Ultimately, the community of users and businesses at large decides. Cryptocurrencies are social contracts that are encoded in software. If everyone agrees that a rule should change, and updates their software in concert, then the rule de-facto changes.
In the absence of total agreement, the decision is made by whoever wins the political fight for user/business mindshare. It's also possible for the question to be answered two different ways (a longstanding fork occurs).
That's called a chain split, isn't it? Usually there are people who follow the new protocol and people who follow the old protocol, so the currency splits into two.
Bitcoin Cash is an example.
But yeah if there's longstanding disagreement, there are two ways for it to go:
* Peaceable fork with replay protection, like Bitcoin Cash
* Fork that seriously fights for the same brand/userbase --- we haven't seen one of these in Bitcoin. Segwit2x would've worked that way, but was canceled
When you change the rules to be less restrictive (add operations, increase block size, etc) then any old version will reject the new blocks, so will keep running their own chain until they are upgraded, potentially forever if there’s demand.
I bought something with BTC after the fork, and all my BCC was emptied within minutes.
1. What if there simply isn't a quorum of the selected signers for a given block, due to people being on vacation, DoS, network failure, etc? Does the whole chain grind to a halt?
2. PoS relies on punishing anyone who signs a block that's not on the main chain. But who defines what the main chain is? If I'm a selected blockmaker or, worse, I control a quorum once, can't I forfeit my deposit to fake someone else out so they sign a block on my bad chain? Then I report both them and myself, bankrupting them and recovering some of my own losses?
3. This whole scheme penalizes cold wallet users. If you don't use your cold wallet funds as mining collateral, you lose the reward. If you do, then anyone who compromises your deposit signing key (which needs to be hot) can wipe you out and steal 1/3 of your deposit.
For your first question, if people are on vacation and there are no validators then Ethereum would be dead. It wouldnt make sense that the global pool of validators would disappear, but even so, it would wait I assume. IF half the validators left, it would choose from the other half.
2. THe main chain is currently defined as the longest defined chain fully validated. To fake someone out, you would need to tell them your vote for a block, and they would need to blindly vote the same without doing any due diligence on the block (which would defeat the purpose) and they would be rightfully burned. That is the same as a miner who just copies the work of other miners, they do nothing. 3. I dont understand this at all. You do not need to stake. You are rewarded but lose liquidity. It is your choice, just as lending compute was your choice with PoW.
Hopefully this is good, the link goes into a lot of detail that should clear things up!
How will the network reach consensus on this? Why is it assumed that a well-coordinated network of decentralized nodes will exist independently of the attacker, and that this honest network partition will be able to reach consensus on which chain to follow?
> Each time the attacker attacks, the network will force them to lose their stake, and their money.
Again, how will this honest network partition coordinate their efforts to reach consensus on which chain to follow?
It seems like in all these scenarios, coordination by honest nodes is assumed, without giving any explanation as to how the nodes are able to reach consensus on which chain to follow of the many valid chains in existence.
acquiring that hardware (whether through direct buying or buying hacked machines) still costs money. it's just that it's easier to liquidate (compared to ASICs) when you're done the attack.
digiconomist [0] estimates that current etherum mining cost is 1.3 billion a year, or 3.6 million a day, or 151,000 an hour, or 2,500 a minute.
Multiply by 5 for cloud on demand premiums and you could dominate the etherum network for an entire day for 18 million. You could also do it for free if you can manage to do it with stolen credit cards. I'm amazed it hasn't happened already.
Lost confidence might negate this price increase.
Sure, if you can identify the location of the hardware and you happen to be a sovereign nation that can field the matériel required, who also doesn't have a problem instigating an international incident.
Now, if you're a single entity or a cartel of entities with the financial and technical resources to launch a majority of hash rate attack, you'll have considered what vectors a counter-attacker would attempt to come at you with.
For example, I expect the (simplistic) response to the hypothetical risk vector of "a physical assault on our mining hardware" to be migrated by distributing hardware across a number of data centers and pooling mining computations. Solved blocks are then propagated to the network through a single, obfuscated network node.
PoW also requires large energy consumption as participants will inevitably enter a race to compute the fastest.
PoS has the benefit that you still lock away capital, much like in PoW, but rather than backing that capital by money put into hardware and electricity, you cut out the middle man and base the capital off of the value of the network token.
Pos also has added security benefits. Vitalik Buterin explains this very well: If someone attacks a PoW blockchain by getting 51+% of the hashing power, the network's only response to recover from the attack is to change the hashing algorithm. This will likely force the network to abandon ASICs since hardware wouldn't exist, and move (at least temporarily) to general-purpose CPUs/GPUs. As soon as the hashing can be done with general purpose computers, the game is over, and the attacker can attack forever without any extra loss of capital, other than electricity.
PoS has the benefit that the network can simply cause the attacker to lose their funds and no longer have a stake. If the attacker wants to continue the attack, they have to buy more of the network token to be equal to 51% of the staked value. This will cause a price increase. Each time the attacker attacks, the network will force them to lose their stake, and their money. The attacker won't be able to keep this going forever.