4 ms·

Proof of Stake is unable to deal with network partitions, the likes of which every so often happen to cut entire countries off the Internet PoS is a disaster

by corv 8y ago

Proof of Stake is unable to deal with network partitions, the likes of which every so often happen to cut entire countries off the Internet PoS is a disaster waiting to happen for Ethereum.

jron 8y ago
Don't worry. They'll just roll the chain back.
dahdum 8y ago
The stakers will lose their stake, and those shards affected will go offline, but the rest of the network will continue on until the interruption is over. Ethereum is designing the system specifically to accommodate that sort of problem, last I remember you needed 95%+ uptime to generate return, so if your internet goes out 5% of the time you should not stake.
- wmf 8y ago
  Why would the minority side of the partition go offline? How would it even know that it's on the minority side of a partition?
  - stale2002 8y ago
    > How would it even know that it's on the minority side of a partition? Because it would be obvious? The world does not exist in a state where a "true" network partition can ever happen. A "true" network partition I would define as preventing all information, of any type, from going into and out of a country. This is a ridiculous scenario that isn't even worth considering.
- jzwinck 8y ago
  I imagine that means it's OK if my network connection goes offline 4% of every hour. But what if I'm offline for 4% of 90 days, all at once?
dlubarov 8y ago
That's not an inherent limitation of PoS; PoS systems can be either AP or CP. Most PoS designs go with CP, but NXT and Ouroboros are examples of AP PoS systems. They use Nakamoto consensus, much like Bitcoin. It's true that Ethereum 2.0 will require 2/3 availability in order to finalize transactions, but that's true of any BFT system; see [1] for a proof. [1] https://zoo.cs.yale.edu/classes/cs426/2017/bib/bracha85asynchronous.pdf https://zoo.cs.yale.edu/classes/cs426/2017/bib/bracha85async...
RexetBlell 8y ago
https://github.com/ethereum/wiki/wiki/Proof-of-Stake-FAQs#so-how-does-this-relate-to-byzantine-fault-tolerance-theory https://github.com/ethereum/wiki/wiki/Proof-of-Stake-FAQs#so...

by corv 8y ago

Proof of Stake is unable to deal with network partitions, the likes of which every so often happen to cut entire countries off the Internet PoS is a disaster waiting to happen for Ethereum.

jron 8y ago

Don't worry. They'll just roll the chain back.

dahdum 8y ago

The stakers will lose their stake, and those shards affected will go offline, but the rest of the network will continue on until the interruption is over. Ethereum is designing the system specifically to accommodate that sort of problem, last I remember you needed 95%+ uptime to generate return, so if your internet goes out 5% of the time you should not stake.

wmf 8y ago
Why would the minority side of the partition go offline? How would it even know that it's on the minority side of a partition?
- stale2002 8y ago
  > How would it even know that it's on the minority side of a partition? Because it would be obvious? The world does not exist in a state where a "true" network partition can ever happen. A "true" network partition I would define as preventing all information, of any type, from going into and out of a country. This is a ridiculous scenario that isn't even worth considering.
jzwinck 8y ago
I imagine that means it's OK if my network connection goes offline 4% of every hour. But what if I'm offline for 4% of 90 days, all at once?

dlubarov 8y ago

That's not an inherent limitation of PoS; PoS systems can be either AP or CP. Most PoS designs go with CP, but NXT and Ouroboros are examples of AP PoS systems. They use Nakamoto consensus, much like Bitcoin. It's true that Ethereum 2.0 will require 2/3 availability in order to finalize transactions, but that's true of any BFT system; see [1] for a proof. [1] https://zoo.cs.yale.edu/classes/cs426/2017/bib/bracha85asynchronous.pdf https://zoo.cs.yale.edu/classes/cs426/2017/bib/bracha85async...

RexetBlell 8y ago

https://github.com/ethereum/wiki/wiki/Proof-of-Stake-FAQs#so-how-does-this-relate-to-byzantine-fault-tolerance-theory https://github.com/ethereum/wiki/wiki/Proof-of-Stake-FAQs#so...