Using money to fund real Debian work
[ Continuing publicly a discussion started in -private, with the agreement
of Pierre ]
The discussion concerns the use of money as a resource within Debian.
Dunc-Tank's principle is to use the money to pay for real work and not
only for travel expenses and reimbursments. This principle is too new to get it
to work within Debian now, people need to familiarize with the advantages and
the drawbacks and we need to find a sustainable model for us.
I also explained that Dunc-Tank's initial experiment of funding release
manager is not a long term model for us. And as a board member, I said
that I don't intend to fund other projects with Dunc-Tank until we have
switched to another model that suits us better in general. Key principles in my
current view of the other model are:
- each DD can register projects for which they'd like to be funded
- all DD can publicly comment the project proposals of everybody (and
hopefully improve the proposals at the same time)
- all donors can donate to any project, but those who have no specific
interest in any of the project would hopefully donate to the most
popular ones (projects could be rated by all the DD)
- and of course, everybody is free to complete one of the registered
projects for free. The fact that project proposals are documented make
that even easier.
In this discussion, on Sun, 08 Oct 2006, Pierre Habouzit wrote:
> when there is a donator, there is two cases:
> (1) the donator just want to thanks debian that he uses a lot, and
> cannot contribute in another way than money because he does not has
> the skills, time, whatever to contribute.
> (2) his work/life/well-beeing/whatever depends upon the fact that the
> task he sponsors has to be done.
> In a wonderful world we would only have money from (1), and such money
> has no smell, and one could dispatch it more easily, as a "resource"
> like you name it.
> But in a real world, a lot of money come through (2): HP that wants to
> see amd64 be supported in Debian, people that want multi-arch (and do
> it in Ubuntu since it seems to be stuck in Debian, etc… In that case,
> there is a — at leat implied — contract between the sponsor — who is
> more an investor — and his sponsoree. Meaning that the sponsor has some
> moral right to expect result.
[.. a paragraph skipped ..]
> So that means that when money comes from (2), people that give the money
> will have some decision power. And money is a bad advisor for us.
The set of projects that can be funded are projects proposed by Debian
developers. I expect Debian developers to propose only projects that are
If some of those projects also serve the private interest of someone external
to Debian, then it's a net win for us to have someone ready to fund the
associated work. The "sponsor" has not gained any decision power, he has only
been accelerating something that a Debian developer already wanted to do.
And of course, the project won't be integrated straight into Debian without
review. If the project changes packages, it will need review and acceptance
of the respective package maintainers.
In the worst scenario, the sponsor will be disappointed and will not give
money any more. But if the rules are clear from the beginning, it's only
> Moreover, I don't want to see the money "flow" be influenced by DDs:
> that's unethical, and deviant.
You don't want donors to decide directly what to fund and you don't want us
to tell them what to fund?
It looks like you're against using money to let us do real work. :-)
> I'm (almost) fine with a Bounty structure, but the sponsors and sponsoree
> should define the terms of the bounty directly, with or without
> contract, that's their call. But I DON'T WANT to see a DD structure in
> the middle, *ESPECIALLY* for "counselling" purposes: that would only
> benefit to the known developers, not the technically good ones.
So you would be okay with the structure that I described, provided that
there's no voting mechanism and that donors are left to themselves to
select the projects to fund?
> Good technical solutions are still what Debian is about right ?
This is not related:
- known developers are not necessarily bad developers
- technically good developers are not necessarily unknown
And even if technicall bad but known developers are paid to do some work,
nothing prevents technically good developers to make sure the project is
sane from the beginning by reviewing the project proposals on the
infrastructure. And nothing prevents further improvements later in the
process. Being paid doesn't mean that we stop doing free software.
Premier livre français sur Debian GNU/Linux :