A Big Fly in the Open-Source Soup (Business Week) [LWN.net]

Cross-post to LWN

Posted Aug 15, 2004 16:34 UTC (Sun) by JLCdjinn (guest, #1905) [Link]

This probably doesn't need to be said, but better safe than sorry. If someone does write such a letter, it would be nice to have it cross-posted here to ensure that LWN readership has an opportunity to read it. Thanks.

Cross-post to LWN

Posted Aug 16, 2004 0:43 UTC (Mon) by felixfix (subscriber, #242) [Link]

First, let's deal with the bugs in your column.

You say:

Intellectual-property questions about Linux came to the forefront
after the SCO Group (SCOX ), which acquired the Unix trademarks,
launched a series of lawsuits against alleged infringers of its
rights.

SCO did NOT acquire UNIX trademarks. The UNIX trademark is owned by
The Open Group.

For someone who claims to be writing about the intellectual property
failures of Linux and the GPL, you seem to have a weak grasp on what
IP is and is not.

You also don't know much about copyright licenses or the GPL's history
if you think it needs court cases to back it up. Without the GPL,
there is no right to redistribute the software at all. Only by
accepting the GPL can software be redistributed, and the GPL has a
long history of companies who violated it seeing the light due to
private discussions with the Free Software Fondation and private
authors. It is much more likely that these companies' lawyers told
them to 'fess up and get their act together because of the GPL's
clarity, rather than try to smear the GPL as murky and vague. If
those company lawyers did see it as vague and murky and wanted to
avoid it, their only choice would be to not distribute it at all.
Showing the GPL to be invalid would end all redistribution rights. It
would not put the licensed code into the public domain.

Besides which, the GPL only covers redistribution. It has nothing to
do with using the code. Micorsoft uses GPL code internally. They
also redistribute plenty of it, and they do it correctly. No company
has shown more interest in dissing the GPL, with all their silly
remarks about it being a cancer and un-American, and their financial
backing of SCO's even more bizarre lawsuits. You'd think if it was
murky and vague, they'd be the first to say so, and considering their
legal history, you'd also think they'd be more than happy to spend a
few million dollars proving it invalid in the courts, rather than just
snipe at it in speeches or thru their proxies. Actions speak louder
than words. Microsoft's actions say the GPL is just fine, thank you.

As for patents threatening GPLs code, it threatens proprietary code
just as much. It is so much easier to hide stolen and patented code
in proprietary software; the source is not available for inspection.
Any idea that GPL programmers should inspect their code for patent
abuse is laughable. I have written plenty of proprietary code in the
last 30 years, and have never ever heard of anyone in any of those
companies even mentioning patented code, let alone looking thru our
own code for patented code.

As a programmer who often has to choose software both for myself at
home and for my employers, I choose GPL whenever possible just for the
very reason you condemn it: I know the company won't be able to take
it away from me, and that others will contribute improvements just as
I do myself.

The BSD style licenses are certainly better than closed source, but if
you look into what is common among the many forked UNIX variants, and
the many forked BSD variants, you will notice they allow proprietary
forks. Linux the kernal has been around about as long as the various
BSD flavors, yet Linux has not forked, and there have been at least 5
BSD variants.

I will give you one example of why I prefer free source software.
Note that I mean free as in speech, not free as in beer. Have you
ever used Framemaker? Many years ago, I worked at a company which
used it, and it was an excellent change from Microsoft Word in every
way. It worked reliably, it was logically organized, it was cross
platform. Several years ago, there was a free (as in beer) beta copy
of a Linux version. I downloaded it, used it a bit, fell in love with
it all over again. But I stopped using it when I found it had an
expiration date as of the end of the year. They had announced
preliminary pricing of $1000 for the program, and I would have paid
for it, it was such a wonderful document editor. I drooled at the
thought of having it again. But I was not about to tie myself to a
program without source, with an expiration date, and sure enough, they
announced that there would not be a production version. If I had
written documents in Framemaker, I would have had to fool around with
my system's clock to use my old files, and I would have had to save
them all and lose all my formatting. I don't need that hassle.

That sums it up right there. I will not have my documents, or my
employers' documents, held hostage to closed source software. It is
mainly free source vs closed source, but I will also choose GPL over
BSD license simply because BSD encourages forking and GPL discourages
it.

I have NEVER had an employer tell me to avoid GPL software. Some have
prefered it. The only business people who hate the GPL are those who
can't take the GPL'd code and turn around and sell it, and that suits
me as an author jsut fine, thank you. I wrote my code; I give it to
the world on the condition that they keep it open to the world. If
you can't understand that simple philosophy, then you have no business
writing about the faults of the GPL.

