Re: Plain text passwords--necessary

[smb () RESEARCH ATT COM (Steven M. Bellovin)]

In message <199904191510.LAA03916 () Iodine Mlink NET>, Phillip Vandry writes:
> > First, plain text passwords are being used is places where they need not
> > be. For example the recent post about the Real Media server storing
> > plain text passwords. There is no reason for the server to store
> > plain text passwords. It can store a hash and authenticate users
> > against the hash.
>
> It's the old PAP versus CHAP debate. *YES*, there is reason for the
> realmedia server to store the password in plaintext (although it
> should still obfuscate it to prevent accidental viewing). I always
> like to compare the types of PPP authentication to show this:
>
> Method  Client     Wire       Server
> ------  ---------  ---------  ---------
> PAP     Clear      Clear      Encrypted
> CHAP    Clear      Encrypted  Clear
>
> And I don't think we can do better than that. We can encrypt at only one
> stage of the process. We have to make a tradeoff.

It's certainly possible to do better -- there's a whole family of protocols
that do that.  See, for example, /http://www.research.att.com/~smb/papers/aeke.ps (or .pdf), which gives
encrypted on the wire and at the server.  (The predecessor paper is
http://www.research.att.com/~smb/papers/neke.ps)  There are related
protocols by others.