You probably know the two technology problems that drive me nuts: email and passwords. They're the worst, the most broken things on the Internet. Google has helped with the email problem, though I still consider email to be well on it's way to obsolescence. Passwords, on the other hand, just keep getting worse.
My least favorite password-using sites are now banks, thanks to some yo-yo legislation that requires a half dozen additional security measures. I would use this as a reason to switch banks, except that every bank must implement something similar by year end. Unfortunately this government imposed mandate doesn't get to the heart of the problem.
The irony of passwords is the more a user needs to remember, the harder it is for them to do so. As a result they do everything they're not supposed to do, including: writing passwords down on sticky notes, using the same password for everything, never changing passwords, and using very simple passwords. This is unavoidable in the current state of authentication: most users will not be bothered to put forth the effort required to secure their logins. You can try to bang it into their heads as much as you want, but it won't work.
That's why the systems used for authentication are so important and tragically neglected. Lots of shops implement their own authentication scheme, even in languages that have decent libraries they could use. The world needs better auth interfaces. Ones that force good security practices on the user by virtue of their design, not by making the user perform extra work. Rewriting this experience over and over is not a great way to achieve this goal.
OpenID is a glimmer of hope in this sea of lousiness. It's by far the best idea yet about how to solve this identification problem. I've written of it before and even use it for comments on my blog, but I recently found this excellent video that doesn't have nearly enough views on YouTube. It is a presentation given by Simon Willison at Google entitled the "Implications of OpenID."
Not only does he do a great job of explaining how it works, but he gets well into the more interesting bits that really have nothing to do with identification. This is the part you should perk up for (starts about 15 minutes in) and then realize that a system like OpenID could attach itself everywhere. I even think it could solve the problems of email by adding an authentication layer atop email clients to verify addresses using message headers. There are numerous other interesting possibilities that Simon mentions.
The video also includes a discussion of OpenID criticisms and general security issues currently faced on the Internet. I especially like the permanent cookie set via out-of-band token mechanism. Anyone interested in web application development and security must watch this video.