Policy-Discussion

[Ian G <iang AT cacert.org>]

I always enjoy responding to Peter's emails, they are an untaxable fringe benefit of being in the policy group, a pleasure so immoral it would be criminal in any other forum :)


On 20/02/13 19:24 PM, Peter Williams wrote:

In your typical public CA, there is appointed a specially-trusted individual. Perhaps call her the policy officer. She is trusted “to do the right thing” - and use her non-reviewable power to manipulate the rules “when the times comes”. She is indoctrinated to obey “at that time” - and otherwise do hum drum stuff of any board secretary. Perhaps she is a he, for all it matters.

 

If you want a historical analogy, s/he awaits Herod’s order to kill every male child (some one of which threatens Herod). And off s/he goes a-killing, without a qualm - for no other reason than Herod commanded it. The main mission is to find the one turncoat who would spite Herod (and save not just a male child, but the very one he was after...). This is why s/he was made ‘policy officer’ - placed to gather the intuition on who could not be trusted “when the times comes”.

 

So the challenge to CAcert is to break the tradition. Solve the problem (find  a disciplined person, who enjoys organization); but prevent the “trust” cycle repeating. It comes down to ensuring transparency and ensuring that the policy officer has no power to act in an covert setting or set a rule that introduces a special caveat (the covert setting rule-making power).

In policy group, we have addressed the Herod's problem with teamwork and a PoP that was modelled loosely after IETF experiences.  A few of the early guys were familiar with IETF's working group model, and certain things about the model were good:  rough consensus was the metric, along with open entry to all in the community.

To some extent, though, IETF was a scary model, and the IETF is tarred with the responsibility for much of what is bad in Internet security (and there's a lot to be responsible for!). Their model tended to promote cartels at war with each other.  To address this, we were quite careful to not create capturable bastions.  The war may go on, but it is open and dynamic, and no territory is holdable.  This helps to focus attention on the issues not the symbols, which unfortunately IETF security is riven with.

One of the 'bastions' was the powerful role of policy officer.  If you look closely at the PoP and any other policy, this so-called Officer has very little power.  She or especially he can't go off and rewrite things, nor block things nor advance things, far less run around and decimate a generation under 5.

According to the wording the policy officer can push some papers around, and can manage a mailing list.  He or she might count decisions, but only 'decides in minor issues' which means nothing because the real authority to rule here is 'rough consensus' which the policy officer has no more power to push around than anyone else.  Everyone has more power to call 'foul' than the PO has to make something happen.

In contrast, in the IETF, there are powers to make and break, and sometimes power has been used to unfortunate consequences.

The policy officer cannot refuse to hold a meeting, cannot refuse to count a decision, and cannot fake a decision.  All for various reasons:  the meeting is here on a maillist, anyone can write the wiki, and if anyone fakes a vote, anyone else can review the record.  (Yes, been there, got the t-shirt, recorded in PolicyDecisions....)

Yet, surprisingly for all that, people tend to trip over the symbols.  And we make mistakes in painting too bright a colour at times.  Hence my post of yesterday - let's wind back to teamwork and joint responsibility, and not lean on the crutch of mystical symbols and authority.  That change to PoP disperses the job of minor editing to the Editors, and plural -- in reality nobody cares two hoots who does the changes, but that they are done.

We have our policy group to see that they are done reasonably, and beyond that we have a community that reads policy and Arbitrators that can rule against.

CACert would be the first CA to succeed at this, if it does. All the others succumbed to relative trivial financial inducements to introduce the “appropriate” trust regime over rule making.

Well, one wonders whether they got value for their money, but I'm sure there are some more chapters to the story :)




iang