And if you can't recognize that the GPL discourages forking, and why
that is good, you have no business writing about technology at all.

Cross-post to LWN

Posted Aug 16, 2004 3:17 UTC (Mon) by dwheeler (guest, #1216) [Link]

If you plan to send that to the magazine, I suggest toning it down so it doesn't look like an attack (e.g., anything with "you"). Yes, there are mistakes, but you can express everything you've said without making it look like a personal attack.

Cross-post to LWN

Posted Aug 16, 2004 3:37 UTC (Mon) by lakeland (subscriber, #1157) [Link]

I think this needs major changes before you send it. Firstly, the goal of
an article critique is to change the author's opinion, or at least to be
published as an alternative opinion. That means your article must be
coherant, internally consistant (it must flow) and reasonable to read (no
personal attacks, no overstating points).

"bugs in your column" says to me that you're a programmer, and any
non-programmer reading your critique is going to feel your opinion doesn't
apply to them.

"Weak grasp on IP" is totally inapproprate -- Personal attacks are the
fastest way to have your opinions sent to /dev/null.

Most of the text relating to GPL in the courts is auxillary to the thesis
of your argument and should be deleted. Saying "The GPL has sucessfully
been informally enforced on a number of occasions" is more than enough --
especially since Stephen's point was that the GPL is ambiguous, _NOT_ that
it was invalid. For instance, exact details of modules vs RPC vs
libraries have not been resolved yet. Stephen had a point and by
misinterpreting it, you appear a fool.

A golden rule is to not argue against a point the author didn't make
(except when creating FUD, but I digress). You do this again even worse
when you talk about the GPL only applying to copying, not use. Did the
author of this article make a mistake here? Because I don't see him
talking about use once.

Next you say patents affect GPL less than prop. software, and then
contradict yourself in the next sentence by saying "It is so much easier
to hide stolen and patented code". If it is much easier to get away with
it when not using linux, then patents are more of a problem for linux.

Next you go on some crazy tangent about cancer and anti-american -- did
the article say anything along these lines? Were I the author reading
your submission I would have deleted it as a loony with an agenda at this
point -- you've clearly read far too much bad journalism about SCO and are
replying to that instead of the article in front of you.

That is unfortunate since you start to make your point in the next
paragraph. I would prefer if your point was made better (and I would
personally delete _everything_ up to there).

Finally, you finish with yet another attack on the author. If they had
managed to read to the end, this would have killed any remaining
credibility you had. I accept it is fun to diss a badly written article,
but if you're posting the criticism to the author then the goal is to get
the author on side, not to ostracise them.

Anyway, as a counter-point to your post, I'll try my hand at writing a
criticism next (I'll post it at the same level as yours on LWN).

Cross-post to LWN

Posted Aug 16, 2004 14:10 UTC (Mon) by sphealey (guest, #1028) [Link]

> You also don't know much about copyright licenses
> or the GPL's history if you think it needs court
> cases to back it up.

That is the fastest and most effective way to turn a reporter against you and your cause for the rest of his career. Reporters are even touchier than doctors about anything they view as an attack on their knowledge or professional ability, and unlike doctors reporters NEVER forget what they consider to be an affront.

Something to consider.

sPh

I did that a couple of days ago...

Posted Aug 16, 2004 3:13 UTC (Mon) by leonbrooks (guest, #1494) [Link]

...and the reply was... odd.

He seems to be reasonably pro-FOSS but anti the GPL. Go figure.

"Murky" is his honest opinion, he doesn't view section 5 or 7 of the GPL as being at all clear. He didn't comment on the Munich restart, so I'm guessing he didn't know at the point where I wrote to him.

I did that a couple of days ago...

Posted Aug 16, 2004 8:01 UTC (Mon) by alextingle (subscriber, #20593) [Link]

> he doesn't view section 5 or 7 of the GPL as being at all clear.

Sections 5 & 7 are the ones that talk about how the license is applied. They look odd to some people because they look nothing like contract clauses: "You don't have to accept this, but nothing else gives you permission to distribute..."

Groklaw have spent a great deal of time on exploring the theory that the GPL is not a contract, but a License (hence the 'L'). Under this theory, sections 5 & 7 make much more sense. Business people are familiar with contracts, so they read the GPL like a contract and it confuses them.

I did that a couple of days ago...

