Re: [tlug] Zurus distributions experience

["Stephen J. Turnbull" <stephen@example.com>]

I'm going to move my responses to this thread elsewhere, to a blog or
something.  But I'll answer a couple of the concrete questions here.

Lars Kotthoff writes:

 > Point taken. It's debatable though whether the possibility of commercial
 > exploitation -- which isn't ruled out by releasing it under the GPL, you can
 > still charge for e.g. consulting -- or the potential benefits to the academic
 > community weigh higher. I would certainly be hesitant to say one or the other
 > for every case.

"Do consulting" is insulting.  A patch to a software program is a gift
that keeps on giving forever, with no further cost to anybody.
Consulting means you have to work your butt off every day, travel on
weekends, and wear a beeper.  That's not comparable compensation for
the contribution you made, that's wage slavery (actually, worse, since
there are no guarantees).

Note that Richard Stallman has never worked for a living.  Nothing
wrong with that, per se[1], but it's kind of galling that somebody who
has never had to work for a living so easily condemns all programmers
and others in the software industry to wage slavery in the name of
software freedom.

As for the potential benefits to the academic community of the
ketchy-ketchy sakusen, approximately zero.  Companies that don't give
back under BSD mostly won't give back under GPL, either.  Yes, I know
about Jeremy Allison's "Freedom Fighters" blog, and the NetApp example
(the company kept its FreeBSD-based product proprietary, but is a big
contributor to Linux).

Guess what?  As evidence it's useless, because NetApp almost certainly
chose FreeBSD *because* it could keep its value-added proprietary and
closed.  Had there been *no* permissively licensed OS, would they have
chosen Linux?  There's a possibility of course, but guess what?  They
might have preferred to (gasp) *pay* for the privilege of keeping
their code secret.  Or they might have decided the risk wasn't worth
the cost, and cancelled the product.  Consider: proprietary code is
not NetApp policy, the way it is with Microsoft; it's something the
company only does when it's really big, otherwise they are happy to
contribute to Linux.  It is not at all obvious that they would not
contribute to Linux if it were permissively licensed.

And note that NetApp are *not* required to contribute to Linux; they
could simply bundle source with their products, and let others do the
nitty gritty work of integrating.  Nope, uh-uh, they are a *presence*,
especially in NFS support.  Hm.  I wonder why?  Oh, yeah, because
there are *lots* of Linux boxes that connect to NetApp storage by NFS.
Do I smell "commercial interest" here?  You bet I do!  Improvements to
NFS are only valuable to NetApp if they are widely disseminated, and
the cheapest way to do that (not to mention smoothing the approval
process for patches that make Linux work better with NetApp storage)
is to pay the lieutenant responsible for NFS.[2]

Let's put it bluntly: Jeremy Allison's argument the the difference in
treatment of FreeBSD and Linux is copyleft has all the intellectual
integrity of the BSA valuing "pirated" copies at retail list price
when computing "damage" to copyright holders.  It only stands up if
you refuse to think carefully about it.

 > I admit that I don't understand all the economics behind it, but
 > are you saying that in general a piece of research, if exploited by
 > a company, benefits the general public? For example a specific
 > algorithm in Mathematica?

Of course, in as much as both the owners of the company, who receive
the profits, and the customers, who clearly value Mathematica more
than the price they pay (else they wouldn't pay), are members of the
general public.  Granted, figuring out what the benefit of a given
algorithm to a given customer is, is hard, but overall that
inequality must hold.

If you think the benefits from academic research (or any social
activity, for that matter) simply diffuse into the "general public"
the way that Chanel No. 5 diffuses into a ballroom, think again.

 > I'm paying tuition fees for, among other things, being able to
 > checkout books and journals from the library. It's debatable
 > whether this is an appropriate amount, but it's not really like a
 > subsidy.

Exactly my point.  If you're not receiving a subsidy, neither is the
commercial company.

 > My point is that everybody's taxes are paying for the research and if one
 > company uses it, they are benefiting from that.

So what?  Disproportionate benefit does not equal subsidy.  *Nobody*
pays in taxes the benefits they receive from public goods (except by
accident).

 > If I release my research software, I can change the license.

No, you cannot.  Not if it incorporates a line of GPLed, or other
strong copyleft, code.  And if your research software includes
incompatibly licensed code, you cannot release it at all.

And even if you write your own software from scratch, by releasing
under the GPL you impose those disablities on all who might build on
your research.


Footnotes: 
[1]  Neither has Mark Knopfler, and let me tell you, that's the way I
like it, uh-huh!  Ditto Yoyo Ma, Picasso, Roger Zelazny, you know the
usual suspects....

[2]  Note that this is not necessarily a bad thing.  There's very
strong social pressure, and even stronger pressure from Linus himself,
to be "honest" about applying quality standards to "own patches".  But
you can short circuit at least one whole level of politics for "own
patches".