There is a huge difference between 'making' reliability and 'checking' reliability. We can't just assume that because we rigorously 'check' reliability that this will procedurally imply that someone else will 'make' reliability. This is like driving your car using the rear vision mirror only. Trying to guide an engineering team on what you 'do want' is not simply telling them what you 'don't want' ... once they fail to meet (ambiguous) guidance.