His recently released book, Schneier on Security, dissects issues like data mining, the industry power struggle over controlling PC security, and why some risks are overestimated while others are underestimated.
In this interview, the security guru discusses a plethora of security topics including how to protect your own PC.
What is the single biggest threat to our technological security at this point?
The single biggest threat is the technology itself. Technological systems, especially newer ones, are exceedingly complexand complexity is the worst enemy of security. This is true for a number of reasons. One is that in our rush to build new systems, we generally ignore security or only pay attention to it at the last minute. But the other is that complex systems, especially non-linear and tightly coupled systems, are naturally less secure.
Theres really no solution to this problem; were not going to give up our new technological systems just because of security concerns, but it is something we need to be constantly aware of.
Fear of identity theft seems to be at exceptionally high levels, with constant headlines about hijacked credit cards and bills run up without the account owner's consent. Is the threat from identity theft as bad as it seems?
In the U.S., not really. The extreme cases get the press, but in the main, identity theft is a solved problem. If someone manages to open a credit card in your name, he makes an average of $1,350 in fraudulent purchasesbut youre not liable for that. Your median out-of-pocket cost for new account fraud is only $40, plus ten hours of grief to clean up the problem. This isnt to say that we shouldnt require companies to be more vigilant with our personal information, though. The privacy issues are much bigger than identity theft.
Are there security risks that are far greater than we know? That is, some issues that dont get much coverage but are in fact quite serious?
Corporate crimeboth fraud and espionagegets less coverage than personal crime. Companies have an incentive to keep incidents out of the public eye, so they are more likely not to talk about them. When mandatory disclosure laws were passed a few years ago, we learned that companies were losing personal data far more often than they admitted. Almost certainly they are suffering other damages as well.
It seems as if there's a national passion for data mining, largely in hopes that it will detect terrorists before they act. Do you agree with our apparent enthusiasm for data mining?
Data mining is great for some things, and terrible for others. Its success story is credit card fraud prevention. Right now, data mining systems are looking through credit card transactions, watching for signs of card theft and other sorts of fraud. This works because 1) there is a large data set of attacks to use to generate predictable patterns, 2) criminals tend to do the same things over and over, 3) fraud reduction is easily quantifiable, and 4) the cost of false alarms is low.
Compare this with detecting terrorism: 1) there are very few attacks, 2) theyre mostly different, 3) its hard to quantify what a reduction in risk looks like, and 4) the cost of false alarms is very expensive. So while I have an enthusiasm for data mining as a security tool, its only in areas where it makes sense to use it.