On Fri, 16 Mar 2012 11:23:06 -0600, Gunnar Wolf <gwolf@gwolf.org> wrote: > Moray Allan dijo [Thu, Mar 15, 2012 at 10:06:34PM +0000]: > > > Select people that are, initially farthest from me (or from all the > > > team members) in the web of trust, probably in small batches, as ding > > > this one at a time will take too long. > > > > > > > I don't have an opinion on whether this bit is sensible without seeing what > > it means in practice > > > > (I suspect, due to mass keysigning, that it will come close to giving the > > people with greatest overall mean-shortest-distance, which could mean > > people who least enjoy attending events themselves, or perhaps people who > > live in the most remote locations. If the results are too warped that way, > > it might work better to select people randomly, with weights proportional > > to, say d^alpha where d is this distance measure.) > > I agree with you here... Although the GPG WoT does carry *some* > measure of whom does each person know, "knowing" is a very subjective > word. I think we are unlikely to get much better results than from a > random sample. I was hoping to make it a random selection, but with a bias towards people who have not attended previous DebConfs (as it will be unlikely for previous attendees to have remained at the rim of the WoT). Then, by asking those people to nominate someone, we bound to get results closer to the the middle of the social network (note: not the GPG WoT, but the social network) so we'll have people that at least one person trusts, and that knows at least one "outcast" an might know more. That sounds like a better candidate than a purely random selection, but is still pretty random, so we should get people that are diverse by whatever measure you happen to care about -- maybe it won't work, but I think it's worth a try. > > > Ask the nominee if they're willing, and not in need of sponsorship > > > themselves. > > > > As I've said in the past, I think the > > not-applying-for-sponsorship-themselves point is sensible/important. It > > should perhaps also be applied to the victims from the previous stage, so > > that they aren't influenced to choose people likely to favour themselves. > > We have in the past decided not to vote for/against ourselves. It seems to me that we have a choice. We can either come up with an increasingly prescriptive set of rules for defining who gets funded, and then have a committee mechanically apply those rules in an unemotional and robotic manner, and fix any issues by adding more caveats and adjustments year on year. Which is a bit like what we've done so far, and doesn't seem like a lot of fun for the people that have to do the work. Or we can attempt to assemble a random sample of representative individuals from the diversity that is Debian, and ask them what they think in as light-weight a manner as possible. If you're doing the former, then its probably fine to allow people to assess one another's applications, as there's supposedly not much wriggle room for favouritism. If you do the latter, I think it's clear that you don't want any of the individuals to have a conflict of interest -- although by saying that you are of course biasing the sample towards affluence (or perhaps disinterest in DebConf attendance). Hmm, perhaps one could automatically grant sponsorship to those individuals if they want it, first -- although that would only work once, as once news of that got out it would then act as an incentive to accept the invitation, which would probably be unhelpful. Cheers, Phil. -- |)| Philip Hands [+44 (0)20 8530 9560] http://www.hands.com/ |-| HANDS.COM Ltd. http://www.uk.debian.org/ |(| 10 Onslow Gardens, South Woodford, London E18 1NE ENGLAND
Attachment:
pgpDZDNdA8dbP.pgp
Description: PGP signature