How does Casper achieve that all validators magically converge to a block without deciding the block deterministacally (stake-grindable)?

How is this implemented in the testnet?

1 Answer 1


Because Casper FFG is an overlay to a PoW blockchain, stake-griding isn't an issue, as validators aren't actually creating blocks (miners still are). The PoW blockchain and usual "highest work chain is best chain" fork-choice rule is enough for validators to converge on the same chain (assuming validators see the same blocks, which we do in PoW).

In general, stake grinding isn't a fundamentally hard problem to solve (like nothing at stake, for example). You can see a nice post about randomness generation in PoS here, where there are a couple options to eliminating it from being a major concern.

  • But the client has to have some kind of implementation of the convergence that the validators can change. This implementation is deterministic and can be exploited/grinded. Exploited definition in this case: Not so much that he/she will get less rewards (assumed that incentives are well defined), more that his choice is forseeable.
    – Ini
    Commented Mar 26, 2018 at 23:10
  • Note that I assume (maybe because of lack of knowledge or ambigious definitions of my learning sources) that there is no validator ordering and "the validators have to figure it out on their own". Is there validator ordering via randomness generation in Casper?
    – Ini
    Commented Mar 26, 2018 at 23:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.