AIUI there are multiple ways to implement proof-of-stake. A friend of mine proposed treating the chain with the most coin days destroyed (along with correct difficulties for the standard proof-of-work function) as the "longest" one rather than only the most difficult chain. Does that not work?