The Trouble With Phishing

Any person who is familiar with even the basics of modern computer threats will know the term phishing. It is an example of the more generic threat known as social engineering, or using psychology as a primary attack vehicle. In general, people tend to be trusting and helpful (although, of course, we can all quickly bring to mind those who are neither). Phishing and other social engineering attacks make use of these traits to trick computer users into giving up valuable information, from user names and passwords to credit card numbers.I could cite a number of recent articles on Dark Reading or other sites referencing phishing attacks. Antiphishing features are present in nearly all of the current batch of desktop security products. Without a doubt, people are concerned about being phished. It is scary to think that somebody could so easily fool you into revealing private information -- and that you wouldn't even realize you had done it.

However, a recent paper from some folks at Microsoft Research (PDF) takes an excellent look at phishing from an economic perspective and concludes it isn't necessarily as dangerous as we might think. The thrust of the paper looks at phishing as an example of a field suffering from the tragedy of the commons. The tragedy of the commons is a principle in which a particular resource suffers rapidly diminishing returns as more and more people start entering the field, ultimately destroying it although everybody may be hurt by the outcome. This is not the case with every resource or, in fact, most resources. In many fields, such as the security software business, overall profit from the field has increased with the growing number of entrants.

As this paper argues, however, the field of phishing is much more like the more conventionally spelled fishing. As cost of entry into phishing has decreased through the introduction of automatic phishing kits and such, the profitability of the field has dropped dramatically. Looking at anecdotal data would not immediately lead to such a conclusion, since it is certainly true that the occasional big score happens. The paper takes a scientific approach and looks at the validity of various studies, attempting to correct the picture of phishing's profitability. Its estimate comes out to approximately $61 million annually, far lower than is typically indicated.

Does this mean phishing is not a significant problem, or one that warrants our attention as security professionals? As the authors point out, the dollar losses are perhaps not the most significant factor when it comes to the damage inflicted by phishing. The loss of trust in online transactions, and the loss of trust in e-mail, is quite likely to be more significant in the long run.

So how do you address that? The vulnerabilities that lead to phishing are, it seems to me, not failures in implementation as much as a failure of design and psychology. The design problem is, I think, at its core a problem of backward compatibility. Protocols designed for the trusting world of the early Internet have gone through many revisions, but they all maintained backward compatibility. As this article discusses, keeping things backward compatible is a great way to keep costs high and slow long-term growth. Perhaps it is time for retiring the support for representing an IP address as a single 32-bit integer, which is one of the classic ways to obfuscate an address in a phishing attack. And, of course, the world is rife with attacks on DNS that allow redirection of connections to legitimate addresses. These outdated designs, combined with the trusting nature of people and the desire to just get things done rapidly, make phishing the threat it is today, even though it isn't a money earner for the bad guys.

- Nathan Spande implemented security in medical systems during the dot-com boom and bust and suffered through federal government security implementations. Special to Dark Reading