It's unclear whether or not the alternative 'proof of stake' technique will be more efficient than proof of work. It is susceptible to abuse, and is likely to require an arms race of CPU power to prevent attackers from abusing it.
It's definitely more efficient than proof-of-work, since no energy is burned off with the only purpose being to calculate a proof of this.
The problem is that it makes consensus harder to reach, since there's no problem with having 10 different chains, because no energy needs to be burned off for each chain. This, effectively, makes your account balance a democratic decision by nodes in the Ethereum network (as opposed to forcing someone to burn off large amounts of energy if they want to alter your balance).
History being decided by democratic vote is exactly what proof-of-work avoids, or at least makes extremely expensive.
That's a bit early to say for certain. Nobody has made PoS work in an adversarial environment yet.
It is misleading to describe PoS as a democratic vote. There is no voting as that could be faked.
If you want an allegory maybe jury duty for the richest. That also captures the basic problem with it, when the chosen few collude. Let's just say it is an active research area.
> It is misleading to describe PoS as a democratic vote. There is no voting as that could be faked.
I disagree. With PoS there is, ultimate, only the democratic vote.
For example, in a network with 1,000 nodes, let's say 95% follow one chain and ignore another chain, and that the remaining 5% follow that other chain. The two chains are completely different, but are both valid. In a PoS-system there would be no way, besides democratic vote, to determine which chain is the right one. You're forced to either ask someone you know which chain he's using, or just go with the 95%-chain. With Bitcoin, the answer is very simple: you follow the chain with the largest amount of work, regardless of how many nodes say the one with less work is the right one.
Bitcoin is inherently non-democratic in this scenario, as all nodes only care about chain work, and not how many nodes follow which chain. As opposed to PoS, where two competing, valid chains only differ in which (trusted/untrusted?) -- and how many -- nodes follow each chain.
What you describe should not be valid in any well working blockchain, not sustainably. Guaranteed global consensus is the whole point, and a simple majority vote would be useless for obvious reasons. I'm not sure if you have something specific in mind, but what you write is not true in the general case.
In a POW network with 1,000 miners (with equal hash power), let's say 95% follow one chain and ignore another chain, and that the remaining 5% follow that other chain. The two chains are completely different, but are both valid.
The second chain is just an alt coin resulting of a fork.
POW is as much a democratic decision by your description. If a group of miners decide to reach a consensus you have the sames results. The real difference is that the right to vote in the system in POW is determined by how many electricity you spend, in POS is determined by how many currency you own.
CPU power doesn't come into it. More ETH = higher returns.
I haven't seen any specific details for Ethereum bonded validators, but the hardware required is likely to be a cheap VPS with a reliable internet connection, at least judging by other active proof of stake projects.
Ethereum wants a PoS that works, so they will need to do something different. We don't know what that will look like.
Other projects using PoS survive by doing things like frequent block chain checkpoints (a notary stamped block if you will), which arguably isn't very decentralized or at least not trustless. If you have a notary anyway there are more efficient ways to go about things.
