While working on tech support at my old job, I came across this confusion regularly. Clients fervently believed that an email address was a website. They also ran Windows 97, and didn't need a modem to access the internet.

On one occasion I received an email from a woman complaining that when she went to a website she was prompted for a username and password, which she did not have. She didn't give me a url for her problem website, but she did give me the company name. I did some research, and found that although the domain was registered, it in fact had no website, not that I could find.

Being young, and full of belief in the capability of users (so cruelly shattered in the months to come), I asked her what the URL was.
Her response? An email address. That was one cause of her problem, but then why was she asked for user name and password?
So I did something I had never even thought of doing before - I typed an email address into my browser's address bar.
And was taken to the login page for that ISP's web based email, which prompted me for a user name and password.

Since then I have done some research, and this process seems to be related to the browser used. A telnet on port 80 to the email address fails. If one telnets to the website (e.g. hotmail.com) and then attempts to GET email@hotmail.com) the request will also fail. This means that at least some web browsers have a feature that detects an @ symbol in the URL given by the user, and visits the domain instead. In some cases, if the domain request fails, some subdomains (mail.domain.com, webmail.domain.com) will be tried. The browsers that have been tested for this support are:

Yet another example of lack of knowledge defining trends and practices.