3 ms·
What they call "moving checkpoints" is a pretty widely-accepted mitigation to long-range attacks. Checkpoints might seem like a security issue, but it really de
by dlubarov 4y ago
What they call "moving checkpoints" is a pretty widely-accepted mitigation to long-range attacks. Checkpoints might seem like a security issue, but it really depends on how recent they need to be (which is a function of staker exit delays) and the process for discovering checkpoints.
E.g. most blockchain clients come with a genesis state baked in, which is effectively a (very old) trusted checkpoint. Someone could try a long fork attack by changing the genesis, but the attack would be obvious from the git history. (Many users might still not notice, but that goes for any attack on the client codebase.)
- ShamelessC 4y agoThe paper I linked discusses moving checkpoints.
- dlubarov 4y agoYes that's why I used their term. It doesn't say much about it though. The Ethereum Wiki [1] has a much better discussion of its security implications IMO. But my point is mainly that moving checkpoints is the standard solution to long forks, so a discussion about PoS security should probably focus on that, vs theoretical alternatives like trusted hardware. [1] https://eth.wiki/en/concepts/proof-of-stake-faqs#what-is-weak-subjectivity https://eth.wiki/en/concepts/proof-of-stake-faqs#what-is-wea...