[Soc-2006-dev] google feedback

[Tom M]

Hi Kent,

semi outsiders opinion...

> Things that really seemed to help this year that we did not do last year.
>        Weekly reports from students.
>        blenderartist.org links for projects.

I would agree, the weekly attendence of the group meeting (ie where
all of the core coders are available to ask questions, give feedback
and get asked questions) were also valuable and I'd recommend for
wider adoption.

> I think that separate cvs branches for each project are useful but I also
> think that improving that would be on my wish list....

Switching to subversion might be the best solution ultimately.

 > It might be worth it to make sure the weekly reports are done on their wiki
> pages.

yep good idea

> I like the flexibility Google gives us, and I think if they had some sort of
> central revision control tree for soc projects, that it would just be more
> additional work for students and merging with current cvs would be more of
> an issue than it is now.

Agreed.

> Anyone have anything else to add?

Can't speak to the pre mentoring process issues experienced but have
some ideas.  Something I was thinking might be useful to reduce the
work of mentors in the pre acceptance stage would be an easy way to
divy up the workload so that each mentor isn't reviewing each project.
 Perhaps every mentor is randomly assigned x% of the applicant pool
with the pool overlapped so two mentors per project.  Each mentor
votes on the applications they've been assigned, but can also vote on
any other applications they find interesting.  Then the applicationst
that are voted above a threshold, or that have conflicting views (ie
one vote high, one vote low) everyone votes on.

It wouldn't reduce much work in the case of most applicants being of
high quality (which was apparently the case this year) but it does
help in the case where you have lots of low quality applicants to weed
out (ie the case last year).

Other issues - perhaps some mention should be made about the initial
selection and allocation process - the current method based on pure
quantity of applicants seems to be a less useful approach, although
admittedly I can't think of an efficient and easy to administer
replacement.

LetterRip