Original URL: http://www.theregister.co.uk/2011/10/11/faith_based_security_fail/

Infosec 'needs warrior cryptoboffins' to beat hackers

Drop and give me 50 better data sets, maggot

By John Leyden

Posted in Security, 11th October 2011 14:05 GMT

RSA Europe The infosec industry needs to move beyond "faith-based security" to an evidence-based approach that takes ideas from battlefield combat if corporations are ever to get ahead of hackers and keep security spending down to manageable levels.

Joshua Corman, director of security intelligence at Akamai, argued that while almost every enterprise attempts to develop security metrics for its environment, these approaches are more akin to "numerology" than hard science.

Raw data and numbers are rarely available in the field of information security and, when they are, they tend to get misquoted or misunderstood, according to Corman. For example, a widely cited misquote from a Verizon data breach report – "90 per cent of breaches in 2008 were due to patchable vulnerabilities where a patch had been available for 12-18 months" – is often taken as a basis by enterprises for developing a security policy.

In truth the sentence needed to be qualified "of the 22 per cent that were patchable [the patch had been available for more than a year]. By 2009, only 6.7 per cent (six of 90) breaches stemmed from patchable vulnerabilities. The following year, zero breaches were due to patchable vulnerabilities.

Despite this, most CISOs have programs to "patch faster" while adversaries have moved elsewhere".

Another problem is that advice designed to address the main security shortfalls in small businesses is sometimes misapplied to large enterprises. Raw data on accidents can be applied to draw up actuarial tables for insurance purposes, but the same approach doesn't work in information security, according to Corman.

"Collecting data and numbers to try to develop actuarial tables for security just doesn't work because the problem space just isn't like that," Corman argued. "Information security is less about actuarial tables and more about game theory."

Vendor-supplied statistics are often misleading, Corman told El Reg. "Vendors pluck out figures that support their sales pitch. They use statistics like a drunk uses lampposts – more for support than illumination."

Rather than taking lessons from industry surveys, analyst reports or vendor-supplied arguments, security managers should look to lessons from military doctrine. The "observe, orient, decide and act" loop can be applied as well to fighting cyber-adversaries with unknown capabilities and tactics as it is in battlefield situations, according to Corman.

Corman is due to expand on his ideas during a conference debate snappily entitled Metrics are Bunk!?: A Zombie Apocalypse, Football/Soccer & Security Metrics at the RSA Conference in Europe on Thursday. ®