Posted Aug 16, 2004 15:35 UTC (Mon) by iabervon (subscriber, #722) [Link]

There is the additional confusing factor that section 5 and probably section 7 are only informational, not actual terms. Section 5 is an explanation of the law as it applies to people who don't accept the license, and therefore, confusingly, neither can be nor needs to be enforced. It is also potentially wrong; if the software is dual-licensed, there may be something else which grants the reader those rights under different terms.

The GPL is slightly marred by RMS's tendancy to stick commentary in the middle of documents, making it difficult to tell what is the actual license and what is not. Sections 5, 7, 9, and 10 are statements which would be true of people who did not accept the license, while the rest are actual license terms and only apply to licensees. Section 4 is probably not strictly-speaking necessary, but it is a standard thing to include in licenses. Section 6 is probably a real clause but it is worded oddly as a grant of rights to someone other than the licensee. Section 8 does apply to the licensee, but only as a potential original author of a derived work. Sections 11 and 12 also couldn't be license terms, in that the people they are meant to inform (i.e., end users) are, as is explicitly stated by section 0, not required to agree to the license or even read it. The actual license portions, 1-3, have the nice feature that they are of the form "you may... provided...", which, if it were not obfuscated by the remaining sections, would make it clear that this is a grant of limited rights without compensation.

OK, it's starting to become clearer now...

Posted Aug 17, 2004 17:33 UTC (Tue) by leonbrooks (guest, #1494) [Link]

He keeps referring to the GPL as "a contract". In the last step of our conversation I told him point blank that the GPL is not a contract. We'll see if he gets it now.

Still no comment on Munich.

He didn't seem to realise that patents will also kill use of a non-GPLed program, all you get is a delay while the patent holder mows down your supplier. Fixed.

Seemed to think that Linus owns "the copyright" on Linux, told him that Linus owns the trademark and some of the copyright, rest is owned by man+dog.

Seemed to think that the FSF somehow "owns" the GPL. Ehhh? Certainly blindsided me with that one! Linus and RMS don't get on, so maybe one day RMS is going to take his licence back? Won't fit in the hole no matter which way I turn it. Told him that Linus might rename his licence, but he couldn't change the terms without the unanimous consent of thousands.

He needed a short summary of TSG's antics, which he got. He'd been mesmerised by the propaganda, not the facts.

We discussed Microsoft's possible actions and agreed that they would be acting in their own best interests, exactly how was difficult to predict but it would be nasty. He didn't react to the idea that Scott might change sides for pay as Sun's cash reserves run low.

EOM, so far.

My attempt

Posted Aug 16, 2004 4:27 UTC (Mon) by lakeland (subscriber, #1157) [Link]

Dear Stephen,

Thank you for your commentary on licencing issues in open-source software.
This is an issue I spend a lot of time thinking about, and it is nice to
see it discussed in the mainstream press. Unfortunately, I have to
disagree with your thesis; let me explain by way of example:

Imagine a company looking for some software to help run their business.
If the company is just planning on using the software then the differences
between the GPL and BSD style licences are irrelevant, so to make the
discussion interesting we can assume the company is going to be modifying
the software.

From a legal perspective, the GPL is extremely simple (just read a
commercial licence some time!) However, the BSD licence is even simpler --
the company can (virtually) do whatever it wants. I think this was why
you preferred it. However, let me say why I prefer the GPL, and why I
think other companies should too.

When somebody develops GPLed software, they know that no company is going
to 'steal their hard work'. The idea of having your hard work taken and
privatised by someone else is a strong disincentive for working on BSD
software. It is no coincidence that GPLed projects have far more
developers than BSD ones.

So, for individual developers there is more incentive to use the GPL. I
think there is more incentive for companies too: if a company chooses to
base their work around GPL licenced software, their contributions and the
contributions of other companies will provide the lifeblood to keep the
software alive. For only a fraction of the cost of maintaining software,
companies can share the maintenance costs and ensure the software
continues to meet their needs. If instead the software was BSD licenced,
why would the companies provide their maintenance work to their
compeditors?

Essentially, I would rather contribute to GPLed software, and so would
most developers. If the projects I contribute to were re-licenced under
the BSD, I lose my incentive to contribute -- especially if the other
users are my compeditors! Forcing everybody to contribute is why GPL
projects have been so successful, and creating a little legal complexity
for companies attempting to profit from GPLed code is a small price to pay
for an active community. You just have to count the number of successful
BSD projects compared to GPL projects to see how much harder it is to
build a community without enforced cooperation.

At the end of the day, the people who make the software will continue to
prefer the GPL. So potential users of the software will have to decide if
they'd rather play by the rules of the GPL, or write their own software.
Complaining that you'd rather have the software without strings is naive
-- the strings are there for very good reasons.

