Frequency Matters

13death

So there are a lot of ways to die. Like a lot. We worry about obscure ways to die. Its gruesome really, to die via an asteroid or “space junk” strike (so much so that we make TV shows about it), hockey puck death, or obscure elevator amputations.

…sort of like the various ways that IT security failures can cause security incidents. Now I’ve argued in the past that not all failure is bad, but in this post I want to talk about an important distinction that is often missed in risk assessments and that’s a focus on temporal factors. Put plainly, time matters.

This article and accompanying graph are a great way to organize some common ways to die (if you are looking for something to do on a Friday night). But it includes something that is missing from a lot of IT risk assessments: time.

Many assessment methods will tell you to assess “likelihood” such that you end up with some values like 80% or “Medium” etc. Now if you’ve been around me for any length of time, you will know that I quote “Fight Club” prodigiously to explain the problem with these values: “On a long enough timeline, the survival rate for everyone drops to zero.” And that’s why frequency matters. 80% what, tomorrow? In the next week? Year? Ever? Imagine if weather forecasts went the same way: just a picture of a rain cloud, no date, no day of the week, just a number that says 80% chance of rain. Should you bring an umbrella tomorrow? Next week? When?

So this is why I let my kid hold a baby alligator. Because, honestly, the odds of death by baby alligator are like, really really rare. Like 1/50 years or more (I dunno, I’m not an alligator expert). Plus, its mouth was taped up so yeah. Controls and such. And now she has a cool life experience and picture she can cherish :-)

So the next time you see likelihood without any reference to time period, call Shenanigans. Its bogus science and you don’t have to tolerate it.

High Accumulation

1280px-Snowy_street_in_Madrid_(Spain)_01I recently relocated to Charlotte from Ohio. Its South, but not so much so that it doesn’t get cold and yes, sometimes there is even snow. As I become acclimated to things down here, I am always surprised at the response that folks from here have to snow. They dislike it immensely and are often fearful of it. Now, I grew up in Pittsburgh which has a lot of snow. Ohio has a lot of snow too. So, this past weekend we had some weather reports that hinted at snow. They used a particular term that peaked my interest. The weather forecasters predicted that there would be a “high accumulation” of snowfall.

This is always the difficulty with verbal labels used to define measurements. Being that I am from the North, where snowfall is frequent, High to me means 6-8 inches or even 1 foot of snow or more. I imagine those from even farther North than I, probably laugh at my ranges and speak only in double-digit feet when measuring “high” amounts of snow. As it turns out, here in Charlotte “high accumulation” means between 1-2 inches. Oh, and that snow was mostly melted about a day and half later (for those that don’t know, this is a marginal amount of snow and the ensuing overreaction is largely comical to us Northerners).

When communicating risk, the same problem is endemic to that process. Just saying high, medium, or low is problematic. No one is able to divorce themselves from their biases and experiences. As a result, when you say “high risk,” there will invariably be those that think $10M, and others that are thinking in terms of $100K or even less. Why? Fundamentally speaking its because there are no numbers. Think about how much more plain it is to speak in terms of dollars or inches. We may disagree about what the relative impact of those units of measure, but one is not likely to argue that an inch isn’t an inch.

So as you go through your risk work, know that if you aren’t speaking plainly in terms that are universal (frequency and magnitude), then know that you may be perceived as shouting at clouds…

It’s always about the money

crumpled_bill[1].rI live in a fairly straightforward world, I guess. I’m often accused of being naive but I’d argue I feel enlightened. Put bluntly, I think risk is always about money.

This has to do with my education from Douglas Hubbard. When I hear nebulous problems I know that I can use the methods of Enrico Fermi to decompose them by asking Fermi questions. Most people think this too simplistic for their complicated world, but I’ll side with the Nobel prize winner on this one thank-you-very-much.

I often have to answer for my position when I stoically say that reputation can be measured in dollars. “But Jack,” folks will say, “its so hard to measure that.” I disagree. Why does anyone care about reputation? Well, for businesses, its about the ability to retain and acquire business. For whatever scenario you are analyzing, make some calibrated estimates of how much business you are likely to loose (use ranges) and suddenly you’ve applied some really complicated quantitative methods in a very practical and straightforward manner.

