There are a number of discussions happening regarding the differences in risk based security vs compliance based security. These mostly have grown from discussions around PCI and other imposed standards of control. My opinion is that risk and compliance are two necessary actions.
Agusto, over at securitybalance blog is the latest to discuss the merits of compliance based security. I share his opinion that creating prescriptive measurable requirements goes a long way to improve the security of a large number of organizations. This is a given - I compare this other compliance programs like laws regarding the use of seat belts in automobiles. They exist because it is better to protect everyone to the same level of protection than it is to measure the specific protections required based on the roads that are being driven on that day, or the specific use of the vehicle, etc.
What this doesn't mean is that there isn't some degree of risk management being performed - its just that its not being performed at the vehicle operator level - where in many cases people would chose not to wear them out of inconvenience.
Like the laws for seatbelts, the larger risk which needs to be managed is not at the corporate level but at the societal level. The consequences of security failures at the organization level do not usually gain enough attention to warrant the appropriate protection, but I would argue that the consequences of systematic security failures across our society's infrastructure are the basis of massive harm.
Protecting our society at this level is the responsibility of our governments - and laws should be enacted to require adequate security protection, and impose legal penalties where they are not sufficient.
The identification of information warranting this protection is a required risk-based process. What types of information need to be protected in order to protect our people, our intellect, our industries and our livelihood?
To drive out the waste of objective-less risk management processes Anton asks the question on his blog-
"What is the risk-driven, correct frequency of changing my email password?"
Attempting to measure specific risks to a combination of the frequency of a control's failure and the existence of a real threat is for sure the wrong way to measure risk. But I think there still is a valid risk discussion regarding the use and standards of password use in protecting certain types of information. Lets rephrase:
"Should passwords be managed for systems that are used to communicate financial transaction data?"
This subjective question is far easier to qualify - and debate the merits and extent of compliance requirements associated with it. Will it make sense in every scenario in every organization? No. But will its application across the majority of scenarios in the majority of organizations help protect our livelihoods as a society? If the answer is yes - it should become a standard.
We need to define a scope of information which should be protected, create the standards to which the information should be protected, and institute formal legal processes to enforce compliance.
This is exactly what the FIPS and NIST standards describe? But these programs need to be extended to more than just federally controlled information types, and begin enforcing these rules on all data we value as a society.