7
  • What if a client malfunctions due to the corruption of data isn't it risky to stake?

With PoW this is no issue but with PoS this seems to be a real issue with the slasher protocol. I would be worried to lose my funds.

I mean the risk to reward is for sure good, but I don't care if I would stake everything.

  • Is this a valid concern?
  • Is there a risk?
2

Slashed funds are the same as money wasted on energy used to mining on a fork and don't achieve becoming the largest (51% attack) .

You'll lose your founds if you behave badly as you lose Kw and CPU/GPU cycles (which traduces in money) while mining not correct blocks or on secondary chains/forks.

PoS carries some kind of bad behaviors (adopted by validators) to take care of ( which means attacks sometimes) like:

  • Nothing at Stake: The Nothing-at-Stake issue is the event of a fork (the “fork” can be a malicious attempt to rewrite the history and do a double spend) the optimal strategy for any validator is to validate every chain so that the validator gets their reward no matter which fork wins. This means that consensus algorithm doesn’t work as intended.

The attacker may be able to send a transaction to an exchange for some digital commodity (usually another cryptocurrency), receive the commodity and then start a fork of the blockchain from one block behind the transaction and send the money to themselves instead.

Even with 1% of the total stake the attacker’s fork would win because everyone else is mining on both.

You can review the CasperFFG details here:

https://github.com/ethereum/casper/tree/master/casper/contracts

Vitalik said in order to solve that problems (by implementing slashing funds to bad behaving validators):

There is a proposal that we are very seriously considering, where a validator only loses their entire deposit if they are faulty around the same time that many other validators are faulty; if something happens to one single validator they would only lose a few percent of their deposit. This should make it safe enough to stake even without a multisig address.

It also has the nice side effect of adding an anti-centralization pressure, as there is an incentive not to stake with the same pool or on the same VPS platform or even with the same software as many other people, as each validator benefits from their failures, being uncorrelated with those of other validators.

But don't worry, as Vitalik said here, staking isn't that risky if you behave well (or you try it).

Read the full discussion on reddit: https://www.reddit.com/r/ethereum/comments/7mmsvt/i_just_read_casper_ffg_paper_and_have_a_question/

If you want to be even more sure that you are protected against slashes, you can implement multisig addresses for example, which make you as an individual more anonymous.

Hope it helps!

-1

Well miners are actually at a risk as well. If a miner goes offline 20% of the time, then they lose 20% of their profits. They could make this up over time, but as difficulty increases they're going to be getting less and less ETH/day with the same hardware.

That said, I completely understand why there would be more concern over staking, because you arent just losing time earning, you're losing your investment. Slashing isn't a one time lose all thing though, you'll slowly get penalties for not voting. You won't just lose your stake when you fail to vote.

  • 3
    Yeah if you data gets corrupted at PoW you loose the reward. We arte not talking about being offline, we talk about data corruption. With PoS you will get a pentalty. I'm in no way pro PoW (exactly the opposite), I want to analyse both. – Ini Nov 15 '17 at 11:35
  • What do you mean "data corruption"? If you need the node losing chain data necessary to mine/validate, then it's the same as the miner/validator going offline. If by "data corruption" you mean data getting corrupted in such a specific way that the miner/validator continues in a way that looks to the user to be working, but not to the network, then you would still basically be offline to the network – flygoing Nov 15 '17 at 12:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.