive implications for a multi-chain system like Quai. If each Quai chain had a long tail for consensus time, cross-chain coordination could only occur at the very end of this long tail. 11/

Additionally, the potential for this uncertainty to occur introduces a cap on the number of blockchains that can be coordinated. In order to coordinate an infinite number of shards/blockchains, Quai Network needs consensus to be immediate and unanimous, 100% of the time. 12/

That's where PoEM comes in. Unlike Nakamoto Consensus, which causes nodes to essentially guess which block will be selected in the case of two simultaneously-proposed blocks, PoEM allows nodes to compare all proposed blocks against each other. 13/

Under PoEM, no two blocks are ever considered "equally valid." Even if both proposed blocks exceed the entropy removal threshold, nodes are able to identify and prefer the block that removed the greater amount of entropy. 14/

This eliminates the period of uncertainty that occurs in Nakamoto Consensus when an orphan block is produced. Orphan/uncle blocks can still be produced in Quai. The difference is that in Quai, these orphan/uncle blocks have no impact on time-to-consensus. 15/

PoEM ensures that at every point of consensus, all nodes can immediately discern which block is "heaviest," and which block they should be mining on. This is a necessary prerequisite to infinite multi-chain coordination. 16/

The ideas shared about PoEM in this thread are just the tip of the iceburg. To learn more about the PoEM consensus mechanism and how it allows Quai to truly scale infinitely, check out the Quai Documentation: Thanks for reading! 17/17



