[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Skolelinux and the "Debian Labs" idea



On Sun, Oct 05, 2003 at 03:47:15PM +1000, Martin Michlmayr - Debian Project Leader wrote:
> * Anthony Towns <aj@azure.humbug.org.au> [2003-09-20 17:07]:
> > By contrast, I wouldn't have a problem in principle with, eg, "HP
> > Debian Labs".
> I'd also have no problems with this.  However, I also think that the
> Debian Labs should not be used to make a profit.  I think a HP Debian
> Labs should be of a similar nature as HP Labs which is a research
> facility.

I suspect that HP shareholders think that they're using HP research labs
to make a profit.

Certainly it'd be reasonable to restrict "Debian Labs" to groups that
are doing R&D rather than sales. Do you want to limit consulting work too
though? 

If I've got $50k to spend getting, say, Gnucash improved to better
handle Australian accounting rules [0], does it make any real sense to
exclude Debian Labs from any possiblity of working on it? If you do want
to limit forms of income like that, then you're going to end up limiting
Debian Labs to big companies that can support research groups internally,
or to groups that can't pay Debian hackers reasonable wages.

> HP Debian Labs could do development, research and market
> research.  HP could of course provide support for Debian and sell it,
> and also publizie their Debian expertise (including ther HP Debian
> Labs), but they shouldn't use HP Debian Labs direcetly to sell any
> services or products.

Not even Debian stable CDs? Why not, exactly? What conflict of interest do
you see here?

(For reference, I joined Debian specifically because it doesn't
discriminate against people trying to make money out of free software,
even by building proprietary software on top of it -- indeed, we go so
far as to explicitly support such uses in a few ways)

> > Is it appropriate for developers to hang "Debian Labs" on the
> > doorknob to their bedrooms or studies?
> I guess not really; it should be bigger than that.  "How big" is a
> difficult question, though.

Well, obvious possible distinctions from the doorknob scenario would be:

	a) More than one researcher
	b) Full-time, paid researchers
	c) Accredited researchers

Is there a problem with a small business setting up a "Debian Lab"
by hiring a single Debian developer to do largely undirected Debian
development, at $200k/year, under the general philosophy that if Debian's
improved they'll be able to sell more hardware to more people, or make
higher profits otherwise? If so, that rules out (a).

(b) and (c) are probably not significantly different in effect.

> > Is it appropriate for Debian Labs guys to do support as well as
> > development? Could you have a Debian Lab that preinstalls Debian systems
> > and fixes them when they break? If so, does everyone in the lab have to
> > be a registered developer? Is it okay to install software from contrib or
> If they don't charge (or perhaps charge but not make a profit?) then
> they can of course install Debian on systems.  

Isn't charging for your time making a profit?

What if someone's employed by the same company to work part-time in a
Debian Lab, and part time doing Debian support, together making up a
single full time job? What benefit does that arrangement have over being
full time employed in a single job, that does research in a Debian Lab
and does Debian support?

> For example, I think it
> would be nice if they would go around and show Debian in schools and
> help them set it up.

If someone does support for schools fulltime, should they get to call
themselves a Debian Lab? If they don't make a profit? If they don't
pay their staff (ie, it's all volunteer)? If they do pay their staff,
is their any salary limit? How about if they do it cheaper than any
competing support organisations, but still make a profit?

I think Debian Labs should emphasise the research component, and set
a minimum level of quality, but leave worrying about profits to the
management and shareholders.

> > non-free on such machines? Unpackaged stuff? Stuff packaged locally? LSB
> > stuff? Proprietary stuff like win4lin or CrossoverOffice?
> non-free/proprietary stuff, better not.  

Even stuff packaged in non-free? How about stuff they maintain in
non-free?

> Unpackaged/LSB, I think so, yeah.

How about unpackaged stuff that they think's free, but debian-legal
doesn't?

How about stuff that's under NDA and thus isn't free now, but will be
free later, potentially like AMD-64 stuff?

How about non-free stuff that they're trying to reimplement, or be
compatible with? If they're working on Samba, or Abiword, or Wine,
or Plex86?

