[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: New-Maintainer - Some Facts

On Tue, Jan 16, 2001 at 06:15:08PM +0100, Gerfried Fuchs wrote:
> On Tue, Jan 16, 2001, Marcus Brinkmann wrote:
> > Regardless of the specific time delays, it is of concern to me that we have
> > only one DAM (it doesn't matter who he is or how much time he has). Even if
> > this DAM had all the time of the world and would do nothing but to process
> > applicants, it is important that we have a reserve in case James becomes
> > suddenly unavailable. It is important that we have somebody who knows how to
> > continue the work then, and why not have this person help out before that
> > critical moment, too?
>  Well, it was pointed out quite a lot of time that the DAM is one of the
> most critical - or let's say most worthful and most to be trusted -
> position in the project.

It's not worth more or less than any job, there are no rewards. However,
especially because it is such a critical job it should not hang on one
person only.

> In fact, it's the person that is responsible
> for all of the developers, in some way or the other.

I beg to differ.

> And now try to find a
> person that you'd trust enough to give the same "power" to.

To be honest, among all Debian developers, James would not be my first choice
probably, but maybe that's only because I don't know him. He grew into this
job, that's why we know he can be trusted to do it. There is no magical
"trust" value that can be applied to persons. A bank account has a certain
balance, a thermometer has acertain temperature, but "trust" is not
measurable. As long as you don't give anybody a chance to proof that he is
trustable, you won't find someone who is.

That's a critical point that everybody who claims that there is no trustable
person is missing. There can't be anybody until you start to trust. Trust is
mutual, it doesn't come from the person you want to be trustable, but
from yourself, you have to trust. You can only trust with confidence, and
confidence grows with experience.

There is nobody growing into this position currently. There is only one
person now remaining from those who grew into it. What we will do when we
loose this person in this position? We will have nothing, and will have to
start from ground up. Is this the situation you prefer?

> It's not like you can mail to -private and say "Hey, someone likes to
> become DAM?"

Why not? An offer like this is not a guarantee that you get the job. You
have to proof to be trustable over time. I am not talking about posting the
root password to debian-private.

> Don't start whining about the DAM, it's just that he might
> read you and get distracted by your mail from the approval job or
> something else....

I am not starting to whine, thank you very much. (Also, you seem to have a
low opinion of the DAM if you suggest that it is so hard for him to focus).

[You see? I can play this game, too. Let's just continue to discuss
this in a rational way, ok?]
> > More than one DAM also distributes the individual waiting times more
> > unifromly among the waiting time interval (because supposedly all DAMs work
> > independently of each other), so it will be a more smooth operation. We
>  I wouldn't like a DAM that works independently from the others.

I meant "independently in time", sorry if this was not clear from the
context. You can think of independent probabilities if this helps you.

The DAM can not work independenly, he should follow the consensus (see
constitution). In this case, I think the consensus is quite adequately
expressed by the recommendations of the AMs (as a whole).

>  I
> would say, yes, there can be more DAM, but more than everywhere else
> they should still make the decisions together than each one for himself.

In the usual case, a little supervision will be sufficient. I think it is a
good idea for the DAMs to organize themselve, and discuss rare controversial
cases among them, as they did in the past.

> > don't have enought numbers to know this for sure, but I think part of the
> > problem might be that the DAM is processing people in bunches, so some wait
> > long and some not at all, resulting in a good average time. But as I said,
> > this can not be concluded from these numbers.
>  Compare the average to the median.  Average can easily be distracted by
> a single high (or low) number.  The median should satisfy you more.
> It's on the front page of <http://nm.debian.org/>, there's no magic
> behind this statistics.

Knowing the median is still not enough to tell much. There is no magic in
statistics, but it requires a lot of careful work to derive meaningful
conclusions from numbers, and people are easily trapped by common
misunderstandings. Often, more results are derived than technically
possible. (Probably because we want to know more than we do).

> > In fact, I find it hard to derive any useful information from the
> > numbers you posted[1]. Just as an example, they don't include the
> > amount of work required for each step. Working through the application
> > process is time intensive (as it depends on the communication between
> > two people), while creating an account is probably time intensive
> > sometimes and often not (James said this depends partly on the review
> > created by the AM). This suggests that the 29 avg time for DAM
> > approval weights more than the 43 avg application process time.
>  In fact you should also here compare the median, which is 27 to 8
> currently.

This seems to suggest that my guess of processing in bunches is correct.
That's all I can read out of it, and even that is far stretched.

Thanks for the info about the AM processing time.


`Rhubarb is no Egyptian god.' Debian http://www.debian.org brinkmd@debian.org
Marcus Brinkmann              GNU    http://www.gnu.org    marcus@gnu.org

Reply to: