What is the downside of not using Proof of Work for validator selection?

4

Bitcoin uses Proof of Work for two things: one is to secure the blockchain with strong cryptographic signatures like a Merkle tree, and the other is to crown the validator which "mines" a solution as THE validator (dictator) who determines which transactions to broadcast in that block. If they blacklist your address (or charge large fees) and don't want to accept your transaction then maybe another validator will eventually accept it.

But why do we need Proof of Work for the second thing? What's wrong with Proof of Correctness as Ripple uses, or Delegated Proof of Stake as LISK uses, for example? Why are they considered "not decentralized" while Bitcoin is said to be "decentralized" despite the arms race that has caused a few mining pools in China to control more than 51% of the mining power in practice?

What is wrong with a community electing 101 random validators like with LISK for example? How can this be gamed to reduce people's freedom to transact or have their transactions confirmed? The worst thing is some address may be blacklisted by a few validators (eg under pressure from a government law) but the same can happen with bitcoin miners, since we know who they are. So what's the disadvantage of Delegated Proof of Stake?

Here are the advantages: less waste of electricity, faster consensus, and possibly more decentralization in practice.

If I am right and this doesn't have any serious disadvantages, it should be rather trivial to make a cryptocurrency with delegated proof-of-stake electing N (odd number) validators.

Gregory Magarshak

Posted 2017-09-06T23:45:09.217

Reputation: 103

Answers

2

Proof of Work is needed to inventivize miners to not fork and, if there is a fork, some metric for determine which fork is the real blockchain (assuming no consensus changes in the fork).

Proof of Work incentivizes miners to not fork (i.e. see someone else's block, ignore it, and create their own block at that height) because of cost. It costs resources in order to produce a fork and stick with it. If their fork has no value, then they are wasting money on something that is worthless. However with other consensus systems like Proof of Stake, there is no cost to make a fork. Making a fork does not cost anything as there are no resources for you to spend to make the fork. In some systems, you can make a fork and still be mining/staking/minting/whatever on the original chain at no extra cost and no opportunity cost.

Proof of Work also provides a metric to determine which chain is the correct chain. Given two blockchains with the same consensus rules, the one with the most work (the sum of the difficulties of all of its blocks) is the one that is considered "the blockchain". In other consensus systems like Proof of Stake, no such metric exists for determining "the blockchain". This results in centralization as someone or something is needed to say "this blockchain is the right one". In some coins, a signer (usually the developer) must sign all blocks for the block to be considered valid, so that signer determines what "the blockchain" is and is a point of centralization.

Andrew Chow

Posted 2017-09-06T23:45:09.217

Reputation: 40 910

Sorry, I fixed up the wording in my question. I didn't mean the miner is the ONLY validator, just that it is the validator that chooses which transactions to include in the broadcast of that block! Please, can you update your answer to reflect my change?Gregory Magarshak 2017-09-13T03:30:01.033

Edited. Your change does not really effect my answer.Andrew Chow 2017-09-13T03:33:16.733

So with LISK for example you have 101 validators that others vote for, delegated proof of stake. Are you saying that this system is more susceptible to forking than Bitcoin? In practice, after the arms race Bitcoin has only a few mining pools that control things and so a fork has already been done - Bitcoin cash. And more to come.Gregory Magarshak 2017-09-13T05:21:09.297

Yes, LISK and other coins using some sort of Proof of Stake are more susceptible to malicious forking. Whoever is the one to mint/forge a new block (e.g. in LISK the current delegate) can create two blocks and thus create a fork. However that person can continue to mint/forge on both forks of the blockchain. In Bitcoin with PoW, that is impossible. You can either point your hardware and mine on one fork. There is no way to mine both simultaneously as you can with PoS systems.Andrew Chow 2017-09-14T04:42:08.513

If one of the 101 delegates creates a fork, how will they broadcast it to the others? If they broadcast different things to different people then their broadcast will simply be IGNORED in the consensus protocol.Gregory Magarshak 2017-09-14T20:20:37.930

How do nodes know to ignore the different blocks broadcast? If I give you block A and give Bob block B, how do you know that I gave Bob block B and vice versa? According to the DPoS system described by bitshares (and cited by lisk), in the event that a fork occurs, witnesses will just forge blocks on top of the most recent one that they have. There is nothing saying that conflicting blocks will be dropped. If list added that, then they have done a horrible job of documenting it (I looked at their docs and they are horribly unspecific).Andrew Chow 2017-09-15T18:10:50.523

You may be right about LISK. But I am still not sure how the nodes can introduce a fork, if after every block the nodes simply merge everything they saw from everyone. If node A broadcasts different things to different nodes, they will eventually compare notes and realize that A said different things (A signs all its broadcasts) and simply ignore A's broadcasts from then on. I think that's what Ripple does (one of its features).Gregory Magarshak 2017-09-17T05:05:00.300

Here's the deal @Andrew Chow .... the 30% defector thing only applies in Byzantine Consensus if messages aren't being signed.Gregory Magarshak 2017-09-19T18:06:16.360