Clientside encryption could be, but clientside encryption is pointless (assuming I'm not running some hurrdurr password collecting scam) because anybody who intercepted the password between the client and server would as good as have the password.
No, anyone intercepting it could NOT as good have the password since they can't reverse the hashing which means that they can't use it to use the account on other sites (assuming same pass, etc).
Interesting, I didn't know that. Subtle difference. I have used AES in previous projects, so I'm used to saying 'encrypted'.
Using AES for passwords would be entirely useless.
So, using the password as a seed? I'm not really sure what the advantage would be, if the user wanted password recovery it would still be impossible.
I assume that you mean "recovering the old password".
No, that would be impossible which is one of the points about it.
However, what you usually do when you use a password recovery feature is that the site sends a new password to your e-mail. That would not be impossible since the encrypted data is already known. What is not known is the encryption KEY. The thing is that you retrieve the encrypted data and the decrypted data from the database. Then you try decrypting it with the password as key and then if it succeeds you compare the decrypted data with the data from the database. If those succeeds, log me in, otherwise, refuse.
Vaguely. Is it that global internet ID thing? I didn't think it was really all that widely used.
Actually some big sites allow both using "their" account system and OpenID. Drupal (which I personally like quite much) ships with an OpenID module by default (although disabled), I'm not sure about how it handles passwords though.
Some big sites (for example StackExchange (StackOverflow, etc) and SuseStudio) only allows login via OpenID.
Also, for example Google and Yahoo acts as OpenID providers which means that any site allowing OpenID-logins can be logged into with your Google/Yahoo account.