PS: While I'm writing, a few nits with your article:

1. SCO did not obtain any trademarks on UNIX from Novell. They obtained
the business of selling UNIX, and possibly the copyrights to ancient
versions of UNIX, but not the trademarks (which are owned by "the open
group") or any patents (which would have expired by now anyway).

2. Patents are just as much a problem for BSD licenced projects as for GPL
licenced projects. Technically, they are just as much a problem for free
proprietary projects too, except in those cases it is somewhat harder to
prove infringement. In any case, I cannot think of a single instance of a
company being sued over patent issues in free software -- for a start it
would be very hard to convince a judge that a company merely using the
software owes you money, and there is little incentive in going after the
developers since they are usually individuals without money. You
acknowledge this yourself when you note that patents are not an immediate
threat.

3. Did Apple really consider using linux at the core? I am very surprised
since by using OpenSTEP they were able to shortcut years of development
that would have been needed if they'd chosen linux. Further, Apple uses a
very large number of non BSD programs throughout Mac OS X (isn't CUPS
under the GPL?), then there is their HTML renderer (under the LGPL), and
their recently-renamed network management software (AFPL).

Anyway, once again thank you for writing your commentary, and I look
forward to reading your response.

Corrin Lakeland

Your attempt

Posted Aug 16, 2004 4:58 UTC (Mon) by dscribner (guest, #7233) [Link]

If I may, I would like to suggest a couple minor corrections before you send this...

"compeditors" should be spelled competitors, and "the open group", being the proper name of a company, should be capitalized (The Open Group). Thanks!

Re: My attempt

Posted Aug 16, 2004 11:49 UTC (Mon) by sdalley (subscriber, #18550) [Link]

My, what a good way to say it ...

Comments

Posted Aug 16, 2004 15:18 UTC (Mon) by dwheeler (guest, #1216) [Link]

Here are some comments, hope they help!

First, you don't seem to really address the article's theses. For example, that the "the GPL is hardly a model of clarity, and few disputes involving it have gotten to court, so case law has done little to clarify its meaning." I'd reply "First, you claim that the GPL isn't clear and few disputes go to court, but most lawyers find that the GPL is extremely clear and thus few are willing to risk going to court. The German court has had no problem enforcing the GPL, and in fact has made it quite clear that it's enforceable. For more information, see Eben Moglen's article at http://www.gnu.org/philosophy/enforcing-gpl.html. And the GPL itself depends on copyright, so it's not somehow illegal due to copyright laws."

The article says that "backers see open-source software as part of a social and political movement that's frankly anti-corporate". That's simply not true; why are so many companies involved?

In "why I think other companies should too", change to "why I think many companies do too."

In "You just have to count the number of successful BSD projects compared to GPL projects to see how much harder it is to build a community without enforced cooperation", append some statistics to justify the claim. There are lots of stats in http://www.dwheeler.com/essays/gpl-compatible.html with links to how to get the latest stats. Note that Freshmeat reports data on http://freshmeat.net/stats and Sourceforge reports on http://sourceforge.net/softwaremap/trove_list.php?form_ca...

Thus you could append this: "For example, according to Freshmeat.net's statistics of projects (dated August 16, 2004), 68.61% use the GPL, while only 9.46% use one of the common BSD-style "licenses" (those licenses with use by 1% or more of the projects: the BSD originial and revised, Artistic, MIT/X, and public domain). Note that 5.66% use the Lesser GPL (LGPL), a compromise license between the two, and the combination of GPL and LGPL accounts for 74.27% of all projects tracked by Freshmeat.net. It's clear there is room for a variety of licenses, but it's also clear that developers tend to choose the GPL instead of various alternatives."

s/compeditors/competitors/

Change "the people who make the software will continue" to "some of the people who make the software will continue". Many smart developers prefer the BSD license, or choose the BSD license for a particular project for a variety of reasons. For example, I've decided to release some code as GPL, and other code as BSD(revised), because I had different motivations for each. Certainly don't imply that all developers prefer the GPL; statistics suggest it's a majority, but it's certainly not all.

Change "privatised by someone else" to "privatised by someone else,
or 'embranced, enhanced, and extinguished',".

You might find the following links interesting:
http://techupdate.zdnet.com/techupdate/stories/main/0,141...
http://techupdate.zdnet.com/techupdate/stories/main/0,141...

You don't want to imply that all developers use the GPL (or that all developers develop OSS), but it IS true that many prefer the GPL, and that needs explaining.