So I was having a discussion about control automation and the reasons organizations spend money on just such a thing. Now, since my worldview on this is that risk = money, it was straightforward for me. Organizations automate controls in order to save money. The howling in protest was fierce. “Jack,” they said, “automation speeds things up so the answer must be efficiency.” “Why do we care about being efficient?” I countered. “Jack, automation also reduces errors, so effectiveness is the answer,” said another. “Why do we care about being effective?” I countered again. Indeed, in all aspects of risk, the answer can be boiled down to money. I want to be faster so I can have the same resources do more work, saving me the cost of buying more. I want to be more effective so that I don’t have more loss events, saving me the cost of responding to them. The other common argument I hear is to blame regulators or auditors, as in “we need to automate to satisfy the auditors.” Once again decomposition helps us analyze this approach–why do we need to satisfy them?

When presented with these sorts of questions I come clean and admit that money drives risk and priorities. We spend money on things we care about, and we spend money to avoid losing money elsewhere. There are countless examples of us using money as a measure of time as well, as in how much would you spend to avoid waiting in line at Disney World? Turns out, there is a service where you can pay to pay someone to wait in line for you, making your holiday that much more pleasurable (although I think its only open to celebrity visitors–and priced for them as well).

Its not cynical, its enlightened. Once you accept this basic premise, all your “hard” problems get that much more easier to model.

Substituting Risk Tolerances

push-button-receive-baconI hate hand dryers in washrooms. I’m not alone: if Wikipedia is to be believed, 63% of people preferred paper towels over hand dryers in restrooms. I’d wager the other 37% choose what they thought was the right answer. Each time I use them, I always end up with cold, wet hands and if I’m forced to stand in front of them, water all over my clothes. I try to stand to the side and I one time watched the blower fling water all the way across the restroom–no small feat. Surely that wet, slick floor I left behind creates a terrible safety hazard. Heck, there is even a dispute about how much more environmentally friendly they are (if full cost environmental impact accounting is to be believed). My problem stems from the simple fact that they largely fail at their stated purpose, that of drying my hands quickly.

So if they are mostly hated, then why do companies implement them? Well, to put it bluntly it’s not like you are going to shop somewhere else because they have hand dryers there. If studies are to believed then I guess companies can save 99% of the cost of paper towels in a single year.

So what does this have to do with risk? Hand dryers (to me at least) are a clear case of substituting risk tolerances. Allow me to explain. When you are done washing your hands, your primary goal is to dry your hands and get out of there as quickly as you can. You are probably not thinking about saving the world with your hand drying choice or even saving money for the business you are at. Your priority here (I often equate priorities with risk) is in direct conflict with the host company. In fact, if its your employer that has the hand dryer, then it means they’d rather you stand there for some indeterminable time until your hands are dry versus getting back to your post as quickly as possible. Okay so may you save a minute or two (I think most people just give up and wipe their hands on their pants, defeating the purpose), but multiply that by how many trips per day times how many people and its no small investment (I used to work with process engineers that thought about stuff like this all the time).

You may be thinking that I’m neurotic about this, and you may be right, but when you think about risk constantly like me you start to see it everywhere. And the hand dryer scenario is not unique. While waiting in line at IKEA at closing time one night, someone in our party asked why they didn’t open up more lanes. The answer is simple–what’s the odds that after spending the last couple hours shopping and schlepping your purchases to the sole closing-time cashier that you would abandon them and sacrifice the last few hours of your life. Slim to none I’d say. Here too is a risk-based decision. They are accepting marginal dissatisfaction in order to save some money on a second or third cashier.

These sorts of trade-offs happen all the time and we hardly notice them. Usually they involve discounting the value of time–yours and mine–in favor of cost avoidance. I try and make these scenarios plain in my mind. I want to know when the value of my time has been discounted. I have less personal tolerance for my time being wasting and I often seek out scenarios where I pay a premium to have more personal time in my life.

How often has your personal risk tolerances been violated without your explicit knowledge? Perhaps its time to manage your resources better…

Private Sector Cyberwar – part 2

I wrote about this last May, namely that so-called cyberwar events are not for the domain of the private sector to defend against. I made an argument that cyberwar perpetrators are in the upper percentiles of attackers (95% +) and that outside of building our organization’s control strength up to that level, let’s just leave cyberwar to governments.

With that a backdrop, I was fascinated by this article that identifies that this exact thing I outlined was actually happening. The bank BB&T has sought help from the government in warding off DDoS attacks believed to be from state-sponsored attackers. This line in particular seemed to reflect my posture on these types of attacks:

“BB&T…and others now say they have spent millions in warding off the attacks and can’t be expected to fend off such attacks from another government.”

If I read between the lines, they have spent a lot of money to raise their control strength, however against attackers in the 95th percentile, its just really outside of their responsibility to defend against it. Like in warfare of earlier times, its time for the generals to step up and keep the farms of the countryside from being destroyed lest they be unable to feed their armies in times of need.