> Moreover, there is a general argument that the attacker will always be able to keep the consensus from finalizing nomatter what the fix is.
> The argument simply comes from the fact, that mathematically provable binary consensus algorithms known in this universe have n2 behavior, and ETH2 is linear in n .
> Therefore, the only way to really fix ETH2 is to make it n2 . Otherwise it is unfixable from the math point of view. There will always be another attack.
> It may be that by continuing patching a fix after a fix after a fix one can end up with something that will work from an engineering point of view.
> This will be security by obscurity.
> But it will not be secure from the math point of view.
So while bodging in patches might work one day, it's an immature approach and a scary place to try store value.
The sheer number of moving parts create a scary amount of emergent complexity and complexity is the enemy of security.
Disclosure: I'm an early but now uncomfortable ETH holder.
On the "fundamental" link, see this reply that appeared after you commented:
"The reason that Ethereum’s consensus can run in n time rather than n^2 is BLS signature aggregation. The attacks in the paper however aren’t attacks on signature aggregation. So I don’t think your argument is valid."
"Change fork choice rule to mitigate balancing and reorging attacks"
https://ethresear.ch/t/change-fork-choice-rule-to-mitigate-b...
And more fundamentally at the heart of the matter
https://ethresear.ch/t/comment-on-three-attacks-on-proof-of-...
> Moreover, there is a general argument that the attacker will always be able to keep the consensus from finalizing nomatter what the fix is.
> The argument simply comes from the fact, that mathematically provable binary consensus algorithms known in this universe have n2 behavior, and ETH2 is linear in n .
> Therefore, the only way to really fix ETH2 is to make it n2 . Otherwise it is unfixable from the math point of view. There will always be another attack.
> It may be that by continuing patching a fix after a fix after a fix one can end up with something that will work from an engineering point of view.
> This will be security by obscurity.
> But it will not be secure from the math point of view.
So while bodging in patches might work one day, it's an immature approach and a scary place to try store value.
The sheer number of moving parts create a scary amount of emergent complexity and complexity is the enemy of security.
Disclosure: I'm an early but now uncomfortable ETH holder.