[Linux-aus] Grant Scheme: libferris, evolution and libxml

[Jonathan Oxer]

On Fri, 2005-06-03 at 14:34 +1000, Anthony Towns wrote:

[Questions about the libferris grant application]

Like AJ I'd be very interested in answers to all his questions. Feedback
on the grants scheme would be very welcome, and it will be a topic of
discussion at the next face to face committee meeting in a couple of
weeks time. If anyone has input on it please speak up now.

>   2) does it make sense to buy equipment instead of just giving out money?

In this context I would say no. If the purpose of the grant application
is for a specific piece of equipment then it may be appropriate for LA
to buy the equipment and provide it. In this specific case though the
equipment is not the purpose of the grant application. The purpose is to
fund software development, and assuming the grant is approved, what the
developer does with the funds shouldn't be directly relevant to the
application. I would much rather see this couched in terms of "I want to
do X, and I want to get $Y for it, then I might spend that money on Z".

So we should attach a dollar value to the work to be completed.

Even where the grant application is for the purpose of obtaining
equipment, it would probably still make more sense for the grant
applicant to request funds for purchase of specific items and LA then
disbursing the necessary funds. LA shouldn't be sending people running
down to the local PC store to buy some gear and ship it off: the people
that need it can make the purchase, LA just covers the cost.

>   4) does it make sense to give the money before the work's done, or 
> would some sort of after-the-face prize scheme make more sense?

I expect both will need to be possible. Where the grant is to cover
third party expenses incurred by the grant applicant it may be necessary
for LA to cover the expense directly if the applicant does not have
sufficient funds available personally. Other times it should be done on
presentation of receipts.

In the case of funding a developer to spend some time, as in the case of
this application, I expect it would be done after-the-fact and once the
required outcomes are demonstrated to have been met.

It would also be necessary to be very careful with outcome definition
and assessment, otherwise we would have situations like the GNOME
bounties that resulted in many very badly implemented solutions claiming
bounties just because they met a couple of bullet-point requirements in
very specific circumstances, but weren't actually adding the value to
the project that was originally intended.

Cheers   :-)

Jonathan Oxer