Re: It's all about trust
On 27/10/08 at 16:01 +0100, Raphael Hertzog wrote:
> Power-hungry might well be interested by this role but when 20 other
> people have the same power, you really don't gain much with your power.
> Unfairness is difficult to avoid as all human judgment have a subjective
> part… but I don't see why this would be exacerbated with this proposial
> compared to the status-quo or to the N-advocations model.
Being one of 20 super-DDs picked amongst ~1000 DDs is still a nice
distinction to have.
> > Since apparently, the NM process doesn't allow people to trust new DDs
> > anymore, I would prefer to move to a system where trust comes from the
> > fact that a large number of normal DDs advocated someone (like what Lars
> > proposed).
> Several (possible) problems with this approach:
> - increasing the number of advocations by DD to increase the trust has a
> real cost. A contributor will usually have interacted with very few
> sponsors that can advocate him at little cost since they already
> reviewed his work. The other advocates will then have to review the work
> by themselves to gain the required confidence. Chances are rather large
> that we will have DD who won't do that job and advocates people just to
> reduce the backlog of applicants that look motivated.
I agree that it would be impractical to require that, for each
applicant, at least 20 DDs have personally reviewed his work. However,
the connections don't have to be direct connections. If both DDs A and B
advocate someone, and I trust A's and B's judgement, I could advocate
the applicant simply because, if A and B advocated him, he must be OK.
Of course, if I state that I advocated him because A and B advocated
him, another C, who trusts me but doesn't trust A and B, will probably
choose not to advocate the applicant (or to review his work).
The current problem is that some DDs, that some of us consider
untrustworthy, advocate people. And then, we don't trust those people,
because there's no chain of trust between us and them.
> - finding many advocations to grant a right might be doable, but removing
> a right is much less fun and will never happen in practice if it follows
> the same rule of a "large numbers of DD"
> - this process might be too heavy with fine-grained privileges as it would
> require the intervention of many DD each time we have to grant a right
> (when trusting the decision of 2 members with special rights would be enough).
That's why I don't think that it's a good idea to play with a lot of
different privileges. Privileges should be granted when they are
necessary to do something, and if you are a DD, you should be able to
easily get any privilege you need to do your work. You are already
trustworthy, so there's no need to double-check everything you do.
| Lucas Nussbaum
| firstname.lastname@example.org http://www.lucas-nussbaum.net/ |
| jabber: email@example.com GPG: 1024D/023B3F4F |