[Soc-coordination] Google Summer of Code 2009: Debian's Shortlist

[Erich Schubert]

FWIW, as I read the discussion about the same topic that happened on the
GSoC lists, the common approach here seems to be:

[Quoting Leslie Hawthorn, who is the key contact at Google = official]

Have the organization rank all other student proposals as though you had
never submitted one. Have them evaluate how many proposals are of such
high quality that they'd like to see them be funded. 

Wait for us to publish slot allocation numbers. 

If the number of proposals that they'd like to see funded is less than
the number of slots allocated, drop out of the application process. It's
therefore completely fair to everyone. 

E.g. 

Your organization has 6 very strong proposals and 4 good proposals. 

Your organization is allocated 6 slots. 

You drop out of the application process. 

or 

Your organization has 6 very strong proposals and 4 good proposals. 

Your organization is allocated 7 slots. 

Assuming your proposal is actually better than one of the 4 good
proposals - and as a domain expert, I'd expect it to be - then yours
could be the 7th project funded. 

I realize everyone is going to have differing opinions on how this
should work. This is how I think you should go about it, should you
choose to pursue applying as a student rather than mentoring. 

---[... back to normal email ...]---

This boils down to giving PRIORITY to getting new, skilled contributors
into the project, but still favors getting funding for an existing
developers project from Google instead of either returning the slot to
Google (*) or giving it to a proposal that we do not consider adequate.

(*) Some people of course claim that this is unfair to other
organizations students, that could have gotten the slot instead.

On a side node, let me quote two FULL *successful* applications for NMAP
according to the GSoC 2008 website (i.e. last year):
---
Title: Feature Creeper / Bug Wrangler
Abstract: I intend to complete small projects that don't take an entire
summer to finish and to fix bugs that come up. One possibility is adding
raw IPv6 scanning to Nmap.

Title: Feature Creeper / Bug Wrangler
Abstract:
This application is for the miscellaneous bug fixer and small feature
adding position. To implement such things as --tcp-ports and an OS
fingerprint assistant. Due to the nature of this task, not to much can
be set in stone for an Abstract.
---

At least one of these 'students' had been in (with the same 'title') the
previous year as well. So at least for me it seems that it is quite
common to just use the funding of Google to get some WORK done, not
necessarily to attract NEW people. We might just have too high
standards ... Not that we should go all the way to NMap, I always found
it very annoying to have to turn down some good applications because we
didn't have any more slots, and seeing one-line applications get through
at NMap... but I can definitely live with seeing a fellow DD *student* -
that is still a requirement! - being funded for a worthy Debian project.

Regards,
Erich