Vitalik: The benefits of complete decentralization and "limited" centralization, the blockchain can do both

Ethereum founder Vitalik Buterin recently shared his views on decentralization and centralization on Twitter. Complete decentralization has a high value, but "limited" centralization is also beneficial. Vitalik believes that blockchain can do both.

Here is what he shared on Twitter:

"Yesterday I talked about the benefits of complete decentralization. Today, I will discuss the benefits of deliberately designed" limited "centralization.

The best argument for supporting centralization is usually about the incentive mechanism of centralization: suppose that each of the 1 million people has only a small interest in the result of a certain thing, and the ability to influence the result is very small. Then they will not care about their own influence (or worse, they will be easily guided by the incentives attached to others) … However, if a person ’s rights and influence in a certain result are great, they will need Time to learn, optimize and use economies of scale to make + execute good decisions.

The best arguments against centralization are the vulnerability of centralization, the risk of abuse of power, the lower "bus factor" and so on.

Can we combine the benefits of both (decentralized and centralized)? Well, the blockchain has already done it. Consider the production mechanism of a block: Basically, every 15 seconds we randomly select a "dictator" who completely controls the transactions (tx) contained in a block. From an economic perspective, he is a block reward "Completely marginal requester".

This architecture actually has many benefits, and is the most resilient of all the bribery / "secret payment" architectures we know so far.

Miners and transaction senders sometimes do private transactions with each other, but the benefits are very low, so it hardly happens.

And it has some disadvantages: if you don't like a certain block producer, please wait 15 seconds for the next one.

Once there are more complex applications on the chain, we will face challenges. Optimal tx inclusion, including capturing "miner extractable value" through frontrunning ( https://arxiv.org/abs/1904.05234 ), will become complicated.

This brings risky centralization to complex block producers.

For this problem, carefully designed "limited centralization" can be solved.

If we divide the consensus into two tasks:

  1. Choose which tx to enter the block (complex, so easy to centralize)
  2. Use these tx to verify other people's blocks (routine work is therefore more decentralized)

* Kinda * in the current eth1-> eth2 roadmap has the following characteristics: eth2 validators must be explicitly registered in order to become a "eth1 friendly" block proposer.

It is proposed that the resource consumption of the block is more, and the resource consumption of the verification block is less. The block verifier does not have to be a block proposer.

If block proposers are centralized, the risk to users is much smaller than that of validators: even if 90% of the proposers are bad, the worst thing they can do is average latency 10 block time slots (~ 2 minutes) for transactions. If they propose wrong blocks, the verifier will ignore them.

Another clearer example is the miner extractable value (MEV) auction .

Basically, in rollup, the right to aggregate and select the order of transactions is auctioned once a day, such as the person with the highest bid.

In this way, the centralization work (tx selection) and the decentralization work (tx verification), which are burdensome with economies of scale, are completely separated, and income can be obtained!

This has limited risks because anyone can directly publish the TX to the chain and force it to be included in the next batch of blocks.

Therefore, a regional concentration and inclusion of economies of scale (and therefore centralization) will exhaust the economies of scale in the surrounding area (and therefore lead to more decentralization). " Next is Vitalik ’s previous share of the benefits of “complete decentralization”:

Those who say "full decentralization" is an exaggeration, and you only need to be non-custodial is not enough. A completely decentralized (or "serverless") application is valuable because it convinces users that "this application will always exist", so you can build on top of this application with peace of mind.

This is why. Projects such as Uniswap, Tornado Cash, etc. can easily inspire people to build an ecosystem around them, but a more centralized alternative fails.

If you build a dapp on a fully decentralized / serverless infrastructure, your dapp can continue to use it even if other communities around that infrastructure disappear. As the network effect is smaller, it may become weaker, but it can still be used.