The issue is in a lower level than that. A proof of stake system will want to randomly choose who gets to mine the next block, weighted by how much everyone is staking. But making this choice depends on everyone agreeing on a source of randomness, which is what the previous posts were talking about.
That might actually work but there is a lot to analyse to make sure that something is actually at stake. Firstly can dishonesty be detected and traced back to a stake holder? I'm not sure about that within specific randomness generating schemes. Secondly even if participants can be detected cheating it may still be economically advantageous for them to be dishonest if there is only a small probability that they lose their stake.