Password reset

David Shaw
Tue Aug 21 18:11:01 2001

On Tue, Aug 21, 2001 at 05:31:01PM +0200, Florian Weimer wrote:

> David Shaw <> writes:
> > Oh, yes, the implementation may make decisions, but the local user
> > always has control.
> There is nothing in RFC 2440 which mandates the possibility of user
> interaction.
Sure, but just because some implementations may not allow the user to modify keys and dates and such, that implementation cannot usually assume that that all implementations do the same. If this locked-down implementation will accept key data from the outside world (say, a keyserver), then someone can still modify the key. What RFC 2440 says or doesn't say is interesting, but not to my point. My point is that there is nothing in the cryptography that prevents this sort of thing from happening. The cryptography *prevents* someone from reading an encrypted document without a key. The cryptography *prevents* a signed document from being modified without also invalidating the signature. RFC 2440 is a social convention between human beings that says "if I'm going to speak OpenPGP, I'm going to speak it this way". Human conventions can be trivially broken (e.g. set your clock back, or remove the revocation). It could be argued that this is a good or a bad thing, but it is nevertheless the case that expiration dates and revocations are human conventions in OpenPGP, and are not backed up by the mathematics. Could there be a controlled OpenPGP installation that only communicates among controlled nodes and never exchanges OpenPGP messages or keys with anyone not under this control? Sure. In such a case, one could expect expirations and revocations to be something more than advisory - but they still wouldn't be backed up by the math. David -- David Shaw | | WWW +---------------------------------------------------------------------------+ "There are two major products that come out of Berkeley: LSD and UNIX. We don't believe this to be a coincidence." - Jeremy S. Anderson