Security Should Embrace Failure

The other day I asked a few folks for suggestions about compelling startups in different tech sectors, including security. Instead of touting a startup, Michele Chubirka, also known as Mrs. Y, replied with a link to a post with a provocative title: Enterprise Security Startups Are Booming - So Why Is Security Getting Worse?

The post, by Tom Foremski, poses an interesting question: When has a business bought enough security? As Foremski writes, "The answer seems to be that there is never enough."

I think the way this question is posed is worth examining, because it's based on a flawed premise. You don’t "buy" security. You buy hardware and software that has specific functions (and often a significant set of limitations).

The security, or lack thereof, emerges from a complex matrix that includes how the products are configured, the operators who implement and monitor the products, the degree to which users ignore or actively thwart any controls put in place by the products, and the organization’s overall operational practices (configuration and patching of related systems, access control, separation of duties, secure application development practices, and so on).

There are other internal factors, such as the degree of support that the security organization gets from management and business units. I haven’t even touched on the attacker’s side of the equation.

Let's Talk About Failure

In his blog, Foremski summarized the comments of a security analyst who said the "enterprise security situation is bad and will worsen further unless there is a radical new approach/technology developed."

But there is no new technology that will save the day. You can't buy security by purchasing a product or service, no matter how radical or innovative it might be, just like you can’t buy health by purchasing a gym membership. The membership is just the first step: you have to show up every day, get on the machines, and work up a sweat.

The notion of buying security also implies buying immunity; in other words, now that you've got products X, Y and Z, you are fully protected from attack. That's not true.

Even organizations with the most powerful tools and dedicated security teams can get beat—there are simply too many ways for an attacker to succeed, whether it's a zero-day exploit or a well-crafted social engineering play.

The problem is that invulnerability is the presumed measure for the security organization. It's an impossible and useless standard. A more sensible measure should account for a variety of factors, including how quickly an attack is discovered, how the damage is contained, and how soon the organization can recover.

Of course, if you're trying to sell a product, or get budget to buy a product, you're probably not going to help your case by talking about failure. This is a terrible Catch-22 for vendors and security pros alike, because unless we can honestly talk about failure, we probably won't get better at it.

Maybe the next radical security startup is the one that will embrace failure, and find ways to help customers recover from it, rather than have everyone pretend it won't happen.