[Soc-coordination] let's look for _new_ Debian contributors!

[Obey Arthur Liu]

On Mon, Mar 15, 2010 at 10:33 AM, Stefano Zacchiroli <zack at debian.org> wrote:
> Let's attack a recurrent GSoC subject, while we are still on time.
>
> On Sun, Mar 14, 2010 at 11:13:25PM +0000, Obey Arthur Liu wrote:
>> The program is a unique opportunity for Debian and eager to learn
>> students to meet each other and work together for the duration of a
>> summer. Often this has been the beginning of a great relation,
>> bringing fresh ideas and enthusiasm to the Project.
>
> Indeed!  So, as a corollary, it is pointless to hire students that are
> already active Debian contributors. That at least is my personal
> feeling, even though past discussions show that it is shared by a good
> deal of project members.
>
> I don't want to rehearse here and now the pro and cons of allowing
> pre-existing contributors, the interested reader can find both positions
> detailed starting from the pointers below (kindly offered by Anthony
> Towns in a recent, though unrelated thread on -vote):
>
>    http://lists.debian.org/debian-devel/2009/04/msg00424.html
>    http://lists.debian.org/debian-devel/2009/04/msg00431.html
>    http://lists.debian.org/debian-devel/2009/04/msg00441.html
>
> My desire is that we take a decision, so that it is clear what is our
> position on this. As a start, I believe that the Debian GSoC admins have
> the right to decide on this, consulting whoever they might like (this
> mailing lists seem like a good start, though). Once they have a
> position, that position can be communicated to -project (looking for
> vetos?). I don't mind discussing that on -project directly neither, I
> simply don't volunteer to follow the thread there, so I'm not posting
> there in the first place.
>
> An important degree of freedom in making the choice is what does qualify
> someone as being a "Debian contributor". Of course it is a gray area,
> but given that we have the two clearly defined status of DM and DD, I
> believe we should state something like "DMs and DDs cannot apply as GSoC
> students" (or the contrary, of course, if the choice will be that we do
> want them).

My wording in the announce mail implied that I was interested in new
contributors.

We can get multiple justifications for wanting non-DDs. Here's an
accountant one:
Basically, from the pool of GSoC students that had no contact with
their organization before, the retention rate is about 1/4 to 1/3. Our
rate is in this norm, meaning we gain about 0.3 long-term contributor
par student.
If we look at our track record of selected students who were DDs, I
believe it is pretty clear that their increased long-term contribution
rate due to GSoC is well below 0.3 (if not zero for many). Therefore,
it doesn't make long-term economic sense to choose DDs.

Why only consider long-term returns? We get about 10 students each
year and spend an awful lot of time, effort and money (although not
entirely ours) on each of them (try comparing it to the help a new
contributor would get by going to debian-mentors). We could try to get
short-term pressing things done, or groom students into full
contributors, and still have them get short-term pressing things done
in the future. The GSoC has so far worked well to efficiently train
contributors in a short period, something which other channels have
not been able to.

I know that some organizations select long time contributors as
student. There exist multiple reasons they do so. It is a way to
reward very active long-time contributors or to guarantee them some
free time by relieving them of monetary needs for a short time.
The important thing to note is that most of these cases happen in
small organizations who have no sponsoring otherwise (like we have
from HP and countless others) or where a single developer can have a
significant impact on the entire project.

We already have sponsoring through multiple channels (hardware,
debconf, etc) so, generally speaking, the average perks of a DD are
much better elsewhere. A GSoC stipend would have much less impact.
Just imagine if you, as a DD, had to personally rent buildd hardware,
bandwidth, debconf housing... You can't compare the help that a GSoC
stipend is to devs of a 3-persons organization to that of Debian.

Secondly, while we have contributors whose participation in certain
task is critical enough because of their knowledge on particular
areas, I don't think any single of them can and should warrant the
diversion of a GSoC student stipend just because of this. Debian is
already too large for this to happen. In a market economics sense, the
offer large and interchangeable enough to not have to treat them
differently.
I know orgs who choose to hire some core developers on donation money
for critical work because this is the only way to get it done
properly. Debian is not in this situation.

To conclude, my position, and I believe it is shared by the majority
of the mentors who actively participate in Google Summer of Code at
Debian, is that we should select those students who will most probably
give us the largest return in terms of *increased long-term
contributions rate*. That means we will strongly prioritize new
contributors.

I don't feel like a formal vote is necessary, partly because I'm lazy,
partly because I think the results are very predictable and partly
because I don't think we need the political backing of a vote at this
point. If some more people feel otherwise, feel free to reply how and
why and I'll organize something (and especially if you donate a
project idea to the wiki page :) ).

Cheers

Arthur