[tei-council] Fwd: TEI licensing issues

[Martin Holmes]

Hi Piotr,

On 11-09-13 12:24 PM, Piotr Bański wrote:
> On 13/09/11 18:27, Martin Holmes wrote:
>> On 11-09-13 05:45 AM, Piotr Bański wrote:
>>> On 13/09/11 14:26, Laurent Romary wrote:
>>>> OK. What I meant is that, if I have understood the situation
>>>> correctly, it was not necessarily planned to have this GPL
>>>> constraints on some of our TEI productions.
>>>
>>> The description of what you call GPL constraints as "viral" comes, as
>>> far as I barely recall, from some Microsoft spokesman. He was very
>>> successful -- this is a strong meme, people catch it unawares. It's very
>>> cunning: it reverses the imagery, from a means to keep stuff free to a
>>> picture where that strategy acquires evil/dangerous connotations, and of
>>> course the meme cleverly plays on the right to possess and to draw
>>> income, although the license actually doesn't prevent either.
>>
>> I don't think I'd quite agree with that. I've been involved in
>> commercial software development for many years, and I've come to dislike
>> the GPL; it's generally very difficult to use GPL code in a commercial
>> project without having to make your own commercial source code public as
>> well, and that acts as a serious deterrent to commercial software
>> development. The LGPL is slightly better, but even that involves lots of
>> compromises and can often force you to restructure your application so
>> that compiled libraries are kept separate and linked at run-time rather
>> than compile-time.
>
> I can't resist remarking that, understandably, this sounds very much
> like it's written from the perspective of someone involved in the
> commercial software side of things :-)

Certainly. I've been involved in commercial software development since 
1995. That was exactly the perspective I was writing from. I think it's 
important that the TEI enables and encourages commercial software 
development resting on its codebase and standards, just as much as it 
encourages development of the various types of free software. oXygen is 
a case in point; it's very helpful to the propagation of TEI that oXygen 
has so much of Sebastian's great work built in to it, and I hope it's 
equally valuable to the oXygen developers that they can advertise it as 
a TEI-enabled tool.

> One fragment I'd like to pick on
> is that GPL, directly or indirectly, is a "serious deterrent to
> commercial software development". How can that be? If you develop
> commercial software, you are not forced to use GPL-ed software -- you
> can develop your own closed equivalent, can't you.

My point is precisely that you can't use GPL'ed code very easily in most 
commercial software projects. You can of course go away and write your 
own libraries, but that can be prohibitively expensive for small 
developers, as well as inefficient. Where good, open, well-tested 
libraries of code exist, I think it's a shame when their licensing 
prevents them from being used in commercial projects; and when 50 small 
developers each have to write their own library doing X, then each will 
have unique bugs and inefficiencies that weren't in the open-source library.

> If you decide to use
> the shortcut provided by the existence of good free GPL-ed software,
> well, there is a price associated with that, because the people who have
> put their time and skills into that software have certain expectations
> with regard to its future fate: it was developed in the open, and it
> shall remain in the open, so that everyone can benefit. "Benefit" also
> monetarily, because open source software need not mean software that
> doesn't provide a way to get revenue, you just need a different business
> model to start benefiting from that.

Exactly. But if my main pleasure in my work lies in writing 
applications, then I don't really want to change my business model so 
that I become a support person or a customizer or something else; I'm 
more likely to abandon the project.

>> The GPL encodes an ideological stance which is strongly opposed to
>> anyone keeping source code secret. Most commercial software developers
>> feel that they must keep at least some of their source code secret,
>> otherwise their software will be recompiled and released for free by
>> hackers or competitors.
>
> One very ideological reply could be that if both the said developers and
> their competition decided to compete in the open, their energy and
> skills would find more efficient use. But of course such opening of
> software happens rarely nowadays, and hasn't happened often in the past,
> for many reasons, not the least important of which is that it may be
> very difficult to ensure that all sides of such a competition would
> behave ethically.

That's the point, really. If I write a little open-source application 
competing with a piece of MS software, then MS can quite legitimately 
recompile my application and give it away for free, removing my market, 
and ensuring that I'll no longer be in business in a few months.

> Still, no one requires that creators of secret code release their code,
> because no one requires them to use GPL-ed code. And the 'release'
> clause only becomes relevant when you decide to distribute. So for
> projects that offer APIs in-situ, for example, without software
> distribution, GPL does nothing -- they are free to use GPL-ed code.

So the rise of the web application has provided a method whereby one 
class of development is able to get around the original intent of the 
GPL by hiding behind the fact that they're not "distributing" their 
application, whereas in fact they're distributing use of their 
application to anyone who wants it (or wants to pay for it) over the web.

>> I understand the motivation behind the GPL, but I think on balance its
>> effects on innovation and openness are actually negative. I much prefer
>> licenses like MPL and BSD which don't really attempt to control what
>> people do with the source code.
>
> In Computational Linguistic circles, this article[1] by Ted Pedersen has
> resounded pretty strongly. Pedersen gives arguments for why the open
> source development model guarantees innovation.
>
> [1]: http://www.aclweb.org/anthology/J/J08/J08-3010.pdf

But he seems quite comfortable with both GPL and MPL; it's the 
distinction between those that concerns me.

> In the end, I think the distinction between BSD and GPL flavours may be
> (brutally) reduced to the distinction between the attitudes of creators
> and the status of the project. For BSD, it is "here, I've created
> something cool, go and use it for your benefit, all I need is a
> copyright statement on your help screen". This is extreme generosity,
> very laudable. A bit like "here, have my kidney, I'll be ok". Super
> cool, if you can afford it. A friend has also remarked that in some
> cases, it may allow software companies to choose stuff under BSD, and
> many of them may be willing to give back. Possibly a kind of law of
> large numbers effect if it's true and measurable.
>
> In GPL, it is "here, I've created something cool, go and use it for your
> benefit, but since I have sweated over it, I would like it to maintain
> its open status (share-alike)". This is a guarantee that if someone put
> their time into an open project, the offspring of this project will
> remain open for others, including the original creator. This is, I
> believe, rather essential in the work on "minor", "under-financed",
> "non-central", "lower-density" languages, to make them open their data
> sheds and contribute towards a common goal, part of which is
> preservation of the given language.

It's not really a guarantee. Using GPL ensures that only unscrupulous 
people, or those who cannot be pursued legally in your jurisdiction, 
will not use your code for their own profit without giving back. It 
deters honest folks with commercial projects from using code, while 
providing no deterrent at all to others.

> This guarantee is not, in my view, essential for the TEI, which, as
> Martin has remarked, can stand on its own, at least among the Humanities
> enterprises. In other words, I believe that the TEI is capable of
> donating a kidney, and maybe even growing a new one to donate it again.
> Not in all areas, but surely in its main areas of application.

Agreed. And I think the TEI can only benefit from enabling commercial 
exploitation of its work. How pleased would we be if Apple or Microsoft 
released an application with built-in TEI support?

> My point regarding CC-BY was only: if they say it's incompatible with
> the license we have used so far (and "they" appear to be serious about
> this, whatever the legalese nitty-gritty is), let us rather use
> something that is compatible with it, not to hit the GPL world that we
> are still, until the change of the licence, part of, and that we still
> want to be able to incorporate the TEI (it's not a small world).

Agreed.

Cheers,
Martin

>
> Best,
>
>    Piotr
>

-- 
Martin Holmes
University of Victoria Humanities Computing and Media Centre
(mholmes at uvic.ca)