Incremental "Gains" Are Just Slower Losses

Anton Chuvakin and I were having a fun debate a couple weeks ago about whether incremental improvements are worthwhile in infosec, or if it's really necessary to "jump to the next curve" (phrase origin: Guy Kawasaki's "Art of Innovation," watch his TedX) in order to make meaningful gains in security practices. Anton even went so far as to write about it a little over a week ago (sorry for the delayed response - work travel). As promised, I feel it's important to counter his arguments a bit.

Anton started by asking, "[Can] you really jump to the "next curve" in security, or do you have to travel the entire journey from the old ways to the cutting edge?"

Of course, this is a sucker's question, and it belies misunderstanding the whole "jump to the next curve" argument, which was conceived by Kawasaki in relation to innovation, but can be applied to strategy in general. In speaking of the notion, Kawasaki says "True innovation happens when a company jumps to the next curve-or better still, invents the next curve, so set your goals high." And this, here, is the point of arguing for organizations to not settle for incremental improvements, but instead to aim higher.

To truly understand this notion in context, let's first think about what would be separate curves in a security practice vertical. Let's take Anton's example of SOCs, SIEM, log mgmt, and threat hunting. To me, the curves might look like this:
- You have no SOC, SIEM, log mgmt
- You start doing some logging, mostly locally
- You start logging to a central location and having a team monitor and manage
- You build or hire a SOC to more efficiently monitor and respond to alerts
- You add in stronger analytics, automation, and threat hunting capabilities

Now, from a security perspective, if you're in one of the first couple stages today (and a lot of companies are!), then a small incremental improvement like moving to central logs might seem like a huge advance, but you'd be completely wrong. Logically, you're not getting much actual risk reduction by simply dumping all your logs to a central place unless you're also adding monitoring, analytics, and response+hunting capabilities at the same time!

In this regard, "jump to the next curve" would likely mean hiring an MSSP to whom you can send all your log data in order to do analytics and proactive threat hunting. Doing so would provide a meaningful leap in security capabilities and would help an organization catch-up. Moreover, even if you spent a year making this a reality, it's a year well-spent, whereas a year spent simply enabling logs without sending them to a central repository for meaningful action doesn't really improve your standing at all.

In Closing

In the interest in keeping this shorter than usual, let's just jump to the key takeaways.

1) The point of "jump to the next curve" is to stop trying to "win" through incremental improvements of the old and broken, instead leveraging innovation to make up lost ground ground by skipping over short-term "gains" that cost you time without actually gaining anything.

2) The farther behind you are, the more important it is to look for curve-jumping opportunities to dig out of technical debt. Go read DevOps literature on how to address technical debt, and realize that with incremental gains, you're at best talking about maintaining your position, not actually catching up. Many organizations are far behind today and cannot afford such an approach.

3) Attacks are continuing to rapidly evolve, which means your resilience relies directly on your agility and ability to make sizable gains in a short period of time. Again, borrowing from DevOps, it's past time to start leveraging automation, cloud services, and agile techniques to reinvent the security program (and, really, organizations overall) to leap out of antiquated, ineffective practices.

4) Anton quipped that "The risks with curve jumping are many: you can jump and miss (wasting resources and time) or you can jump at the wrong curve or you simply have no idea where to jump and where the next curve is." To a degree, yes, this is true. But, in many ways, for organizations that are 5-10 years behind in practices (again, this applies to a LOT of you!), we know exactly where you should go. Even Gartner advice can be useful in this regard! ;) The worst thing you can do is decide not to take an aggressive approach to getting out of technical security debt for fear of choosing the "wrong" path.

5) If you're not sure where the curves are, here's a few suggestions:
- Identity as Perimeter - move toward Zero Trust, heavily leveraging federated identity/IDaaS
- Leverage an MSSP to central manage and monitor log data, including analytics and threat hunting
- Automate, automate, automate! You don't need to invest in expensive security automation tools. You can do a lot with general purpose IT automation tools (like Ansible, Chef, Puppet, Jenkins, Travis, etc.). If you think you need a person staring a dashboard, clicking a button when a color changes, then I'm sorry to tell you that this can and should be automated.
- If your org writes code, then adopt DevOps practices, getting a CI/CD pipeline built, with appsec testing integrated and automated.
- Heavily leverage cloud services for everything!

Good luck, and may the odds be ever in your favor! :)

My Other Pages

Support Me

Support EFF


Bloggers' Rights at EFF

Creative Commons License
This blog is licensed under a Creative Commons License.
Powered by Movable Type 5.2.10