It isn't just taking on risk by trying something new. Even a known better solution can result in assuming risk. Take a long existing systematic problem which

It isn't just taking on risk by trying something new. Even a known better solution can result in assuming risk. Take a long existing systematic problem which produces 100 bad effects a day (people getting double billed, database errors, what have you). Someone thinks of an improvement to reduce this to 10 bad effects a day instead. A major improvement. But they don't do it. Why? Because anyone who approves of the new system is now the person responsible for the 10 errors a day. It goes from 100 errors per day because that is just how the system works to 10 errors a day because of Bob two cubes over. Bob ends up being blamed for the 10 errors far more than being congratulated for the 90 that no longer occur.