Defending The Kingdom

Here’s the story:

Who among us doesn’t love a good hack? After putting forth a $10,000 come-and-get-us challenge, it’s possible that StrongWebmail CEO Darren Berkovitz is rethinking his stance on that. The company, which makes voice-based authentication software, dared hackers to break into Mr. Berkovitz’s Web-mail account and report back details from an upcoming date on his calendar. A week later, a team of high-profile security researchers contacted a reporter with precisely that information.

Once again, it’s worth pointing out that there is no such thing as perfect security. You have to choose a level that is good enough. It can be uncomfortable to know and accept that your email address could get hacked, but there’s no way around it. All you can do is decrease the chances in a way that doesn’t cramp your style too much.

I advocate cramping your style a bit more than others in your category of “target juiciness”. If you have typical assets to protect, put just a bit more effort into security than the typical person. If you are atypical, put just a bit more effort into security than those with your level of assets.

Strong passwords are important, and I recommend using eight or nine digits whenever you can. Sometimes, however, you can’t avoid using a short password. For example, many ATMs outside North America will not accept long passwords, so you have to use a short ATM password if you live or travel outside of North America. In such an instance, is using a four digit password unsafe?

The answer, as far as I can tell, is no. A longer password would be better, but a four digit password for your ATM card is good enough.

Why is a four digit password okay for your ATM card, but not for other accounts? Many ATMs limit the number of failed entries for a given card, eating the card if a user enters an incorrect password four times in a row. This reduces the chance that someone would be able to guess your password.

What are the chances, exactly?

Suppose an ATM limits the number of failed password entries to four, after which it will eat the card. Let’s calculate the probability of guessing a four digit password in four tries.

First, we need to know the number of four digit passwords that can be created from a keypad that includes numbers from 0-9. There are ten usable numbers, each of which can be used as the first, second, third, and fourth digits. That means there are ten ways you can choose the first digit of your password, ten ways to choose the second, ten ways to choose the third, and ten ways to choose the fourth.

These ways multiply to give us 10*10*10*10 = 10,000 different four digit passwords that can be made from ten numbers. Your password is one of those 10,000 passwords. The probability, then, that someone could guess your password in one try is 1/10,000 = 0.0001. The probability that someone could guess your password in four tries is additive: 0.0001+0.0001+0.0001+0.0001 = 0.0004.

Some perspective

If you’re like me, you need some way to interpret this risk. We know 0.0004 is a small number, but can we do better than that? To put that figure into perspective, we can calculate the expected loss (a term that describes the probability of an event multiplied by its cost). Consider the following events: you lose your ATM card, your card is found by someone who tries to extract cash from your savings account, and you don’t notice that your card is missing for a week. What is your expected loss in this case?

We start by calculating your maximum possible loss. Someone who correctly guesses your ATM card password would be able to withdraw or spend up to your daily limit on each of the seven days you are unaware of your missing card. Let’s say this limit is $3,000 and the person who has your ATM card knows it (perhaps he learned it by starting with an attempt to withdraw $5,000, then trying smaller and smaller amounts until the machine capitulated). Your maximum loss in this instance is 7*3,000 = $21,000.

Now all we have to do is multiply the maximum loss by the probability of experiencing that loss. We get 21,000*0.0004 = $8.40. You will probably agree with me that this is no big deal compared to the other threats you face. It’s too bad that you are sometimes forced to use shorter passwords than you would like, but at least in this instance, it’s not worth worrying about.

The Economist, reporting research by Symantec, has an interesting chart of the most common goods and services offered by cybercriminals.

You can use the prices on the right of the chart as a sort of risk indicator: if a criminal steals your bank account details, you can expect to lose the amount another criminal is willing to pay (plus the value of the second criminal’s time) to get those details. Keep in mind that these values represent the average (mean) amounts victims will lose and criminals will gain - in reality, some victims will lose a lot more and some a lot less.

Most interesting feature of the chart: email passwords sell for more than full identities. If you think your email password isn’t very valuable, you should know that cybercriminals think otherwise!

The previous post discussed the amount of money you ought to be willing to pay per year to avoid getting phished. By using statistics about the average cost of phishing and the probability of experiencing it, it was possible to come up with a meaningful figure. Given the right statistics, this type of analysis is possible for any type of risk.

What Should You Pay to Avoid Viruses and Spyware?

In Consumer Reports’ 2008 State of the Net summary, the odds of contracting a serious computer virus problems are given to be 1 in 7, the yearly costs $2.9 billion. The odds of a serious spyware problem are 1 in 14, with a yearly cost of $3.6 billion. (Note that these figures are for both businesses and consumers.)

From these statistics, it is possible to calculate the amount that the typical person ought to be willing to pay, yearly, in the form of insurance or a preventative product or service, to avoid the consequences of viruses and spyware.

If 1 in 7 computer users had major virus problems, it means that 26 million people suffered expenses of about $110 each. If 1 in 14 computer users had a major spyware problem, it means that about 13 million people took a hit of $275.

Using these numbers and a formula for expected costs (expected cost = average cost per incident multiplied by probability of incidence) we can conclude that the expected yearly loss per person from virus and spyware threats totals $35. Put another way, each of us should be willing to spend up to $35 per year on insurance, services, or products that would shield us from the costs of viruses and spyware.

The Value of Anti-Virus Software

Of course, my calculations could be wrong. But it’s interesting to note that McAfee and Symantec, two of the most popular anti-virus and anti-spyware providers, price their mainstay products at $40, $5 more than our calculation says is reasonable.

Is that extra $5 per year for peace of mind or is it down to overpricing? Or maybe the cost figures that Consumer Reports noted do not include the psychological cost of annoyance and time spent getting rid of viruses and spyware, which could bring the total cost per person higher than what was reported. I’m inclined to give the benefit of the doubt to the millions of consumers who indicate, by their willingness to pay, that a $40 anti-virus solution is worth it to them, but I could be off the mark.

In the previous post, I calculated the cost, in statistical terms, of identity theft for the typical person. But identity theft is not the only danger - what about the risks of phishing?

Consumer Reports, in their 2008 State of The Net report, claims that the likelihood of getting phished this year is 1 in 94, or just over 1%. The total amount lost to phishers nation-wide is estimated to be $2 billion.

Worry or Keep Cool?

If 1 in 94 American adults lost money to phishers, it means that $2 billion in costs were distributed amongst 2.4 million victims. From that statistic, we can figure that the average cost per person was about $835. If your chances of getting phished are 1 in 94, you can expect to lose (in statistical terms) $9 per year to phishers.

Now, knowing that you are likely to lose $9 per year in statistical terms is a bit of strange concept. In any given year, you will either lose a large sum like $835 or nothing at all. It might be easier to think of the $9 per year as something each person should be willing to spend to avoid the consequences of phishing.

For example, everyone in the country could contribute $9 per year into a phishing fund and distribute the money to the victims of phishing. Those who contribute but don’t fall victim to phishing get peace of mind out of the deal. The victims get compensated for what they lose. Everyone wins as long as peace of mind doesn’t cost more than $9 per year. Beyond that, it’s best to take your chances!