Comment by kmoser

The next logical step would be to somehow inform users so they could take action to replace the bad memory. I realize this is a challenge given the anonymized nature of the crash data, but I might be willing to trade some anonymity in exchange for stability.