> > Do we want to allow people to earmark donations for a particular
> > Debian Lab, as a way of indirectly allowing users to sponsor
> > particular developers or projects? Is it possible for SPI or similar
> Yes, I think we could encourage people to give 1/2 of their donation
> to SPI And 1/2 to a Debian Lab.

I agree (presuming the .5:.5 ratio is determined by the contributor, and can
be 0:1 or 1:0 etc).

There are tax implications here [1]. The money goes:

	donator * -> SPI * -> Lab -> employee *
	                          -> expenses
	                          -> profits *

with each "*" representing a point at which the government could end up
taking a cut. SPI being a tax-exempt non-profit allows the first to *'s to
disappear, if the appropriate rules are followed. This is as opposed to
what happens without SPI, which would be:

	donator * -> Lab -> ... *

with the donator not being able to claim the donation against tax.

The rules for donations are usually something like requiring they further
the organisations interest, and aren't a quid-pro-quo arrangement. So
if you go to a charity ball and pay $1000 to attend, that's not tax
deductible because it was a payment for something. The extra $1000 that
you donate when they bring a hat around is a donation. The $1000 raffle
ticket you buy isn't.

If for-profit companies are allowed to run Debian Labs and receive donations,
there's no way to avoid that increasing their profits. If they do:

	Descr.		Income		Expense

	Donations	  $30000
	Printer Profits	$1000000
	Salaries			$800000
	Profits				$230000

it's impossible to tell whether without the donation their salaries
would have been $770k or their profits would have been $200k. If the Debian
Labs happen to not be a profit centre, you can likewise end up with:

	Descr.		Income		Expense

	SPI Donations	  $30000
	Subsidy		 $800000
	Expenses			 $50000
	Salaries			$780000

and the problem of working out whether removing the $30k donations would
have dropped the expenses down to $20k (no bandwidth or new computers for
anyone), the salaries down to $750k (no Christmas bonusses), or increased
the subsidy from the corporations profit centres to $830k -- and even
in that case, it's impossible to say whether that's just a bonus for
shareholders, or the only thing stopping the Lab from being downsized.
Note that the above can just as easily be a statement from a non-profit
organisation as one from a subdivision of a for-profit corporation.

Anyway.

The main issue for tax is likely to be ensuring that donations aren't used
as a tax avoidance scheme. Someone saying to HP, "hey, instead of paying
you directly, I'll make a donation to your labs via SPI. muahahaha!" will
get us all into trouble. And if we can avoid that, we'll probably avoid
the other problems in so doing.

Allowing people to earmark funds for project areas rather than companies
might be one way. Limiting the size of donations might be another. Having
Debian take a cut (ie, have us tax donations instead of the government)
might be another. 

The other side of the coin is that we don't want to devote to many
resources to looking after all this stuff. We don't want to spend
donations on getting accoutants to audit Debian Labs' books to make
sure they're not making a profit, or having developers spend their time
micromanaging them, or spend too much time working out which Labs are
doing a good job and which aren't; far better to leave that to the people
in the Labs themselves, and to the people thinking about earmarking a
donation for them.

FWIW, I'm in favour of requiring Debian Labs to be R&D focussed and to
employ full time researchers, to provide debs of everything they develop,
to focus on getting their successful research into main, and letting
pretty much everything else develop as best it may -- having a focus
doesn't mean to the exclusion of everything else. I'm also in favour
of letting people do tax-deductible donations to Debian Labs via SPI,
if it's possible. Obviously Debian Labs will have to say "no" to some
contracts if they want to keep the moniker -- ones that say "here's $5M;
now make all your staff wear Microsoft suits and ties, and play Windows
Solitaire all year", eg. But we shouldn't force them to ever say "no"
to money for working on things that actually improve Debian, imo.

Cheers,
aj

[0] I don't, for reference.

[1] aka: Here be Dragons

-- 
Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
I don't speak for anyone save myself. GPG signed mail preferred.

Australian DMCA (the Digital Agenda Amendments) Under Review!
	-- http://azure.humbug.org.au/~aj/blog/copyright/digitalagenda

Attachment: pgpom0g5Q4YFn.pgp
Description: PGP signature


Reply to: