Re: Unit 193: Declaration of intent

2016-07-25 Thread Gunnar Wolf
Holger Levsen dijo [Tue, Jul 26, 2016 at 01:58:49AM +]:
> > But, again, find two DDs with active keys in the keyring with personal
> > policies different than mine, and I will accept it. Hell, I won't even
> > be able to know about it :)
> 
> while I appreciate that you accept keys signed with other policies than
> yours, I don't think keyring maintainers should be willing to accept
> *all* signatures done by DDs. 
> 
> (And I do see the problem that you cannot know everything…)

Ack

> But still, if you *hear* some signatures have been done under fishy
> circumstances, I *do* think you should object.
> 
> Else I^wsomeones may be tempted to try to game the system…
> 
> IOW: please don't state you'd be willing to accept *any* signatures done
> by two DDs… maybe just adding a single word and saying "you'd *almost* be
> willing…" is enough to make the difference I think is important here.
> 
> I fully understand your POV but if I were to take a similar stance,
> namely "I will sign any key presented under any ID to me, because I have
> no means whatsoever to properly verify IDs anyway" and if there then
> were several DDs with that policy… I dont think that would be good. And
> it would be worse if our keyring maintainers were to accept those IDs
> into Debian.

Just for the record: I agree with Holger here. We expect every Debian
Project Member to be a responsible user of their key, and that
includes not blindly signing anybody's key. We do not currently have a
policy on what to do in the event somebody is known to misuse the
trust model we work with, but I think that a key used improperly could
very well be treated as if it had been compromised, and removed from
the keyring (as happened many years ago, when many people were storing
their .gnupg/secring.gpg in project machines).

In short: The system can be gamed. We assume good will and best
intentions from Debian people. Bad intentions will be punished.


signature.asc
Description: Digital signature


Re: Unit 193: Declaration of intent

2016-07-25 Thread Holger Levsen
On Mon, Jul 25, 2016 at 11:46:45AM -0500, Gunnar Wolf wrote:
> [...] Signing an identity must
> mean that you verified the identity in a nontrivial way. Signing
> somebody you have not directly interacted with at all is wrong in my
> eyes.

I agree.

> But, again, find two DDs with active keys in the keyring with personal
> policies different than mine, and I will accept it. Hell, I won't even
> be able to know about it :)

while I appreciate that you accept keys signed with other policies than
yours, I don't think keyring maintainers should be willing to accept
*all* signatures done by DDs. 

(And I do see the problem that you cannot know everything…)

But still, if you *hear* some signatures have been done under fishy
circumstances, I *do* think you should object.

Else I^wsomeones may be tempted to try to game the system…

IOW: please don't state you'd be willing to accept *any* signatures done
by two DDs… maybe just adding a single word and saying "you'd *almost* be
willing…" is enough to make the difference I think is important here.

I fully understand your POV but if I were to take a similar stance,
namely "I will sign any key presented under any ID to me, because I have
no means whatsoever to properly verify IDs anyway" and if there then
were several DDs with that policy… I dont think that would be good. And
it would be worse if our keyring maintainers were to accept those IDs
into Debian.


-- 
cheers,
Holger, with no clear signing policy… (I mostly only sign keys
from people I know offline+online, but I do make
frequent exceptions from that… and what means knowing a
person anyway…)


signature.asc
Description: Digital signature


Re: Unit 193: Declaration of intent

2016-07-25 Thread Gunnar Wolf
Christian Kastner dijo [Wed, Jul 13, 2016 at 07:45:36PM +0200]:
> On 2016-07-13 10:16, Filippo Rusconi wrote:
> > To the community : Should we accept pure pseudo "identifications" in
> > Debian ?
> 
> Personally, I wouldn't care about the pseudonym, but I would very much
> care about an in-person meeting.
> 
> There was a longer thread about this on -project last year:
> 
> https://lists.debian.org/debian-project/2015/02/msg00017.html

Absolutely. But signing somebody's ID is done out of personal
policy. I always make clear to people how my personal signing policy
differs from some other people's (and why I am sometimes flexible upon
what I say, if the reality pushes that way). Signing an identity must
mean that you verified the identity in a nontrivial way. Signing
somebody you have not directly interacted with at all is wrong in my
eyes.

But, again, find two DDs with active keys in the keyring with personal
policies different than mine, and I will accept it. Hell, I won't even
be able to know about it :)


signature.asc
Description: Digital signature