# Consensus definitions

###### tags: `thread`

In this thread, I want to clarify my misunderstanding of consensus mechanisms and clarify their definitions

- PoW, PoS, and PoA are not consensus protocols
- PoW, PoS, and PoA are essentially Sybil resistance mechanisms
- Consensus protocol and consensus mechanism are not equal
- For PoW-based Ethereum, Nakamoto is its consensus protocol
- And for PoS-based Ethereum, Gasper is its consensus protocol

## Proof-of-work and Proof-of-stake

Proof of work and Proof of stake are not consensus protocols in themselves, people calling them consensus protocols is a lazy way of saying

They are essentially Sybil resistance mechanisms, and understanding this requires first understanding "what is a Sybil attack"

"Sybil attack" refers to an attack by creating vast numbers of duplicate identities at a low (or zero) cost

For blockchain, we need a majority of entities (generally 51%) to agree on a blockchain history

But if there is a Sybil (an attacker who creates a vast number of entities at low cost), this attacker will become the majority (with more than 51% entities), which eventually overwhelms the entire network


## Anti-Sybil mechanism

There are two ways to defend the Sybil attack:

- One is to build a permissioned network that allows only a few participants to enter (only they have voting rights, e.g. pBFT)
- Another way is to build a permissionless network (allowing everyone to enter), but with some limited resources to prove the voting rights of the participants

The second is the way Bitcoin and Ethereum use, PoW uses computing power (a limited resource) to verify the voting rights of participants (the cost of creating duplicate identities for an attacker is huge)

PoS verifies the voting rights of participants through money (chain’s coin, also a limited resource), and the cost of creating duplicate identities is greater than PoW (PoW can rent computing power, PoS must buy coins in the market)

So PoW and PoS are actually Sybil resistance mechanisms

## Consensus mechanism

In addition, the consensus protocol ≠ the consensus mechanism

- The consensus protocol is a set of algorithms that can be defined at the protocol layer (in code)
- The consensus mechanism is a combination of a set of components, both protocol-level and non-protocol-level

For PoW-based Ethereum, Nakamoto is its consensus protocol

For PoS-based Ethereum, Gasper is its consensus protocol (i.e. a combination of finality tool “Casper FFG” and fork-choice rule “LMD-GHOST”)

For the consensus mechanism of Ethereum, when attacked by 67%, the hard fork through the social layer (and legitimacy of the Ethereum community), are all part of the consensus mechanism, but cannot be defined at the protocol layer

## Conclusions

Anyways, the consensus is really an interesting thing, and it is these studies that make me see that it’s different from any permissioned network in history

For example, religious networks, nation-states, and military, these networks surround a core authority figure (the Pope, head of government), and restrict access to all

The closer you get to the core, the more you get, and these core characters monopolize all key interests

As a drawback, these networks limit innovation. Just look at the impact of the market economy on human society (fifty years from the biplane to the Saturn V), it is a relatively permissionless network

[Thread](https://twitter.com/LuozhuZhang/status/1577954739613941762?s=20&t=hf32T6kwnUQiTjAV1MfiLA)