Security measures to overcome 51% attack

Question

I was reading During a 51% attack, What Can the Attacker Actually Accomplish?

from there, i understand the after-effects are something serious, so what are the security measures which ethereum have and what if all they fail?

Answer 1

If a full-frontal and obvious 51% attack occurs, the live Ethereum network will essentially have failed. There is no real way to protect against this eventuality other than to watch for orphan rates to spike and then immediately notify everyone that the network is unreliable.

The main thing protecting the Ethereum network (and the Bitcoin network, and any other public blockchain network) is the lack of incentives to perform such an attack in the first place. To perform such an attack the attacker must be willing to dedicate a large amount of resources simply to destroy something. Unless they have a large financial advantage to gain from that destruction, those resources will largely be lost permanently to them when the network is destroyed. Against this type of attacker, the greatest weapon Ethereum has is the market price and issuance rate of Ether. The higher the market price, the more hashing power gets paid for, and the more expensive an attack becomes. In the past two weeks Ethereum has gotten almost twice as secure from this type of attack just due to rises in the market price of ether.

From a full-frontal attack, disaster recovery might still be possible, and the Ethereum economy might yet live again. The pre-attack blocks wouldn't be gone, even if the attacker reversed old blocks. If the community could agree on a "last good state" they could fork the network at that point, change the proof of work function, and begin again (on a PoS network this is even easier: all that has to be done is to make a fork which treats the attacker's coins as invalid).

But it's unlikely a real 51% attack would be a full frontal attack. A more subtle 51% attack would be much more likely to profit an attacker. We might detect it as an unusual pattern in the orphan rate, or (for very subtle and transitive types of attacks) we might not even notice. If a subtle 51% attacker were able to operate a long time without being detected, disaster recovery would likely be nearly impossible. In the end, performing such an attack, "getting away clean" with significant resources, and then publicly announcing the attack with proof would probably be a far more effective way of killing the network.

Again, the only protection against such an attacker is to attempt to make the cost of the attack as high as possible. Casper will help with that, a higher issuance rate in Homestead will help with that, and a higher market price will help with that. We can also take solace in the fact that the Bitcoin network has survived without an attack for over 7 years at this point. It may well be the case that no one has a strong enough incentive to put forth the resources that would be needed. But ultimately, blockchain technologies can't make us invulnerable to arbitrarily powerful adversaries. No blockchain technology in the world can protect the Ethereum network from planetary annihilation, for example. The only thing we can do is raise the height of the line on the following sign:

---

----------------- You must be this powerful to destroy the Ethereum network