GSoC 2013 procedures?

[rsbivand]

I would like to ask for clarification on the running of GSoC 2013, and that the clarification be copied to the R-Wiki page. I'm asking because I am very confused, and because the threaded nature of this list means that Brian's reply to Tilak Raj Singh may not have been read by the apparent addressees. I have experience of mentoring a project proposed by me, recruiting a student through a test process, and successfully completing in GSoC 2010 (rgeos with Colin Rundel), and I'm trying to repeat that process, but the procedures are not recognizable.

As a proposer of a project, I need clarification on a number of questions.

1) R has been accepted as a GSoC organisation in 2013. This has not been announced here or on the R wiki page. Will this be rectified?

2) Proposals should provide tests for students, but no deadline for preparing tests has been given. How are candidate students to know what they must do in order to enter into consideration?

3) Since as I understand it tests are needed, is it the case that mentors propose and students apply to pass tests set by mentors, or what? The phrasing on our wiki is unclear.

4) How do student replies to tests now, and by when?

I recall in 2010 that organisations were first verified, then got slots, prospective mentors voted proposals into the available slots, and then students were recruited.

The proposal to replace akima has attracted a lot of interest, but I have not followed up approaches I have received so far from interested students off-list, because this reduces transparency. As far as I can see, proposals should have a test with clear pass criteria, and that the test will decide which candidate student is offered the studentship. Test answers should be sent to the prospective mentor rather than the list, but other contacts should be on-list to promote transparency.

I'd be grateful for guidance,

Roger

[Brian G. Peterson]

04/26/2013 03:05 AM, rsbivand wrote:
> I would like to ask for clarification on the running of GSoC 2013, and
> that the clarification be copied to the R-Wiki page. I'm asking because
> I am very confused, and because the threaded nature of this list means
> that Brian's reply to Tilak Raj Singh may not have been read by the
> apparent addressees. I have experience of mentoring a project proposed
> by me, recruiting a student through a test process, and successfully
> completing in GSoC 2010 (rgeos with Colin Rundel), and I'm trying to
> repeat that process, but the procedures are not recognizable.
>
> As a proposer of a project, I need clarification on a number of questions.
>
> 1) R has been accepted as a GSoC organisation in 2013. This has not been
> announced here or on the R wiki page. Will this be rectified?

Google announced accepted organizations, as in prior years.

R was among the organizations accepted, as in prior years.

Many R-related blogs have also commented on R's acceptance.

The Google melange home page for R's participation in GSoC 2013 is here:

http://www.google-melange.com/gsoc/org/google/gsoc2013/rproject

The application template for students is on this page, as in prior
years, and contains quite a bit of information for students about things
needed for a successful application.

> 2) Proposals should provide tests for students, but no deadline for
> preparing tests has been given. How are candidate students to know what
> they must do in order to enter into consideration?

Student Application Deadline:
May 03 at 19:00 UTC

This is the Google denominated hard stop.
(apologies for misreading my calendar yesterday)
With any and all timeline related inquiries, the Google Calendar posted
on melange is the final word.

The test is only one small part of a successful student application.

As in prior years, we've asked students to include test information in
their application where practical, and to communicate off-list with the
proposed project's mentors regarding the test.

> 3) Since as I understand it tests are needed, is it the case that
> mentors propose and students apply to pass tests set by mentors, or
> what? The phrasing on our wiki is unclear.

The student needs to fill out an entire application, not just a test.
Passing the test is a necessary but not sufficient step for consideration.

Project mentors will need to indicate on melange in the application
comments, as in prior years, that a prospective student has passed the
test for their proposed project.

> 4) How do student replies to tests now, and by when?

Before the student application deadline.

For some proposed projects, the requested tests seem small enough to be
included directly in the application. For more involved tests there
will likely need to be some back and forth among prospective students
and mentors, and I would suggest that the code be placed in some public
repository (github, google code, etc.) so that other R mentors can
review the test in addition to the proposed project's mentor.

> I recall in 2010 that organisations were first verified, then got slots,
> prospective mentors voted proposals into the available slots, and then
> students were recruited.

The student application period is part of the schedule dictated by Google.

After the application period closes, no new student applications can be
accepted. This means that interested students should apply soon, and
refine their applications while the application period is still open.

Google has extended the post-application steps this year,so the process
after the application period closes is somewhat longer this year than it
was in prior years.

After the student application process closes, the R project
administrators will request slots from Google based on the number of
project ideas we've proposed as an organization and on the number of
strong applications we have received as an organization.

From the close of the student application period to (at the latest) May
22nd, the entire R mentor community will review, vote on, and rank
proposals. As in prior years, this will happen on melange, in public
comments on student applications where more detail is needed from the
prospective student, and in private comments by the mentors for
discussion amongst the mentors.

Accepted student proposals will be announced via email to the student
and on the Google Summer of Code 2013 site on Monday, May 27.

> The proposal to replace akima has attracted a lot of interest, but I
> have not followed up approaches I have received so far from interested
> students off-list, because this reduces transparency. As far as I can
> see, proposals should have a test with clear pass criteria, and that the
> test will decide which candidate student is offered the studentship.
> Test answers should be sent to the prospective mentor rather than the
> list, but other contacts should be on-list to promote transparency.

I see that you've posted a test for the akima replacement proposal here
to this list. I also recommend that you follow up with students who
have contacted you off-list.

While I see nothing wrong with responding to public inquiries on-list, I
also see nothing wrong with prospective students and mentors
communicating in other, more appropriate venues, whether that be another
R-related list, on irc, or via private email. Many prospective students
have been in touch with mentors for several months now.

I'd urge mentors of proposed projects to offer as much guidance as seems
reasonable to prospective students as the students prepare their
applications. Don't do their application for them, obviously, but
provide information that will help the student write a better
application. My personal policy in prior years, and again this year,
has been to put most project guidance in the R project idea on the R
wiki, but also to reply to specific questions from prospective students
as they come up.

Regards,

Brian

--
Brian G. Peterson
http://braverock.com/brian/
Ph: 773-459-4973
IM: bgpbraverock

[rsbivand]

See also this blog posting. This list and the Wiki should be the prime resources, things shouldn't just float around, really.

Roger