[cryptography] Security Pop-Up of the Day

Joe St Sauver joe at oregon.uoregon.edu
Wed Sep 21 18:20:04 EDT 2011

Chris Palmer commented:

#> Well, its obviously not quite that easy yet, but users can currently get
#> a free client cert by visiting a web page and filling out a form, and
#IanG's point was that there should be no web page, no form. You know
#how sshd generates a host key when there isn't one yet? Like that.

Understood that would be the "zipless" ideal, but how would the binding
of the private/public keypair to the email address occur then, eh?
I suppose that if the web email provider was also the RA for the client's
cert, they could handle tieing the two together at account provisioning
time, but:

1) Would you be comfortable that the private key that's part of that
key pair is adequately secure if thats the model were to be deployed? 

2) That point may be moot, since I doubt we'll be seeing that sort of model
happen any time soon (although, in a way, note that many email providers 
are already transparently offering weaker assertions about the origins
of the email they emit via use of DKIM -- but clearly, DKIM does not offer 
a foundation for true end-to-end payload encryption)



Disclaimer: all opinions strictly my own

More information about the cryptography mailing list