Having just returned from a west coast conference - Big Data/HADOOP analytic constructs are in place at large cloud providers to overcome the human information processing bottleneck when it comes to security operations. For example, Salesforce Operation reported 15TB of logs per day. In the speakers analysis, the 100GB per day per 2 people on-task linearly scales poorly. Given job scarcity for a SECOPS staffer, the direction is recruit for people who training the cloud-based ML capability to perform log reduction to stream only important anomalous events. The human challenge occurs with defining use cases, developing model dimensions, developing and refining data models, and mitigating data volume, veracity, volatility, fidelity, and integrity. Follow-on challenges reveal human-directed analytical effort to develop model clusters, risk cluster scoring, and defining the rule set and actions to occur when machine data reveals risk behaviors falling outside of expected data cluster behavior.

The Cyber Human Error Assessment Tool (CHEAT®) might be of interest: https://www.thalesgroup.com/en/cyber-human-error-assessment-tool

Agree, this is an open umbrella problem. It may protect you from rain, but not from running water under you feet. Both the proper security policies and regulation are critical, in conjunction with the legal measure that may need to be implement if necessary. This may drive the Cyber Security discussions fundamental towards issues related to Security vs. Privacy. How Secure and Safe we are, and at the same time how much ours and anyone's Privacy is respected and untouched. The Castle Strategy of building a strong defenses and high walls may protect the Establishment from malicious intruders who may wish to jump or climb over the fence. However the strong defenses and high walls may not be enough for the undercurrents and high winds.