SSL and the Security Myth

There is a persistent myth about SSL. The premise is that if you use cryptography to protect communications, all security problems will be solved. This belief is very strong among non-practitioners (the sight of a padlock on a web page is often enough to create a sense of trust) and, surprisingly, among many security practitioners. Even Bruce Schneier believed in it wholeheartedly when he wrote “Cryptography Applied”. He quickly realized that security is much more complex, and as you can see in his later books, his view evolved. We're talking about the mid-'90s here!

Encrypting communications protects the content from the prying eyes of Eve. In the case of web communications, the protocol used (SSL or TLS) wraps the message to protect in an opaque envelope (encryption). Only the source and destination of the communication can see its content, in this case the web browser and the web server (there are more complex scenarios involving proxies and accelerators, but for the purposes of this explanation I'll stick to a simple communication). Incidentally, it is possible to validate the identity of the source and/or the destination. In the web world, only the web server is authenticated, by comparing the domain name in the certificate with the one actually being accessed.

Now, a protective measure exists to guard against a threat. In the case of encrypting web communications, it ensures the information isn't accessible to an unauthorized third party, Eve. So far, so good. That said, you have to put things in context. For Eve to be able to intercept the communication, she has to be able to place herself between the source and the destination. The most common case is through unprotected wireless networks. In that case, it is essential that the communication be encrypted (see Hide that security I cannot bear to see), since WiFi is a shared medium, i.e. all communications are accessible to every device within range or plugged into the shared medium (non-switched Ethernet). For a communication over a network without a shared medium, it is hard to position yourself to see the traffic, and it is flat-out illegal to do so (read section 184 of the Criminal Code of Canada to avoid rotting in prison for 5 years). In these scenarios, you also have to consider the value of the information in transit, because if it's just a comment form, nobody will spend energy specifically intercepting that information. If it's a little-known site, it's unlikely anyone will care. For better-known sites, like Facebook for example, it's likely that some people will be tempted to intercept communications to hijack sessions, as Firesheep demonstrated so well.

Security is a combination of things which, taken individually, aren't much use. Nothing is completely black or white in security.