No. The required amount is greater than 2/3 (67%) not greater than 1/2 (51%) on PoS for Ethereum.

Also, see this tweet from Vitalik: https://twitter.com/VitalikButerin/status/130129808602782105...

So even if there was a successful first attack from some organised body that's all the time they get.

> The required amount is greater than 2/3 (67%)

I'm not sure that's correct (the tweet doesn't get into any details so I went to ethereum's website https://ethereum.org/en/developers/docs/consensus-mechanisms...)

> The threat of a 51% attack still exists in proof-of-stake but it's even more risky for the attackers. To do so, you'd need to control 51% of the staked ETH. Not only is this a lot of money but it would probably cause ETH's value to drop. There's very little incentive to destroy the value of a currency you have a majority stake in. There are stronger incentives to keep the network secure and healthy.

The keypoint seems to be that if your attack fails your stake gets destroyed so besides the positive incentives (a good stable network working for all) this system also relies on punishing failed attacks.

Correct me if I'm wrong, but you would only need greater than 2/3 of the amount of Ether currently staked, not 2/3 of the total amount of Ether. The amount of staked Ether probably would be less than 10% of the total issued Ether, maybe much less as staking rewards go down.

Correct it is of the staked Ether not total supply. It's an important detail of course and easily missed, a good way for future people stumbling across here to think about it is that any attack only involves those participating in consensus. For Bitcoin consensus is mining. For the beacon Ethereum chain consensus is staking, therefore only staking Ether is relevant.

Great explanation. One interesting aspect of PoS is that attackers really only have 1 shot, and will get slashed or lose their assets. In PoW, you in theory can keep attacking again and again.

So the endgame is that someone owns 99% of all ETH and everyone just accepts it because it's not a threat? Why have a blockchain at all?

A 51% staker who attempted to reverse transactions on ETH2 would be automatically penalized by the destruction of their stake. The network would keep running and the end result would be a large sudden deflation of ETH. On PoW this would be like an attacker's mining rig burning down.

A 51% staker who just censored transactions could hold out longer. If the problem were severe, the community would have to decide whether they want to manually fork off the attacker. The equivalent for PoW would be changing the hash function.

Gaining 51% can be more expensive to do on PoS than on PoW. If 10% of the tokens are staked, you need to accrue another 10% of the total market cap. On PoW, if the annual inflation rate is 2%, the hardware is good for two years, and half the mining cost is electricity, then the total value of mining equipment is only 2% of the market cap, and that's how much you'll have to spend to get 51%. (If miners are rentable, then much less for a brief attack.)

Yes. But thats the point of PoS. Why would someone who owns >50% of something want to destroy it? Literally hurting yourself more just to hurt others a lesser amount.

Correct me if I'm wrong, but if you own > 50% and get the ability to reverse transactions, it can still go undetected. So it's possible to hurt others without hurting yourself.

How would it go undetected? The people whose transactions are being reversed would definitely notice, as would anyone else paying close attention to said transactions.

In theory yes, but in practice the people most involved (holding large stakes) would pull out of the system en masse, crashing the value and (probably) the whole network with it. Same with BTC. In theory anyway.

What if those people had more to lose by pulling out than pretending nothing happened as long as the 51% attacker does not abuse it "too much"?

It definitely doesn't go undetected, as the conflicting transactions would have to be in blocks both forking from a common block, both signed by the attacker, which is a slashing condition that would burn their staked ether.

Also, it would be kind of visible and hard to hide. With sufficient high profile stake holders, this would be very unlikely to happen and even harder to hide.

Ok, but suppose someone owns 50% of the tokens. Doesn't this mean the system is de facto centralized. Even if the majority holder behave well, how is it better than a centralized solution?

Unrealistic surely but, "Some just wants to watch the world burn"

Another aspect of PoS: it makes these attacks incredibly expensive. Currently staked value is over half a billion USD.

Someone (very rich) could bet even more money on that thing falling..?

If a large government decides it's a problem, would they hold it or destroy it?

Vitaliks recent blog post is probably the best place for a quick overview of Pos/PoW attack costs and why PoS looks to be a more secure approach https://vitalik.ca/general/2020/11/06/pos2020.html

I think PoS is much more likely to get wider adoption than PoW with the rest of the Ethereum ecosystem, FWIW. Glad Ethereum is making the switch.

Imagine somehow a single party gets 66% of all ETH, enough for them to execute the equivalent of a "51% attack".

The community will notice and can decide to do a hard fork of the network where they "delete" the attackers coins.

So the network would have experience a hickup, but the hacker has lost billions of dollars worth of ETH and can't attack anymore.

This is different from Proof of Work & Bitcoin. If an attacker gets 51% of the "mining power" (physical hardware), there's nothing the community can do to "delete" their hardware.

Yes