Gnupg-users Digest, Vol 91, Issue 30

[Grant Olson]

On 4/19/11 3:17 PM, Mike Acker wrote:
> On 04/19/2011 14:35, gnupg-users-request at gnupg.org wrote:
>>  Maybe because, since this is the support list for GnuPG, we are all
>> thinking more about how to protect an encrypted file than about how to
>> protect a server account.
> relevance?
> 
> what difference does it make if I am discussing a server logon or the
> password for a .zip?  3 strikes, you're out would be good on the server
> but for the .zip the delay after bad makes more sense
> 
> if i delay responding to a bad password for 1 second the speed of your
> processor become irrelevant: you now need 1000 vm's to get to 1m
> tries/sec. and there's no real reason i wouldn't make it 10 sec after
> the 2d bad try, and then 30 min after the 3d -- like the Novell server
> used to do
> 

For an OS or some rpc call, a three strikes rule makes sense.  An
attacker is attacking from an outside system, you still have control of
your system, and the login is a barrier between the two.

But an encrypted file can be on the attacker's system.  We could
conceivably add a three-strikes option to gnupg, but since the OpenPGP
standard is published and gnupg is open source, a malicious user could
just write their own program that doesn't have a delay, or run a
modified copy of gpg.  It's the same with a zip file.  You can't enforce
the rule in any reliable way.

-- 
Grant

"I am gravely disappointed. Again you have made me unleash my dogs of war."

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 570 bytes
Desc: OpenPGP digital signature
URL: </pipermail/attachments/20110419/f7a7f4e1/attachment.pgp>